[AMDGPU] Autogenerate wqm.ll

[critson]

Summary

Switch wqm.ll to be autogenerated.
Replace gfx6 and gfx8 targets with gfx9 (wave64) and gfx10 (wave32).

[kmitropoulou]

LGTM

llvm/test/CodeGen/AMDGPU/wqm.ll

-; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=CHECK %s
-; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=CHECK %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX9-W64 %s
+; RUN: llc -march=amdgcn -mcpu=gfx1030 -verify-machineinstrs -mattr=+wavefrontsize32,-wavefrontsize64 < %s | FileCheck -check-prefixes=GFX10-W32 %s
 
 ; Check that WQM isn't triggered by image load/store intrinsics.
-;
-;CHECK-LABEL: {{^}}test1:
-;CHECK-NOT: s_wqm
 define amdgpu_ps <4 x float> @test1(<8 x i32> inreg %rsrc, i32 %c) {
+; GFX9-W64-LABEL: test1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v0
+; GFX9-W64-NEXT:    image_load v[0:3], v0, s[0:7] dmask:0xf unorm
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_store v[0:3], v4, s[0:7] dmask:0xf unorm
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v0
+; GFX10-W32-NEXT:    image_load v[0:3], v0, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_1D unorm
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_store v[0:3], v4, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_1D unorm
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %tex = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i32(i32 15, i32 %c, <8 x i32> %rsrc, i32 0, i32 0)
   call void @llvm.amdgcn.image.store.1d.v4f32.i32(<4 x float> %tex, i32 15, i32 %c, <8 x i32> %rsrc, i32 0, i32 0)
   ret <4 x float> %tex
 }
 
 ; Check that WQM is triggered by code calculating inputs to image samples and is disabled as soon as possible
-;
-;CHECK-LABEL: {{^}}test2:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: interp
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK-NOT: interp
-;CHECK: image_sample
-;CHECK-NOT: exec
-;CHECK: .size test2
 define amdgpu_ps <4 x float> @test2(i32 inreg, i32 inreg, i32 inreg, i32 inreg %m0, <8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, <2 x float> %pos) #6 {
+; GFX9-W64-LABEL: test2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b32 m0, s3
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    v_interp_p1_f32_e32 v2, v0, attr0.x
+; GFX9-W64-NEXT:    v_interp_p1_f32_e32 v3, v0, attr0.y
+; GFX9-W64-NEXT:    v_interp_p2_f32_e32 v2, v1, attr0.x
+; GFX9-W64-NEXT:    v_interp_p2_f32_e32 v3, v1, attr0.y
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    image_sample v[0:3], v[2:3], s[4:11], s[12:15] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 m0, s3
+; GFX10-W32-NEXT:    v_interp_p1_f32_e32 v2, v0, attr0.x
+; GFX10-W32-NEXT:    v_interp_p1_f32_e32 v3, v0, attr0.y
+; GFX10-W32-NEXT:    v_interp_p2_f32_e32 v2, v1, attr0.x
+; GFX10-W32-NEXT:    v_interp_p2_f32_e32 v3, v1, attr0.y
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    image_sample v[0:3], v[2:3], s[4:11], s[12:15] dmask:0xf dim:SQ_RSRC_IMG_2D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %inst23 = extractelement <2 x float> %pos, i32 0
   %inst24 = extractelement <2 x float> %pos, i32 1
   %inst25 = tail call float @llvm.amdgcn.interp.p1(float %inst23, i32 0, i32 0, i32 %m0)
   %inst26 = tail call float @llvm.amdgcn.interp.p2(float %inst25, float %inst24, i32 0, i32 0, i32 %m0)
   %inst28 = tail call float @llvm.amdgcn.interp.p1(float %inst23, i32 1, i32 0, i32 %m0)
   %inst29 = tail call float @llvm.amdgcn.interp.p2(float %inst28, float %inst24, i32 1, i32 0, i32 %m0)
   %tex = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f32(i32 15, float %inst26, float %inst29, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   ret <4 x float> %tex
 }
 
 ; ... but disabled for stores (and, in this simple case, not re-enabled) ...
-;
-;CHECK-LABEL: {{^}}test3:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK: store
-;CHECK-NOT: exec
-;CHECK: .size test3
 define amdgpu_ps <4 x float> @test3(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, float %c) {
+; GFX9-W64-LABEL: test3:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dwordx4 v[0:3], v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test3:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dwordx4 v[0:3], v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex.1 = bitcast <4 x float> %tex to <4 x i32>
   %tex.2 = extractelement <4 x i32> %tex.1, i32 0
 
   call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> %tex, <4 x i32> undef, i32 %tex.2, i32 0, i32 0, i32 0)
 
   ret <4 x float> %tex
 }
 
 ; ... and disabled for export.
-;
-;CHECK-LABEL: {{^}}test3x:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK: exp
-;CHECK-NOT: exec
-;CHECK: .size test3x
 define amdgpu_ps void @test3x(i32 inreg, i32 inreg, i32 inreg, i32 inreg %m0, <8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, <2 x float> %pos) #6 {
+; GFX9-W64-LABEL: test3x:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b32 m0, s3
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    v_interp_p1_f32_e32 v2, v0, attr0.x
+; GFX9-W64-NEXT:    v_interp_p1_f32_e32 v3, v0, attr0.y
+; GFX9-W64-NEXT:    v_interp_p2_f32_e32 v2, v1, attr0.x
+; GFX9-W64-NEXT:    v_interp_p2_f32_e32 v3, v1, attr0.y
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    image_sample v[0:3], v[2:3], s[4:11], s[12:15] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    exp mrt0 v0, v1, v2, v3 done vm
+; GFX9-W64-NEXT:    s_endpgm
+;
+; GFX10-W32-LABEL: test3x:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 m0, s3
+; GFX10-W32-NEXT:    v_interp_p1_f32_e32 v2, v0, attr0.x
+; GFX10-W32-NEXT:    v_interp_p1_f32_e32 v3, v0, attr0.y
+; GFX10-W32-NEXT:    v_interp_p2_f32_e32 v2, v1, attr0.x
+; GFX10-W32-NEXT:    v_interp_p2_f32_e32 v3, v1, attr0.y
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    image_sample v[0:3], v[2:3], s[4:11], s[12:15] dmask:0xf dim:SQ_RSRC_IMG_2D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    exp mrt0 v0, v1, v2, v3 done vm
+; GFX10-W32-NEXT:    s_endpgm
 main_body:
   %inst23 = extractelement <2 x float> %pos, i32 0
   %inst24 = extractelement <2 x float> %pos, i32 1
   %inst25 = tail call float @llvm.amdgcn.interp.p1(float %inst23, i32 0, i32 0, i32 %m0)
   %inst26 = tail call float @llvm.amdgcn.interp.p2(float %inst25, float %inst24, i32 0, i32 0, i32 %m0)
   %inst28 = tail call float @llvm.amdgcn.interp.p1(float %inst23, i32 1, i32 0, i32 %m0)
   %inst29 = tail call float @llvm.amdgcn.interp.p2(float %inst28, float %inst24, i32 1, i32 0, i32 %m0)
   %tex = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f32(i32 15, float %inst26, float %inst29, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex.0 = extractelement <4 x float> %tex, i32 0
   %tex.1 = extractelement <4 x float> %tex, i32 1
   %tex.2 = extractelement <4 x float> %tex, i32 2
   %tex.3 = extractelement <4 x float> %tex, i32 3
   call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %tex.0, float %tex.1, float %tex.2, float %tex.3, i1 true, i1 true)
   ret void
 }
 
 ; Check that WQM is re-enabled when required.
-;
-;CHECK-LABEL: {{^}}test4:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: v_mul_lo_u32 [[MUL:v[0-9]+]], v0, v1
-;CHECK: image_sample
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK: store
 define amdgpu_ps <4 x float> @test4(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, float addrspace(1)* inreg %ptr, i32 %c, i32 %d, float %data) {
+; GFX9-W64-LABEL: test4:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mul_lo_u32 v4, v0, v1
+; GFX9-W64-NEXT:    image_sample v0, v4, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dwordx4 v[0:3], v4, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test4:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mul_lo_u32 v4, v0, v1
+; GFX10-W32-NEXT:    image_sample v0, v4, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dwordx4 v[0:3], v4, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %c.1 = mul i32 %c, %d
 
   call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> undef, <4 x i32> undef, i32 %c.1, i32 0, i32 0, i32 0)
   %c.1.bc = bitcast i32 %c.1 to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.1.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   ret <4 x float> %dtex
 }
 
 ; Check that WQM is triggered by the wqm intrinsic.
-;
-;CHECK-LABEL: {{^}}test5:
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
 ; WQM was inserting an unecessary v_mov to self after the v_add. Make sure this
 ; does not happen - the v_add should write the return reg directly.
-;CHECK-NOT: v_mov_b32_e32
 define amdgpu_ps float @test5(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test5:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    buffer_load_dword v0, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX9-W64-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test5:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v0, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX10-W32-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.wqm.f32(float %out)
   ret float %out.0
 }
 
 ; Check that the wqm intrinsic works correctly for integers.
-;
-;CHECK-LABEL: {{^}}test6:
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
 define amdgpu_ps float @test6(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test6:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    buffer_load_dword v0, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX9-W64-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test6:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v0, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX10-W32-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %out = fadd float %src0, %src1
   %out.0 = bitcast float %out to i32
   %out.1 = call i32 @llvm.amdgcn.wqm.i32(i32 %out.0)
   %out.2 = bitcast i32 %out.1 to float
   ret float %out.2
 }
 
 ; NOTE: llvm.amdgcn.wwm is deprecated, use llvm.amdgcn.strict.wwm instead.
 
 ; Check that WWM is triggered by the wwm intrinsic.
-;
-;CHECK-LABEL: {{^}}test_wwm1:
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
 define amdgpu_ps float @test_wwm1(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_wwm1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.wwm.f32(float %out)
   ret float %out.0
 }
 
 ; Same as above, but with an integer type.
-;
-;CHECK-LABEL: {{^}}test_wwm2:
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_{{[iu]}}32_e32
 define amdgpu_ps float @test_wwm2(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_wwm2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_u32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %src0.0 = bitcast float %src0 to i32
   %src1.0 = bitcast float %src1 to i32
   %out = add i32 %src0.0, %src1.0
   %out.0 = call i32 @llvm.amdgcn.wwm.i32(i32 %out)
   %out.1 = bitcast i32 %out.0 to float
   ret float %out.1
 }
 
 ; Check that we don't leave WWM on for computations that don't require WWM,
 ; since that will lead clobbering things that aren't supposed to be clobbered
 ; in cases like this.
 ; We enforce this by checking that v_add gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_wwm3:
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: v_add_f32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_wwm3(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_wwm3:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB9_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX9-W64-NEXT:  .LBB9_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm3:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB9_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX10-W32-NEXT:  .LBB9_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.wwm.f32(float %out)
   %out.1 = fadd float %src, %out.0
   br label %endif
 
 endif:
   %out.2 = phi float [ %out.1, %if ], [ 0.0, %main_body ]
   ret float %out.2
 }
 
 ; Check that WWM writes aren't coalesced with non-WWM writes, since the WWM
 ; write could clobber disabled channels in the non-WWM one.
 ; We enforce this by checking that v_mov gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_wwm4:
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK-NEXT: v_mov_b32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_wwm4(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_wwm4:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB10_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB10_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm4:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB10_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB10_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.wwm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Make sure the transition from Exact to WWM then WQM works properly.
-;
-;CHECK-LABEL: {{^}}test_wwm5:
-;CHECK: buffer_load_dword
-;CHECK: buffer_store_dword
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: s_wqm_b64 exec, exec
 define amdgpu_ps float @test_wwm5(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_wwm5:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm5:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %src0, <4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %src1, %src1
   %temp.0 = call float @llvm.amdgcn.wwm.f32(float %temp)
   %out = fadd float %temp.0, %temp.0
   %out.0 = call float @llvm.amdgcn.wqm.f32(float %out)
   ret float %out.0
 }
 
 ; Check that WWM is turned on correctly across basic block boundaries.
 ; if..then..endif version
-;
-;CHECK-LABEL: {{^}}test_wwm6_then:
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endif
 define amdgpu_ps float @test_wwm6_then() {
+; GFX9-W64-LABEL: test_wwm6_then:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[0:1], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB12_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB12_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm6_then:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s0, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB12_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB12_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.wwm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Check that WWM is turned on correctly across basic block boundaries.
 ; loop version
-;
-;CHECK-LABEL: {{^}}test_wwm6_loop:
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %loop
-;CHECK: s_or_saveexec_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endloop
 define amdgpu_ps float @test_wwm6_loop() {
+; GFX9-W64-LABEL: test_wwm6_loop:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], 0
+; GFX9-W64-NEXT:  .LBB13_1: ; %loop
+; GFX9-W64-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_add_u32_e32 v3, -1, v3
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v3
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_or_b64 s[0:1], vcc, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    s_andn2_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_cbranch_execnz .LBB13_1
+; GFX9-W64-NEXT:  ; %bb.2: ; %endloop
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm6_loop:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:  .LBB13_1: ; %loop
+; GFX10-W32-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v3, -1, v3
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v3
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_or_b32 s0, vcc_lo, s0
+; GFX10-W32-NEXT:    s_andn2_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_cbranch_execnz .LBB13_1
+; GFX10-W32-NEXT:  ; %bb.2: ; %endloop
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   br label %loop
 
 loop:
   %counter = phi i32 [ %lo, %main_body ], [ %counter.1, %loop ]
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.wwm.f32(float %out)
   %counter.1 = sub i32 %counter, 1
   %cc = icmp ne i32 %counter.1, 0
   br i1 %cc, label %loop, label %endloop
 
 endloop:
   ret float %out.0
 }
 
 ; Check that @llvm.amdgcn.set.inactive disables WWM.
-;
-;CHECK-LABEL: {{^}}test_wwm_set_inactive1:
-;CHECK: buffer_load_dword
-;CHECK: s_not_b64 exec, exec
-;CHECK: v_mov_b32_e32
-;CHECK: s_not_b64 exec, exec
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: v_add_{{[iu]}}32_e32
 define amdgpu_ps void @test_wwm_set_inactive1(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_wwm_set_inactive1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    v_add_u32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, v0
+; GFX9-W64-NEXT:    buffer_store_dword v2, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_endpgm
+;
+; GFX10-W32-LABEL: test_wwm_set_inactive1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, v0
+; GFX10-W32-NEXT:    buffer_store_dword v2, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_endpgm
 main_body:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %src.0 = bitcast float %src to i32
   %src.1 = call i32 @llvm.amdgcn.set.inactive.i32(i32 %src.0, i32 0)
   %out = add i32 %src.1, %src.1
   %out.0 = call i32 @llvm.amdgcn.wwm.i32(i32 %out)
   %out.1 = bitcast i32 %out.0 to float
   call void @llvm.amdgcn.struct.buffer.store.f32(float %out.1, <4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   ret void
 }
 
 ; Check that Strict WQM is triggered by the strict_wqm intrinsic.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm1:
-;CHECK: s_mov_b64 s{{\[[0-9]+:[0-9]+\]}}, exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
 define amdgpu_ps float @test_strict_wqm1(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wqm1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wqm.f32(float %out)
   ret float %out.0
 }
 
 ; Same as above, but with an integer type.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm2:
-;CHECK: s_mov_b64 s{{\[[0-9]+:[0-9]+\]}}, exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_{{[iu]}}32_e32
 define amdgpu_ps float @test_strict_wqm2(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wqm2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_u32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %src0.0 = bitcast float %src0 to i32
   %src1.0 = bitcast float %src1 to i32
   %out = add i32 %src0.0, %src1.0
   %out.0 = call i32 @llvm.amdgcn.strict.wqm.i32(i32 %out)
   %out.1 = bitcast i32 %out.0 to float
   ret float %out.1
 }
 
 ; Check that we don't leave Strict WQM on for computations that don't require it,
 ; since that will lead clobbering things that aren't supposed to be clobbered
 ; in cases like this.
 ; We enforce this by checking that v_add gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm3:
-;CHECK: %if
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: v_add_f32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wqm3(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_strict_wqm3:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB17_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_mov_b64 s[4:5], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX9-W64-NEXT:  .LBB17_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm3:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB17_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX10-W32-NEXT:  .LBB17_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.strict.wqm.f32(float %out)
   %out.1 = fadd float %src, %out.0
   br label %endif
 
 endif:
   %out.2 = phi float [ %out.1, %if ], [ 0.0, %main_body ]
   ret float %out.2
 }
 
 ; Check that Strict WQM writes aren't coalesced with non-strict writes, since
 ; the Strict WQM write could clobber disabled channels in the non-strict one.
 ; We enforce this by checking that v_mov gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm4:
-;CHECK: %if
-;CHECK: s_mov_b64 s{{\[[0-9]+:[0-9]+\]}}, exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK-NEXT: v_mov_b32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wqm4(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_strict_wqm4:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB18_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_mov_b64 s[4:5], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB18_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm4:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB18_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB18_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.strict.wqm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Make sure the transition from Exact to Strict WQM then WQM works properly.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm5:
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: buffer_store_dword
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: s_wqm_b64 exec, exec
 define amdgpu_ps float @test_strict_wqm5(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wqm5:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 s[4:5], exec
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm5:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %src0, <4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %src1, %src1
   %temp.0 = call float @llvm.amdgcn.strict.wqm.f32(float %temp)
   %out = fadd float %temp.0, %temp.0
   %out.0 = call float @llvm.amdgcn.wqm.f32(float %out)
   ret float %out.0
 }
 
 ; Check that Strict WQM is turned on correctly across basic block boundaries.
 ; if..then..endif version
-;
-;CHECK-LABEL: {{^}}test_strict_wqm6_then:
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %if
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wqm6_then() {
+; GFX9-W64-LABEL: test_strict_wqm6_then:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[0:1], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB20_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB20_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm6_then:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s0, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB20_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_mov_b32 s1, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB20_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wqm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Check that Strict WQM is turned on correctly across basic block boundaries.
 ; loop version
-;
-;CHECK-LABEL: {{^}}test_strict_wqm6_loop:
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %loop
-;CHECK: s_mov_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endloop
 define amdgpu_ps float @test_strict_wqm6_loop() {
+; GFX9-W64-LABEL: test_strict_wqm6_loop:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], 0
+; GFX9-W64-NEXT:  .LBB21_1: ; %loop
+; GFX9-W64-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_add_u32_e32 v3, -1, v3
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v3
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_or_b64 s[0:1], vcc, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    s_andn2_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_cbranch_execnz .LBB21_1
+; GFX9-W64-NEXT:  ; %bb.2: ; %endloop
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm6_loop:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:  .LBB21_1: ; %loop
+; GFX10-W32-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX10-W32-NEXT:    s_mov_b32 s1, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v3, -1, v3
+; GFX10-W32-NEXT:    s_mov_b32 s1, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v3
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_or_b32 s0, vcc_lo, s0
+; GFX10-W32-NEXT:    s_andn2_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_cbranch_execnz .LBB21_1
+; GFX10-W32-NEXT:  ; %bb.2: ; %endloop
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   br label %loop
 
 loop:
   %counter = phi i32 [ %lo, %main_body ], [ %counter.1, %loop ]
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wqm.f32(float %out)
   %counter.1 = sub i32 %counter, 1
   %cc = icmp ne i32 %counter.1, 0
   br i1 %cc, label %loop, label %endloop
 
 endloop:
   ret float %out.0
 }
 
 ; Check that enabling WQM anywhere enables WQM for the set.inactive source.
-;
-;CHECK-LABEL: {{^}}test_set_inactive2:
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
 define amdgpu_ps void @test_set_inactive2(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_set_inactive2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    ; kill: def $vgpr1 killed $vgpr1 def $scc killed $exec
+; GFX9-W64-NEXT:    ; kill: def $vgpr2 killed $vgpr2 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_u32_e32 v1, v2, v1
+; GFX9-W64-NEXT:    buffer_store_dword v1, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_endpgm
+;
+; GFX10-W32-LABEL: test_set_inactive2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    ; kill: def $vgpr2 killed $vgpr2 def $scc killed $exec
+; GFX10-W32-NEXT:    ; kill: def $vgpr1 killed $vgpr1 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    buffer_store_dword v1, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_endpgm
 main_body:
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %src1.0 = bitcast float %src1 to i32
   %src1.1 = call i32 @llvm.amdgcn.set.inactive.i32(i32 %src1.0, i32 undef)
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src0.0 = bitcast float %src0 to i32
   %src0.1 = call i32 @llvm.amdgcn.wqm.i32(i32 %src0.0)
   %out = add i32 %src0.1, %src1.1
   %out.0 = bitcast i32 %out to float
   call void @llvm.amdgcn.struct.buffer.store.f32(float %out.0, <4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   ret void
 }
 
 ; Check a case of one branch of an if-else requiring WQM, the other requiring
 ; exact.
-;
 ; Note: In this particular case, the save-and-restore could be avoided if the
 ; analysis understood that the two branches of the if-else are mutually
 ; exclusive.
-;
-;CHECK-LABEL: {{^}}test_control_flow_0:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: %ELSE
-;CHECK: s_and_saveexec_b64 [[SAVED:s\[[0-9]+:[0-9]+\]]], [[ORIG]]
-;CHECK: store
-;CHECK: s_mov_b64 exec, [[SAVED]]
-;CHECK: %IF
-;CHECK: image_sample
-;CHECK: image_sample
 define amdgpu_ps float @test_control_flow_0(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %c, i32 %z, float %data) {
+; GFX9-W64-LABEL: test_control_flow_0:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_xor_b64 s[14:15], exec, s[14:15]
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB23_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %ELSE
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[16:17], s[12:13]
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[16:17]
+; GFX9-W64-NEXT:  .LBB23_2: ; %Flow
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[14:15], s[14:15]
+; GFX9-W64-NEXT:    s_xor_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB23_4
+; GFX9-W64-NEXT:  ; %bb.3: ; %IF
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v2, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:  .LBB23_4: ; %END
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_control_flow_0:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_ne_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_xor_b32 s13, exec_lo, s13
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB23_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %ELSE
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s14, s12
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s14
+; GFX10-W32-NEXT:  .LBB23_2: ; %Flow
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s13, s13
+; GFX10-W32-NEXT:    s_xor_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB23_4
+; GFX10-W32-NEXT:  ; %bb.3: ; %IF
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v2, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:  .LBB23_4: ; %END
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %cmp = icmp eq i32 %z, 0
   br i1 %cmp, label %IF, label %ELSE
 
 IF:
   %c.bc = bitcast i32 %c to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %data.if = extractelement <4 x float> %dtex, i32 0
   br label %END
 
 ELSE:
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data, <4 x i32> undef, i32 %c, i32 0, i32 0, i32 0)
   br label %END
 
 END:
   %r = phi float [ %data.if, %IF ], [ %data, %ELSE ]
   ret float %r
 }
 
 ; Reverse branch order compared to the previous test.
-;
-;CHECK-LABEL: {{^}}test_control_flow_1:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: %IF
-;CHECK: image_sample
-;CHECK: image_sample
-;CHECK: %Flow
-;CHECK-NEXT: s_or_saveexec_b64 [[SAVED:s\[[0-9]+:[0-9]+\]]],
-;CHECK-NEXT: s_and_b64 exec, exec, [[ORIG]]
-;CHECK-NEXT: s_and_b64 [[SAVED]], exec, [[SAVED]]
-;CHECK-NEXT: s_xor_b64 exec, exec, [[SAVED]]
-;CHECK-NEXT: s_cbranch_execz [[END_BB:.LBB[0-9]+_[0-9]+]]
-;CHECK-NEXT: ; %bb.{{[0-9]+}}: ; %ELSE
-;CHECK: store_dword
-;CHECK: [[END_BB]]: ; %END
-;CHECK: s_or_b64 exec, exec,
-;CHECK: v_mov_b32_e32 v0
-;CHECK: ; return
 define amdgpu_ps float @test_control_flow_1(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %c, i32 %z, float %data) {
+; GFX9-W64-LABEL: test_control_flow_1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_xor_b64 s[14:15], exec, s[14:15]
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB24_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %IF
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v2, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr0
+; GFX9-W64-NEXT:  .LBB24_2: ; %Flow
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_and_b64 s[0:1], exec, s[0:1]
+; GFX9-W64-NEXT:    s_xor_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB24_4
+; GFX9-W64-NEXT:  ; %bb.3: ; %ELSE
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:  .LBB24_4: ; %END
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_control_flow_1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_ne_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_xor_b32 s13, exec_lo, s13
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB24_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %IF
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v2, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0
+; GFX10-W32-NEXT:  .LBB24_2: ; %Flow
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_and_b32 s0, exec_lo, s0
+; GFX10-W32-NEXT:    s_xor_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB24_4
+; GFX10-W32-NEXT:  ; %bb.3: ; %ELSE
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:  .LBB24_4: ; %END
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %cmp = icmp eq i32 %z, 0
   br i1 %cmp, label %ELSE, label %IF
 
 IF:
   %c.bc = bitcast i32 %c to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %data.if = extractelement <4 x float> %dtex, i32 0
   br label %END
 
 ELSE:
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data, <4 x i32> undef, i32 %c, i32 0, i32 0, i32 0)
   br label %END
 
 END:
   %r = phi float [ %data.if, %IF ], [ %data, %ELSE ]
   ret float %r
 }
 
 ; Check that branch conditions are properly marked as needing WQM...
-;
-;CHECK-LABEL: {{^}}test_control_flow_2:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: store
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: load
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: store
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: v_cmp
 define amdgpu_ps <4 x float> @test_control_flow_2(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, <3 x i32> %idx, <2 x float> %data, i32 %coord) {
+; GFX9-W64-LABEL: test_control_flow_2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    buffer_store_dword v3, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    buffer_store_dword v4, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-W64-NEXT:    v_cmp_nlt_f32_e32 vcc, 0, v0
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_xor_b64 s[14:15], exec, s[14:15]
+; GFX9-W64-NEXT:  ; %bb.1: ; %ELSE
+; GFX9-W64-NEXT:    v_lshlrev_b32_e32 v0, 2, v5
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr5
+; GFX9-W64-NEXT:  ; %bb.2: ; %Flow
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[14:15], s[14:15]
+; GFX9-W64-NEXT:    s_xor_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:  ; %bb.3: ; %IF
+; GFX9-W64-NEXT:    v_mul_lo_u32 v0, v5, 3
+; GFX9-W64-NEXT:  ; %bb.4: ; %END
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_control_flow_2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    buffer_store_dword v3, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_cmp_nlt_f32_e32 vcc_lo, 0, v0
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    buffer_store_dword v4, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_xor_b32 s13, exec_lo, s13
+; GFX10-W32-NEXT:  ; %bb.1: ; %ELSE
+; GFX10-W32-NEXT:    v_lshlrev_b32_e32 v0, 2, v5
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr5
+; GFX10-W32-NEXT:  ; %bb.2: ; %Flow
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s13, s13
+; GFX10-W32-NEXT:    s_xor_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:  ; %bb.3: ; %IF
+; GFX10-W32-NEXT:    v_mul_lo_u32 v0, v5, 3
+; GFX10-W32-NEXT:  ; %bb.4: ; %END
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %idx.1 = extractelement <3 x i32> %idx, i32 0
   %data.1 = extractelement <2 x float> %data, i32 0
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data.1, <4 x i32> undef, i32 %idx.1, i32 0, i32 0, i32 0)
 
   ; The load that determines the branch (and should therefore be WQM) is
   ; surrounded by stores that require disabled WQM.
   %idx.2 = extractelement <3 x i32> %idx, i32 1

 END:
   %coord.END = phi i32 [ %coord.IF, %IF ], [ %coord.ELSE, %ELSE ]
   %coord.END.bc = bitcast i32 %coord.END to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord.END.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   ret <4 x float> %tex
 }
 
 ; ... but only if they really do need it.
-;
-;CHECK-LABEL: {{^}}test_control_flow_3:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: image_sample
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK-DAG: v_cmp
-;CHECK-DAG: store
 define amdgpu_ps float @test_control_flow_3(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %idx, float %coord) {
+; GFX9-W64-LABEL: test_control_flow_3:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    image_sample v1, v1, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v1, v1, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_cmp_nlt_f32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    buffer_store_dword v1, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[0:1], vcc
+; GFX9-W64-NEXT:    s_xor_b64 s[0:1], exec, s[0:1]
+; GFX9-W64-NEXT:  ; %bb.1: ; %ELSE
+; GFX9-W64-NEXT:    v_mul_f32_e32 v0, 4.0, v1
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr1
+; GFX9-W64-NEXT:  ; %bb.2: ; %Flow
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], s[0:1]
+; GFX9-W64-NEXT:    s_xor_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:  ; %bb.3: ; %IF
+; GFX9-W64-NEXT:    v_mul_f32_e32 v0, 0x40400000, v1
+; GFX9-W64-NEXT:  ; %bb.4: ; %END
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_control_flow_3:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    image_sample v1, v1, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v1, v1, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_cmp_nlt_f32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    buffer_store_dword v1, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s0, vcc_lo
+; GFX10-W32-NEXT:    s_xor_b32 s0, exec_lo, s0
+; GFX10-W32-NEXT:  ; %bb.1: ; %ELSE
+; GFX10-W32-NEXT:    v_mul_f32_e32 v0, 4.0, v1
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr1
+; GFX10-W32-NEXT:  ; %bb.2: ; %Flow
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, s0
+; GFX10-W32-NEXT:    s_xor_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:  ; %bb.3: ; %IF
+; GFX10-W32-NEXT:    v_mul_f32_e32 v0, 0x40400000, v1
+; GFX10-W32-NEXT:  ; %bb.4: ; %END
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %dtex.1 = extractelement <4 x float> %dtex, i32 0
   call void @llvm.amdgcn.struct.buffer.store.f32(float %dtex.1, <4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
 
   %cc = fcmp ogt float %dtex.1, 0.0
   br i1 %cc, label %IF, label %ELSE
 
 IF:
   %tex.IF = fmul float %dtex.1, 3.0
   br label %END
 
 ELSE:
   %tex.ELSE = fmul float %dtex.1, 4.0
   br label %END
 
 END:
   %tex.END = phi float [ %tex.IF, %IF ], [ %tex.ELSE, %ELSE ]
   ret float %tex.END
 }
 
 ; Another test that failed at some point because of terminator handling.
-;
-;CHECK-LABEL: {{^}}test_control_flow_4:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: %IF
-;CHECK: s_and_saveexec_b64 [[SAVE:s\[[0-9]+:[0-9]+\]]],  [[ORIG]]
-;CHECK: load
-;CHECK: store
-;CHECK: s_mov_b64 exec, [[SAVE]]
-;CHECK: %END
-;CHECK: image_sample
-;CHECK: image_sample
 define amdgpu_ps <4 x float> @test_control_flow_4(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, float %coord, i32 %y, float %z) {
+; GFX9-W64-LABEL: test_control_flow_4:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB27_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %IF
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[16:17], s[12:13]
+; GFX9-W64-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, 1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v1, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[16:17]
+; GFX9-W64-NEXT:  .LBB27_2: ; %END
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_control_flow_4:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB27_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %IF
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s14, s12
+; GFX10-W32-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, 1
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v1, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s14
+; GFX10-W32-NEXT:  .LBB27_2: ; %END
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %cond = icmp eq i32 %y, 0
   br i1 %cond, label %IF, label %END
 
 IF:
   %data = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> undef, i32 0, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data, <4 x i32> undef, i32 1, i32 0, i32 0, i32 0)
   br label %END
 
 END:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   ret <4 x float> %dtex
 }
 
 ; Kill is performed in WQM mode so that uniform kill behaves correctly ...
-;
-;CHECK-LABEL: {{^}}test_kill_0:
-;CHECK-NEXT: ; %main_body
-;CHECK-NEXT: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK-NEXT: s_wqm_b64 exec, exec
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK: buffer_store_dword
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: v_cmp_
-;CHECK: image_sample
-;CHECK: s_and_b64 exec, exec, [[ORIG]]
-;CHECK: image_sample
-;CHECK: buffer_store_dword
 define amdgpu_ps <4 x float> @test_kill_0(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, float addrspace(1)* inreg %ptr, <2 x i32> %idx, <2 x float> %data, float %coord, float %coord2, float %z) {
+; GFX9-W64-LABEL: test_kill_0:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    image_sample v[7:10], v4, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_ngt_f32_e32 vcc, 0, v6
+; GFX9-W64-NEXT:    s_andn2_b64 s[12:13], s[12:13], vcc
+; GFX9-W64-NEXT:    s_cbranch_scc0 .LBB28_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %main_body
+; GFX9-W64-NEXT:    s_andn2_b64 exec, exec, vcc
+; GFX9-W64-NEXT:    image_sample v0, v5, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[11:14], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v7, v11
+; GFX9-W64-NEXT:    buffer_store_dword v3, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v8, v12
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v9, v13
+; GFX9-W64-NEXT:    v_add_f32_e32 v3, v10, v14
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_branch .LBB28_3
+; GFX9-W64-NEXT:  .LBB28_2:
+; GFX9-W64-NEXT:    s_mov_b64 exec, 0
+; GFX9-W64-NEXT:    exp null off, off, off, off done vm
+; GFX9-W64-NEXT:    s_endpgm
+; GFX9-W64-NEXT:  .LBB28_3:
+;
+; GFX10-W32-LABEL: test_kill_0:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    image_sample v[7:10], v4, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_ngt_f32_e32 vcc_lo, 0, v6
+; GFX10-W32-NEXT:    s_andn2_b32 s12, s12, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_scc0 .LBB28_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %main_body
+; GFX10-W32-NEXT:    s_andn2_b32 exec_lo, exec_lo, vcc_lo
+; GFX10-W32-NEXT:    image_sample v0, v5, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[11:14], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    buffer_store_dword v3, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v4, v8, v12
+; GFX10-W32-NEXT:    v_add_f32_e32 v5, v10, v14
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v7, v11
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v9, v13
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, v4
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, v5
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    s_branch .LBB28_3
+; GFX10-W32-NEXT:  .LBB28_2:
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, 0
+; GFX10-W32-NEXT:    exp null off, off, off, off done vm
+; GFX10-W32-NEXT:    s_endpgm
+; GFX10-W32-NEXT:  .LBB28_3:
 main_body:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %idx.0 = extractelement <2 x i32> %idx, i32 0
   %data.0 = extractelement <2 x float> %data, i32 0
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data.0, <4 x i32> undef, i32 %idx.0, i32 0, i32 0, i32 0)
 
   %z.cmp = fcmp olt float %z, 0.0
   call void @llvm.amdgcn.kill(i1 %z.cmp)
 
   %idx.1 = extractelement <2 x i32> %idx, i32 1
   %data.1 = extractelement <2 x float> %data, i32 1
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data.1, <4 x i32> undef, i32 %idx.1, i32 0, i32 0, i32 0)
   %tex2 = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord2, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex2.0 = extractelement <4 x float> %tex2, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex2.0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %out = fadd <4 x float> %tex, %dtex
 
   ret <4 x float> %out
 }
 
 ; ... but only if WQM is necessary.
-;
 ; CHECK-LABEL: {{^}}test_kill_1:
 ; CHECK-NEXT: ; %main_body
 ; CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
 ; CHECK: image_sample
 ; CHECK: s_and_b64 exec, exec, [[ORIG]]
 ; CHECK: image_sample
 ; CHECK-NOT: wqm
 ; CHECK-DAG: buffer_store_dword
 ; CHECK-DAG: v_cmp_
 define amdgpu_ps <4 x float> @test_kill_1(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %idx, float %data, float %coord, float %coord2, float %z) {
+; GFX9-W64-LABEL: test_kill_1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v2
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v5, v0
+; GFX9-W64-NEXT:    image_sample v0, v1, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf
+; GFX9-W64-NEXT:    v_cmp_ngt_f32_e32 vcc, 0, v4
+; GFX9-W64-NEXT:    s_andn2_b64 s[12:13], s[12:13], vcc
+; GFX9-W64-NEXT:    buffer_store_dword v5, off, s[0:3], 0
+; GFX9-W64-NEXT:    s_cbranch_scc0 .LBB29_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %main_body
+; GFX9-W64-NEXT:    s_andn2_b64 exec, exec, vcc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_branch .LBB29_3
+; GFX9-W64-NEXT:  .LBB29_2:
+; GFX9-W64-NEXT:    s_mov_b64 exec, 0
+; GFX9-W64-NEXT:    exp null off, off, off, off done vm
+; GFX9-W64-NEXT:    s_endpgm
+; GFX9-W64-NEXT:  .LBB29_3:
+;
+; GFX10-W32-LABEL: test_kill_1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v2
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v5, v0
+; GFX10-W32-NEXT:    image_sample v0, v1, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    v_cmp_ngt_f32_e32 vcc_lo, 0, v4
+; GFX10-W32-NEXT:    buffer_store_dword v5, off, s[0:3], 0
+; GFX10-W32-NEXT:    s_andn2_b32 s12, s12, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_scc0 .LBB29_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %main_body
+; GFX10-W32-NEXT:    s_andn2_b32 exec_lo, exec_lo, vcc_lo
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    s_branch .LBB29_3
+; GFX10-W32-NEXT:  .LBB29_2:
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, 0
+; GFX10-W32-NEXT:    exp null off, off, off, off done vm
+; GFX10-W32-NEXT:    s_endpgm
+; GFX10-W32-NEXT:  .LBB29_3:
 main_body:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %coord, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
 
   call void @llvm.amdgcn.raw.buffer.store.f32(float %data, <4 x i32> undef, i32 0, i32 0, i32 0)
 
   %z.cmp = fcmp olt float %z, 0.0
   call void @llvm.amdgcn.kill(i1 %z.cmp)
 
   ret <4 x float> %dtex
 }
 
 ; Check prolog shaders.
-;
 ; CHECK-LABEL: {{^}}test_prolog_1:
 ; CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
 ; CHECK: v_add_f32_e32 v0,
 ; CHECK: s_and_b64 exec, exec, [[ORIG]]
 define amdgpu_ps float @test_prolog_1(float %a, float %b) #5 {
+; GFX9-W64-LABEL: test_prolog_1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_prolog_1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %s = fadd float %a, %b
   ret float %s
 }
 
 ; CHECK-LABEL: {{^}}test_loop_vcc:
 ; CHECK-NEXT: ; %entry
 ; CHECK-NEXT: s_mov_b64 [[LIVE:s\[[0-9]+:[0-9]+\]]], exec

 ; CHECK: v_add_f32_e32 [[CTR]], 2.0, [[CTR]]
 ; CHECK: [[LOOP:.LBB[0-9]+_[0-9]+]]: ; %loop
 ; CHECK: v_cmp_lt_f32_e32 vcc, [[SEVEN]], [[CTR]]
 ; CHECK: s_cbranch_vccz [[LOOPHDR]]
 
 ; CHECK: ; %break
 ; CHECK: ; return
 define amdgpu_ps <4 x float> @test_loop_vcc(<4 x float> %in) nounwind {
+; GFX9-W64-LABEL: test_loop_vcc:
+; GFX9-W64:       ; %bb.0: ; %entry
+; GFX9-W64-NEXT:    s_mov_b64 s[8:9], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v7, v3
+; GFX9-W64-NEXT:    v_mov_b32_e32 v6, v2
+; GFX9-W64-NEXT:    v_mov_b32_e32 v5, v1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v0
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[8:9]
+; GFX9-W64-NEXT:    s_mov_b32 s0, 0
+; GFX9-W64-NEXT:    s_mov_b32 s1, s0
+; GFX9-W64-NEXT:    s_mov_b32 s2, s0
+; GFX9-W64-NEXT:    s_mov_b32 s3, s0
+; GFX9-W64-NEXT:    s_mov_b32 s4, s0
+; GFX9-W64-NEXT:    s_mov_b32 s5, s0
+; GFX9-W64-NEXT:    s_mov_b32 s6, s0
+; GFX9-W64-NEXT:    s_mov_b32 s7, s0
+; GFX9-W64-NEXT:    image_store v[4:7], v0, s[0:7] dmask:0xf unorm
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v8, 0
+; GFX9-W64-NEXT:    s_mov_b32 s10, 0x40e00000
+; GFX9-W64-NEXT:    s_branch .LBB31_2
+; GFX9-W64-NEXT:  .LBB31_1: ; %body
+; GFX9-W64-NEXT:    ; in Loop: Header=BB31_2 Depth=1
+; GFX9-W64-NEXT:    s_mov_b32 s1, s0
+; GFX9-W64-NEXT:    s_mov_b32 s2, s0
+; GFX9-W64-NEXT:    s_mov_b32 s3, s0
+; GFX9-W64-NEXT:    s_mov_b32 s4, s0
+; GFX9-W64-NEXT:    s_mov_b32 s5, s0
+; GFX9-W64-NEXT:    s_mov_b32 s6, s0
+; GFX9-W64-NEXT:    s_mov_b32 s7, s0
+; GFX9-W64-NEXT:    image_sample v[4:7], v0, s[0:7], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:    v_add_f32_e32 v8, 2.0, v8
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], 0
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB31_4
+; GFX9-W64-NEXT:  .LBB31_2: ; %loop
+; GFX9-W64-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v4
+; GFX9-W64-NEXT:    v_cmp_lt_f32_e32 vcc, s10, v8
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, v5
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, v6
+; GFX9-W64-NEXT:    v_mov_b32_e32 v3, v7
+; GFX9-W64-NEXT:    s_and_b64 vcc, exec, vcc
+; GFX9-W64-NEXT:    s_cbranch_vccz .LBB31_1
+; GFX9-W64-NEXT:  ; %bb.3:
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], -1
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr4_vgpr5_vgpr6_vgpr7
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr8
+; GFX9-W64-NEXT:  .LBB31_4: ; %break
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[8:9]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_loop_vcc:
+; GFX10-W32:       ; %bb.0: ; %entry
+; GFX10-W32-NEXT:    s_mov_b32 s8, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v8, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s8
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    image_store v[0:3], v0, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_1D unorm
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_branch .LBB31_2
+; GFX10-W32-NEXT:    .p2align 6
+; GFX10-W32-NEXT:  .LBB31_1: ; %body
+; GFX10-W32-NEXT:    ; in Loop: Header=BB31_2 Depth=1
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    v_add_f32_e32 v8, 2.0, v8
+; GFX10-W32-NEXT:    image_sample v[0:3], v4, s[0:7], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_mov_b32 s1, 0
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB31_4
+; GFX10-W32-NEXT:  .LBB31_2: ; %loop
+; GFX10-W32-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX10-W32-NEXT:    v_cmp_lt_f32_e32 vcc_lo, 0x40e00000, v8
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v7, v3
+; GFX10-W32-NEXT:    v_mov_b32_e32 v6, v2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v5, v1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v0
+; GFX10-W32-NEXT:    s_and_b32 vcc_lo, exec_lo, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_vccz .LBB31_1
+; GFX10-W32-NEXT:  ; %bb.3:
+; GFX10-W32-NEXT:    s_mov_b32 s1, -1
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0_vgpr1_vgpr2_vgpr3
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr8
+; GFX10-W32-NEXT:  .LBB31_4: ; %break
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s8
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v4
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, v5
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, v6
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, v7
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 entry:
   call void @llvm.amdgcn.image.store.1d.v4f32.i32(<4 x float> %in, i32 15, i32 undef, <8 x i32> undef, i32 0, i32 0)
   br label %loop
 
 loop:
   %ctr.iv = phi float [ 0.0, %entry ], [ %ctr.next, %body ]
   %c.iv = phi <4 x float> [ %in, %entry ], [ %c.next, %body ]
   %cc = fcmp ogt float %ctr.iv, 7.0
   br i1 %cc, label %break, label %body
 
 body:
   %c.iv0 = extractelement <4 x float> %c.iv, i32 0
   %c.next = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.iv0, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   %ctr.next = fadd float %ctr.iv, 2.0
   br label %loop
 
 break:
   ret <4 x float> %c.iv
 }
 
 ; Only intrinsic stores need exact execution -- other stores do not have
 ; externally visible effects and may require WQM for correctness.
-;
 ; CHECK-LABEL: {{^}}test_alloca:
 ; CHECK: s_mov_b64 [[LIVE:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
 
 ; CHECK: s_and_b64 exec, exec, [[LIVE]]
 ; CHECK: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0
 ; CHECK: s_wqm_b64 exec, exec
 ; CHECK: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:4{{$}}
 ; CHECK: s_and_b64 exec, exec, [[LIVE]]
 ; CHECK: buffer_store_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 idxen
 ; CHECK: s_wqm_b64 exec, exec
 ; CHECK: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 offen
 
 ; CHECK: s_and_b64 exec, exec, [[LIVE]]
 ; CHECK: image_sample
 ; CHECK: buffer_store_dwordx4
 define amdgpu_ps void @test_alloca(float %data, i32 %a, i32 %idx) nounwind {
+; GFX9-W64-LABEL: test_alloca:
+; GFX9-W64:       ; %bb.0: ; %entry
+; GFX9-W64-NEXT:    s_mov_b32 s8, SCRATCH_RSRC_DWORD0
+; GFX9-W64-NEXT:    s_mov_b32 s9, SCRATCH_RSRC_DWORD1
+; GFX9-W64-NEXT:    s_mov_b32 s10, -1
+; GFX9-W64-NEXT:    s_mov_b32 s11, 0xe00000
+; GFX9-W64-NEXT:    s_add_u32 s8, s8, s0
+; GFX9-W64-NEXT:    s_addc_u32 s9, s9, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    buffer_store_dword v1, off, s[8:11], 0 offset:4
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, 1
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 4
+; GFX9-W64-NEXT:    v_lshl_add_u32 v0, v2, 2, v0
+; GFX9-W64-NEXT:    buffer_load_dword v0, v0, s[8:11], 0 offen
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_mov_b32 s0, 0
+; GFX9-W64-NEXT:    s_mov_b32 s1, s0
+; GFX9-W64-NEXT:    s_mov_b32 s2, s0
+; GFX9-W64-NEXT:    s_mov_b32 s3, s0
+; GFX9-W64-NEXT:    s_mov_b32 s4, s0
+; GFX9-W64-NEXT:    s_mov_b32 s5, s0
+; GFX9-W64-NEXT:    s_mov_b32 s6, s0
+; GFX9-W64-NEXT:    s_mov_b32 s7, s0
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
+; GFX9-W64-NEXT:    s_endpgm
+;
+; GFX10-W32-LABEL: test_alloca:
+; GFX10-W32:       ; %bb.0: ; %entry
+; GFX10-W32-NEXT:    s_mov_b32 s8, SCRATCH_RSRC_DWORD0
+; GFX10-W32-NEXT:    s_mov_b32 s9, SCRATCH_RSRC_DWORD1
+; GFX10-W32-NEXT:    s_mov_b32 s10, -1
+; GFX10-W32-NEXT:    s_mov_b32 s11, 0x31c16000
+; GFX10-W32-NEXT:    s_add_u32 s8, s8, s0
+; GFX10-W32-NEXT:    s_addc_u32 s9, s9, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, 1
+; GFX10-W32-NEXT:    v_lshl_add_u32 v2, v2, 2, 4
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_store_dword v1, off, s[8:11], 0 offset:4
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    buffer_store_dword v0, v3, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v0, v2, s[8:11], 0 offen
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
+; GFX10-W32-NEXT:    s_endpgm
 entry:
   %array = alloca [32 x i32], align 4, addrspace(5)
 
   call void @llvm.amdgcn.raw.buffer.store.f32(float %data, <4 x i32> undef, i32 0, i32 0, i32 0)
 
   %s.gep = getelementptr [32 x i32], [32 x i32] addrspace(5)* %array, i32 0, i32 0
   store volatile i32 %a, i32 addrspace(5)* %s.gep, align 4
 
   call void @llvm.amdgcn.struct.buffer.store.f32(float %data, <4 x i32> undef, i32 1, i32 0, i32 0, i32 0)
 
   %c.gep = getelementptr [32 x i32], [32 x i32] addrspace(5)* %array, i32 0, i32 %idx
   %c = load i32, i32 addrspace(5)* %c.gep, align 4
   %c.bc = bitcast i32 %c to float
   %t = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   call void @llvm.amdgcn.raw.buffer.store.v4f32(<4 x float> %t, <4 x i32> undef, i32 0, i32 0, i32 0)
 
   ret void
 }
 
 ; Must return to exact at the end of a non-void returning shader,
 ; otherwise the EXEC mask exported by the epilog will be wrong. This is true
 ; even if the shader has no kills, because a kill could have happened in a
 ; previous shader fragment.
-;
 ; CHECK-LABEL: {{^}}test_nonvoid_return:
 ; CHECK: s_mov_b64 [[LIVE:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
-;
 ; CHECK: s_and_b64 exec, exec, [[LIVE]]
 ; CHECK-NOT: exec
 define amdgpu_ps <4 x float> @test_nonvoid_return() nounwind {
+; GFX9-W64-LABEL: test_nonvoid_return:
+; GFX9-W64:       ; %bb.0:
+; GFX9-W64-NEXT:    s_mov_b32 s0, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[8:9], exec
+; GFX9-W64-NEXT:    s_mov_b32 s1, s0
+; GFX9-W64-NEXT:    s_mov_b32 s2, s0
+; GFX9-W64-NEXT:    s_mov_b32 s3, s0
+; GFX9-W64-NEXT:    s_mov_b32 s4, s0
+; GFX9-W64-NEXT:    s_mov_b32 s5, s0
+; GFX9-W64-NEXT:    s_mov_b32 s6, s0
+; GFX9-W64-NEXT:    s_mov_b32 s7, s0
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[0:3] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[8:9]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[0:7], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_nonvoid_return:
+; GFX10-W32:       ; %bb.0:
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s8, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[0:3] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s8
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    ; return to shader part epilog
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float undef, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   ret <4 x float> %dtex
 }
 
 ; CHECK-LABEL: {{^}}test_nonvoid_return_unreachable:
 ; CHECK: s_mov_b64 [[LIVE:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
-;
 ; CHECK: s_and_b64 exec, exec, [[LIVE]]
 ; CHECK-NOT: exec
 define amdgpu_ps <4 x float> @test_nonvoid_return_unreachable(i32 inreg %c) nounwind {
+; GFX9-W64-LABEL: test_nonvoid_return_unreachable:
+; GFX9-W64:       ; %bb.0: ; %entry
+; GFX9-W64-NEXT:    s_mov_b32 s4, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_mov_b32 s5, s4
+; GFX9-W64-NEXT:    s_mov_b32 s6, s4
+; GFX9-W64-NEXT:    s_mov_b32 s7, s4
+; GFX9-W64-NEXT:    s_mov_b32 s8, s4
+; GFX9-W64-NEXT:    s_mov_b32 s9, s4
+; GFX9-W64-NEXT:    s_mov_b32 s10, s4
+; GFX9-W64-NEXT:    s_mov_b32 s11, s4
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    image_sample v0, v0, s[4:11], s[0:3] dmask:0x1
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[4:11], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:    s_cmp_lt_i32 s0, 1
+; GFX9-W64-NEXT:    s_cbranch_scc0 .LBB34_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %else
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_branch .LBB34_3
+; GFX9-W64-NEXT:  .LBB34_2: ; %if
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:  .LBB34_3:
+;
+; GFX10-W32-LABEL: test_nonvoid_return_unreachable:
+; GFX10-W32:       ; %bb.0: ; %entry
+; GFX10-W32-NEXT:    s_mov_b32 s4, 0
+; GFX10-W32-NEXT:    s_mov_b32 s1, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s5, s4
+; GFX10-W32-NEXT:    s_mov_b32 s6, s4
+; GFX10-W32-NEXT:    s_mov_b32 s7, s4
+; GFX10-W32-NEXT:    s_mov_b32 s8, s4
+; GFX10-W32-NEXT:    s_mov_b32 s9, s4
+; GFX10-W32-NEXT:    s_mov_b32 s10, s4
+; GFX10-W32-NEXT:    s_mov_b32 s11, s4
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    image_sample v0, v0, s[4:11], s[0:3] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[4:11], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_cmp_lt_i32 s0, 1
+; GFX10-W32-NEXT:    s_cbranch_scc0 .LBB34_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %else
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_branch .LBB34_3
+; GFX10-W32-NEXT:  .LBB34_2: ; %if
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:  .LBB34_3:
 entry:
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float undef, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   %cc = icmp sgt i32 %c, 0
   br i1 %cc, label %if, label %else
 
 if:
   store volatile <4 x float> %dtex, <4 x float> addrspace(1)* undef
   unreachable
 
 else:
   ret <4 x float> %dtex
 }
 
 ; Test awareness that s_wqm_b64 clobbers SCC.
-;
 ; CHECK-LABEL: {{^}}test_scc:
 ; CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
 ; CHECK: s_wqm_b64 exec, exec
 ; CHECK: s_cmp_
 ; CHECK-NEXT: s_cbranch_scc
 ; CHECK: ; %else
 ; CHECK: image_sample
 ; CHECK: ; %if
 ; CHECK: image_sample
 ; CHECK: ; %end
 ; CHECK: s_and_b64 exec, exec, [[ORIG]]
 define amdgpu_ps <4 x float> @test_scc(i32 inreg %sel, i32 %idx) #1 {
+; GFX9-W64-LABEL: test_scc:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v0
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_cmp_lt_i32 s0, 1
+; GFX9-W64-NEXT:    s_cbranch_scc0 .LBB35_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %else
+; GFX9-W64-NEXT:    s_mov_b32 s4, 0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, 1
+; GFX9-W64-NEXT:    s_mov_b32 s5, s4
+; GFX9-W64-NEXT:    s_mov_b32 s6, s4
+; GFX9-W64-NEXT:    s_mov_b32 s7, s4
+; GFX9-W64-NEXT:    s_mov_b32 s8, s4
+; GFX9-W64-NEXT:    s_mov_b32 s9, s4
+; GFX9-W64-NEXT:    s_mov_b32 s10, s4
+; GFX9-W64-NEXT:    s_mov_b32 s11, s4
+; GFX9-W64-NEXT:    image_sample v[0:3], v[0:1], s[4:11], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB35_3
+; GFX9-W64-NEXT:    s_branch .LBB35_4
+; GFX9-W64-NEXT:  .LBB35_2:
+; GFX9-W64-NEXT:    ; implicit-def: $vgpr0_vgpr1_vgpr2_vgpr3
+; GFX9-W64-NEXT:  .LBB35_3: ; %if
+; GFX9-W64-NEXT:    s_mov_b32 s4, 0
+; GFX9-W64-NEXT:    s_mov_b32 s5, s4
+; GFX9-W64-NEXT:    s_mov_b32 s6, s4
+; GFX9-W64-NEXT:    s_mov_b32 s7, s4
+; GFX9-W64-NEXT:    s_mov_b32 s8, s4
+; GFX9-W64-NEXT:    s_mov_b32 s9, s4
+; GFX9-W64-NEXT:    s_mov_b32 s10, s4
+; GFX9-W64-NEXT:    s_mov_b32 s11, s4
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    image_sample v[0:3], v0, s[4:11], s[0:3] dmask:0xf
+; GFX9-W64-NEXT:  .LBB35_4: ; %end
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v5, 1.0
+; GFX9-W64-NEXT:    buffer_store_dword v5, v4, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_scc:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v0
+; GFX10-W32-NEXT:    s_mov_b32 s8, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_cmp_lt_i32 s0, 1
+; GFX10-W32-NEXT:    s_cbranch_scc0 .LBB35_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %else
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, 1
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    image_sample v[0:3], v[0:1], s[0:7], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_2D
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB35_3
+; GFX10-W32-NEXT:    s_branch .LBB35_4
+; GFX10-W32-NEXT:  .LBB35_2:
+; GFX10-W32-NEXT:    ; implicit-def: $vgpr0_vgpr1_vgpr2_vgpr3
+; GFX10-W32-NEXT:  .LBB35_3: ; %if
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:    s_mov_b32 s1, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, s0
+; GFX10-W32-NEXT:    s_mov_b32 s3, s0
+; GFX10-W32-NEXT:    s_mov_b32 s4, s0
+; GFX10-W32-NEXT:    s_mov_b32 s5, s0
+; GFX10-W32-NEXT:    s_mov_b32 s6, s0
+; GFX10-W32-NEXT:    s_mov_b32 s7, s0
+; GFX10-W32-NEXT:    image_sample v[0:3], v0, s[0:7], s[0:3] dmask:0xf dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:  .LBB35_4: ; %end
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s8
+; GFX10-W32-NEXT:    v_mov_b32_e32 v5, 1.0
+; GFX10-W32-NEXT:    buffer_store_dword v5, v4, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %cc = icmp sgt i32 %sel, 0
   br i1 %cc, label %if, label %else
 
 if:
   %r.if = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float 0.0, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   br label %end
 
 else:
   %r.else = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f32(i32 15, float 0.0, float bitcast (i32 1 to float), <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) #0
   br label %end
 
 end:
   %r = phi <4 x float> [ %r.if, %if ], [ %r.else, %else ]
   call void @llvm.amdgcn.struct.buffer.store.f32(float 1.0, <4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   ret <4 x float> %r
 }
 
 ; Check a case of a block being entirely WQM except for a bit of WWM.
 ; There was a bug where it forgot to enter and leave WWM.
-;
-;CHECK-LABEL: {{^}}test_wwm_within_wqm:
-;CHECK: %IF
-;CHECK: s_or_saveexec_b64 {{.*}}, -1
-;CHECK: ds_swizzle
-;
 define amdgpu_ps float @test_wwm_within_wqm(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %c, i32 %z, float %data) {
+; GFX9-W64-LABEL: test_wwm_within_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB36_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %IF
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_cvt_i32_f32_e32 v2, v0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, 0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_cvt_f32_i32_e32 v1, v0
+; GFX9-W64-NEXT:  .LBB36_2: ; %ENDIF
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wwm_within_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB36_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %IF
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_cvt_i32_f32_e32 v2, v0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, 0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_cvt_f32_i32_e32 v1, v0
+; GFX10-W32-NEXT:  .LBB36_2: ; %ENDIF
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %c.bc = bitcast i32 %c to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %cmp = icmp eq i32 %z, 0
   br i1 %cmp, label %IF, label %ENDIF
 
 IF:
   %dataf = extractelement <4 x float> %dtex, i32 0
   %data1 = fptosi float %dataf to i32
   %data2 = call i32 @llvm.amdgcn.set.inactive.i32(i32 %data1, i32 0)
   %data3 = call i32 @llvm.amdgcn.ds.swizzle(i32 %data2, i32 2079)
   %data4 = call i32 @llvm.amdgcn.wwm.i32(i32 %data3)
   %data4f = sitofp i32 %data4 to float
   br label %ENDIF
 
 ENDIF:
   %r = phi float [ %data4f, %IF ], [ 0.0, %main_body ]
   ret float %r
 }
 
 ; Check that WWM is triggered by the strict_wwm intrinsic.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm1:
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
 define amdgpu_ps float @test_strict_wwm1(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wwm1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wwm.f32(float %out)
   ret float %out.0
 }
 
 ; Same as above, but with an integer type.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm2:
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: buffer_load_dword
-;CHECK: buffer_load_dword
-;CHECK: v_add_{{[iu]}}32_e32
 define amdgpu_ps float @test_strict_wwm2(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wwm2:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_nop 0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_u32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm2:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_clause 0x1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %src0.0 = bitcast float %src0 to i32
   %src1.0 = bitcast float %src1 to i32
   %out = add i32 %src0.0, %src1.0
   %out.0 = call i32 @llvm.amdgcn.strict.wwm.i32(i32 %out)
   %out.1 = bitcast i32 %out.0 to float
   ret float %out.1
 }
 
 ; Check that we don't leave WWM on for computations that don't require WWM,
 ; since that will lead clobbering things that aren't supposed to be clobbered
 ; in cases like this.
 ; We enforce this by checking that v_add gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm3:
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: v_add_f32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wwm3(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_strict_wwm3:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB39_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX9-W64-NEXT:  .LBB39_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm3:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB39_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v1, v0
+; GFX10-W32-NEXT:  .LBB39_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.strict.wwm.f32(float %out)
   %out.1 = fadd float %src, %out.0
   br label %endif
 
 endif:
   %out.2 = phi float [ %out.1, %if ], [ 0.0, %main_body ]
   ret float %out.2
 }
 
 ; Check that WWM writes aren't coalesced with non-WWM writes, since the WWM
 ; write could clobber disabled channels in the non-WWM one.
 ; We enforce this by checking that v_mov gets emitted in the same block as
 ; WWM computations.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm4:
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK-NEXT: v_mov_b32_e32
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wwm4(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_strict_wwm4:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB40_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB40_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm4:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s1, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB40_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s2
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB40_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %out = fadd float %src, %src
   %out.0 = call float @llvm.amdgcn.strict.wwm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Make sure the transition from Exact to WWM then WQM works properly.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm5:
-;CHECK: buffer_load_dword
-;CHECK: buffer_store_dword
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: s_wqm_b64 exec, exec
 define amdgpu_ps float @test_strict_wwm5(i32 inreg %idx0, i32 inreg %idx1) {
+; GFX9-W64-LABEL: test_strict_wwm5:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-W64-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[4:5], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[4:5]
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[2:3]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm5:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-W32-NEXT:    s_mov_b32 s2, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v2, v0, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    buffer_load_dword v1, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    ; kill: def $vgpr0 killed $vgpr0 killed $exec killed $exec
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s2
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %src0, <4 x i32> undef, i32 %idx0, i32 0, i32 0, i32 0)
   %src1 = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %src1, %src1
   %temp.0 = call float @llvm.amdgcn.strict.wwm.f32(float %temp)
   %out = fadd float %temp.0, %temp.0
   %out.0 = call float @llvm.amdgcn.wqm.f32(float %out)
   ret float %out.0
 }
 
 ; Check that WWM is turned on correctly across basic block boundaries.
 ; if..then..endif version
-;
-;CHECK-LABEL: {{^}}test_strict_wwm6_then:
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %if
-;CHECK: s_or_saveexec_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], -1
 ;SI-CHECK: buffer_load_dword
 ;VI-CHECK: flat_load_dword
-;CHECK: v_add_f32_e32
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endif
 define amdgpu_ps float @test_strict_wwm6_then() {
+; GFX9-W64-LABEL: test_strict_wwm6_then:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX9-W64-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX9-W64-NEXT:    v_cmp_gt_u32_e32 vcc, 32, v0
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[0:1], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB42_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %if
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:  .LBB42_2: ; %endif
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm6_then:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v0, -1, 0
+; GFX10-W32-NEXT:    v_mbcnt_hi_u32_b32 v0, -1, v0
+; GFX10-W32-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 32, v0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s0, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB42_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %if
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v1, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:  .LBB42_2: ; %endif
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   %cc = icmp uge i32 %hi, 32
   br i1 %cc, label %endif, label %if
 
 if:
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wwm.f32(float %out)
   br label %endif
 
 endif:
   %out.1 = phi float [ %out.0, %if ], [ 0.0, %main_body ]
   ret float %out.1
 }
 
 ; Check that WWM is turned on correctly across basic block boundaries.
 ; loop version
-;
-;CHECK-LABEL: {{^}}test_strict_wwm6_loop:
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;SI-CHECK: buffer_load_dword
-;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-;CHECK: %loop
-;CHECK: s_or_saveexec_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], -1
-;SI-CHECK: buffer_load_dword
-;VI-CHECK: flat_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-;CHECK: %endloop
 define amdgpu_ps float @test_strict_wwm6_loop() {
+; GFX9-W64-LABEL: test_strict_wwm6_loop:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    global_load_dword v1, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], 0
+; GFX9-W64-NEXT:  .LBB43_1: ; %loop
+; GFX9-W64-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    global_load_dword v2, v[3:4], off glc
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    v_add_u32_e32 v3, -1, v3
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v3
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_or_b64 s[0:1], vcc, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    s_andn2_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    s_cbranch_execnz .LBB43_1
+; GFX9-W64-NEXT:  ; %bb.2: ; %endloop
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[0:1]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm6_loop:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    global_load_dword v1, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mbcnt_lo_u32_b32 v3, -1, 0
+; GFX10-W32-NEXT:    s_mov_b32 s0, 0
+; GFX10-W32-NEXT:  .LBB43_1: ; %loop
+; GFX10-W32-NEXT:    ; =>This Inner Loop Header: Depth=1
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    global_load_dword v2, v[3:4], off glc dlc
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v3, -1, v3
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v1, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v3
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_or_b32 s0, vcc_lo, s0
+; GFX10-W32-NEXT:    s_andn2_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    s_cbranch_execnz .LBB43_1
+; GFX10-W32-NEXT:  ; %bb.2: ; %endloop
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %src0 = load volatile float, float addrspace(1)* undef
   ; use mbcnt to make sure the branch is divergent
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   br label %loop
 
 loop:
   %counter = phi i32 [ %lo, %main_body ], [ %counter.1, %loop ]
   %src1 = load volatile float, float addrspace(1)* undef
   %out = fadd float %src0, %src1
   %out.0 = call float @llvm.amdgcn.strict.wwm.f32(float %out)
   %counter.1 = sub i32 %counter, 1
   %cc = icmp ne i32 %counter.1, 0
   br i1 %cc, label %loop, label %endloop
 
 endloop:
   ret float %out.0
 }
 
 ; Check that @llvm.amdgcn.set.inactive disables WWM.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm_set_inactive1:
-;CHECK: buffer_load_dword
-;CHECK: s_not_b64 exec, exec
-;CHECK: v_mov_b32_e32
-;CHECK: s_not_b64 exec, exec
-;CHECK: s_or_saveexec_b64 s{{\[[0-9]+:[0-9]+\]}}, -1
-;CHECK: v_add_{{[iu]}}32_e32
 define amdgpu_ps void @test_strict_wwm_set_inactive1(i32 inreg %idx) {
+; GFX9-W64-LABEL: test_strict_wwm_set_inactive1:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    v_add_u32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, v0
+; GFX9-W64-NEXT:    buffer_store_dword v2, v1, s[0:3], 0 idxen
+; GFX9-W64-NEXT:    s_endpgm
+;
+; GFX10-W32-LABEL: test_strict_wwm_set_inactive1:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    v_add_nc_u32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, v0
+; GFX10-W32-NEXT:    buffer_store_dword v2, v1, s[0:3], 0 idxen
+; GFX10-W32-NEXT:    s_endpgm
 main_body:
   %src = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   %src.0 = bitcast float %src to i32
   %src.1 = call i32 @llvm.amdgcn.set.inactive.i32(i32 %src.0, i32 0)
   %out = add i32 %src.1, %src.1
   %out.0 = call i32 @llvm.amdgcn.strict.wwm.i32(i32 %out)
   %out.1 = bitcast i32 %out.0 to float
   call void @llvm.amdgcn.struct.buffer.store.f32(float %out.1, <4 x i32> undef, i32 %idx, i32 0, i32 0, i32 0)
   ret void
 }
 
 ; Check a case of a block being entirely WQM except for a bit of WWM.
 ; There was a bug where it forgot to enter and leave WWM.
-;
-;CHECK-LABEL: {{^}}test_strict_wwm_within_wqm:
-;CHECK: %IF
-;CHECK: s_or_saveexec_b64 {{.*}}, -1
-;CHECK: ds_swizzle
-;
 define amdgpu_ps float @test_strict_wwm_within_wqm(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %c, i32 %z, float %data) {
+; GFX9-W64-LABEL: test_strict_wwm_within_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB45_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %IF
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_cvt_i32_f32_e32 v2, v0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, 0
+; GFX9-W64-NEXT:    s_not_b64 exec, exec
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_cvt_f32_i32_e32 v1, v0
+; GFX9-W64-NEXT:  .LBB45_2: ; %ENDIF
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm_within_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB45_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %IF
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v0, v0, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_cvt_i32_f32_e32 v2, v0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, 0
+; GFX10-W32-NEXT:    s_not_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_cvt_f32_i32_e32 v1, v0
+; GFX10-W32-NEXT:  .LBB45_2: ; %ENDIF
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v1
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %c.bc = bitcast i32 %c to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %cmp = icmp eq i32 %z, 0
   br i1 %cmp, label %IF, label %ENDIF
 
 IF:
   %dataf = extractelement <4 x float> %dtex, i32 0
   %data1 = fptosi float %dataf to i32
   %data2 = call i32 @llvm.amdgcn.set.inactive.i32(i32 %data1, i32 0)
   %data3 = call i32 @llvm.amdgcn.ds.swizzle(i32 %data2, i32 2079)
   %data4 = call i32 @llvm.amdgcn.strict.wwm.i32(i32 %data3)
   %data4f = sitofp i32 %data4 to float
   br label %ENDIF
 
 ENDIF:
   %r = phi float [ %data4f, %IF ], [ 0.0, %main_body ]
   ret float %r
 }
 
 ; Check a case of a block being entirely WQM except for a bit of STRICT WQM.
-;
-;CHECK-LABEL: {{^}}test_strict_wqm_within_wqm:
-;CHECK: %IF
-;CHECK: s_mov_b64 s{{\[[0-9]+:[0-9]+\]}}, exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: ds_swizzle
-;
 define amdgpu_ps float @test_strict_wqm_within_wqm(<8 x i32> inreg %rsrc, <4 x i32> inreg %sampler, i32 %c, i32 %z, float %data) {
+; GFX9-W64-LABEL: test_strict_wqm_within_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[12:13], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b64 s[14:15], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, v0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[14:15]
+; GFX9-W64-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-W64-NEXT:    s_and_saveexec_b64 s[14:15], vcc
+; GFX9-W64-NEXT:    s_cbranch_execz .LBB46_2
+; GFX9-W64-NEXT:  ; %bb.1: ; %IF
+; GFX9-W64-NEXT:    s_mov_b64 s[16:17], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    image_sample v2, v2, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    image_sample v2, v2, s[0:7], s[8:11] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_cvt_i32_f32_e32 v2, v2
+; GFX9-W64-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[16:17]
+; GFX9-W64-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_cvt_f32_i32_e32 v0, v0
+; GFX9-W64-NEXT:  .LBB46_2: ; %ENDIF
+; GFX9-W64-NEXT:    s_or_b64 exec, exec, s[14:15]
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[12:13]
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm_within_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s12, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s13, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, v0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s13
+; GFX10-W32-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, 0
+; GFX10-W32-NEXT:    s_and_saveexec_b32 s13, vcc_lo
+; GFX10-W32-NEXT:    s_cbranch_execz .LBB46_2
+; GFX10-W32-NEXT:  ; %bb.1: ; %IF
+; GFX10-W32-NEXT:    s_mov_b32 s14, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    image_sample v2, v2, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    image_sample v2, v2, s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_cvt_i32_f32_e32 v2, v2
+; GFX10-W32-NEXT:    ds_swizzle_b32 v2, v2 offset:swizzle(SWAP,2)
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s14
+; GFX10-W32-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    v_cvt_f32_i32_e32 v0, v0
+; GFX10-W32-NEXT:  .LBB46_2: ; %ENDIF
+; GFX10-W32-NEXT:    s_or_b32 exec_lo, exec_lo, s13
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s12
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   %c.bc = bitcast i32 %c to float
   %tex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %c.bc, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %tex0 = extractelement <4 x float> %tex, i32 0
   %dtex = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %tex0, <8 x i32> %rsrc, <4 x i32> %sampler, i1 false, i32 0, i32 0) #0
   %cmp = icmp eq i32 %z, 0
   br i1 %cmp, label %IF, label %ENDIF
 
 IF:
   %dataf = extractelement <4 x float> %dtex, i32 0
   %data1 = fptosi float %dataf to i32
   %data2 = call i32 @llvm.amdgcn.ds.swizzle(i32 %data1, i32 2079)
   %data3 = call i32 @llvm.amdgcn.strict.wqm.i32(i32 %data2)
   %data3f = sitofp i32 %data3 to float
   br label %ENDIF
 
 ENDIF:
   %r = phi float [ %data3f, %IF ], [ 0.0, %main_body ]
   ret float %r
 }
 
-;CHECK-LABEL: {{^}}test_strict_wqm_strict_wwm_wqm:
-;CHECK: buffer_store_dword
-
-;CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-
-;CHECK: s_or_saveexec_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-
-;CHECK: s_mov_b64 [[ORIG3:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: v_add
-;CHECK: s_mov_b64 exec, [[ORIG3]]
-
 ;TODO: StrictWQM -> WQM transition could be improved. WQM could use the exec from the previous state instead of calling s_wqm again.
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: image_sample
-
 define amdgpu_ps float @test_strict_wqm_strict_wwm_wqm(i32 inreg %idx0, i32 inreg %idx1, <4 x i32> inreg %res, <4 x i32> inreg %res2, float %inp, <8 x i32> inreg %res3) {
+; GFX9-W64-LABEL: test_strict_wqm_strict_wwm_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[28:29], exec
+; GFX9-W64-NEXT:    s_mov_b32 s19, s17
+; GFX9-W64-NEXT:    s_mov_b64 s[30:31], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b32 s23, s5
+; GFX9-W64-NEXT:    s_mov_b32 s22, s4
+; GFX9-W64-NEXT:    s_mov_b32 s21, s3
+; GFX9-W64-NEXT:    s_mov_b32 s20, s2
+; GFX9-W64-NEXT:    s_mov_b32 s27, s9
+; GFX9-W64-NEXT:    s_mov_b32 s26, s8
+; GFX9-W64-NEXT:    s_mov_b32 s25, s7
+; GFX9-W64-NEXT:    s_mov_b32 s24, s6
+; GFX9-W64-NEXT:    s_mov_b32 s18, s16
+; GFX9-W64-NEXT:    s_mov_b32 s17, s15
+; GFX9-W64-NEXT:    s_mov_b32 s16, s14
+; GFX9-W64-NEXT:    s_mov_b32 s15, s13
+; GFX9-W64-NEXT:    s_mov_b32 s14, s12
+; GFX9-W64-NEXT:    s_mov_b32 s13, s11
+; GFX9-W64-NEXT:    s_mov_b32 s12, s10
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[30:31]
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[20:23], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    buffer_load_dword v2, v1, s[20:23], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-W64-NEXT:    buffer_load_dword v3, v3, s[24:27], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v2, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v3
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v4
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[28:29]
+; GFX9-W64-NEXT:    image_sample v0, v0, s[12:19], s[20:23] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[20:23], 0 idxen
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[20:23], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wqm_strict_wwm_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s28, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s19, s17
+; GFX10-W32-NEXT:    s_mov_b32 s29, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s23, s5
+; GFX10-W32-NEXT:    s_mov_b32 s22, s4
+; GFX10-W32-NEXT:    s_mov_b32 s21, s3
+; GFX10-W32-NEXT:    s_mov_b32 s20, s2
+; GFX10-W32-NEXT:    s_mov_b32 s27, s9
+; GFX10-W32-NEXT:    s_mov_b32 s26, s8
+; GFX10-W32-NEXT:    s_mov_b32 s25, s7
+; GFX10-W32-NEXT:    s_mov_b32 s24, s6
+; GFX10-W32-NEXT:    s_mov_b32 s18, s16
+; GFX10-W32-NEXT:    s_mov_b32 s17, s15
+; GFX10-W32-NEXT:    s_mov_b32 s16, s14
+; GFX10-W32-NEXT:    s_mov_b32 s15, s13
+; GFX10-W32-NEXT:    s_mov_b32 s14, s12
+; GFX10-W32-NEXT:    s_mov_b32 s13, s11
+; GFX10-W32-NEXT:    s_mov_b32 s12, s10
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s29
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[20:23], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 s1, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v2, v1, s[20:23], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, s0
+; GFX10-W32-NEXT:    buffer_load_dword v3, v3, s[24:27], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s1
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(1)
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v2, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v3
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v4
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s28
+; GFX10-W32-NEXT:    image_sample v0, v0, s[12:19], s[20:23] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[20:23], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[20:23], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   call void @llvm.amdgcn.struct.buffer.store.f32(float %inp, <4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   %reload = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %reload, %reload
   %temp2 = call float @llvm.amdgcn.strict.wqm.f32(float %temp)
   %temp3 = fadd float %temp2, %temp2
   %reload_wwm = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res2, i32 %idx0, i32 0, i32 0, i32 0)
   %temp4 = call float @llvm.amdgcn.strict.wwm.f32(float %reload_wwm)
   %temp5 = fadd float %temp3, %temp4
   %tex = call float @llvm.amdgcn.image.sample.1d.f32.f32(i32 1, float %temp5, <8 x i32> %res3, <4 x i32> %res, i1 false, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %tex, <4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   %out = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   ret float %out
 }
 
-;CHECK-LABEL: {{^}}test_strict_wwm_strict_wqm_wqm:
-;CHECK: buffer_store_dword
-
-;CHECK: s_or_saveexec_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG]]
-
-;CHECK: s_mov_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-
-;CHECK: s_or_saveexec_b64 [[ORIG3:s\[[0-9]+:[0-9]+\]]], -1
-;CHECK: v_add
-;CHECK: s_mov_b64 exec, [[ORIG3]]
-
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: image_sample
 define amdgpu_ps float @test_strict_wwm_strict_wqm_wqm(i32 inreg %idx0, i32 inreg %idx1, <4 x i32> inreg %res, float %inp, <8 x i32> inreg %res2) {
+; GFX9-W64-LABEL: test_strict_wwm_strict_wqm_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[20:21], exec
+; GFX9-W64-NEXT:    s_mov_b32 s15, s13
+; GFX9-W64-NEXT:    s_mov_b64 s[22:23], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b32 s19, s5
+; GFX9-W64-NEXT:    s_mov_b32 s18, s4
+; GFX9-W64-NEXT:    s_mov_b32 s17, s3
+; GFX9-W64-NEXT:    s_mov_b32 s16, s2
+; GFX9-W64-NEXT:    s_mov_b32 s14, s12
+; GFX9-W64-NEXT:    s_mov_b32 s13, s11
+; GFX9-W64-NEXT:    s_mov_b32 s12, s10
+; GFX9-W64-NEXT:    s_mov_b32 s11, s9
+; GFX9-W64-NEXT:    s_mov_b32 s10, s8
+; GFX9-W64-NEXT:    s_mov_b32 s9, s7
+; GFX9-W64-NEXT:    s_mov_b32 s8, s6
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[22:23]
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-W64-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-W64-NEXT:    buffer_load_dword v2, v2, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[2:3]
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    buffer_load_dword v3, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_or_saveexec_b64 s[0:1], -1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-W64-NEXT:    v_add_f32_e32 v2, v2, v2
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v4, v3
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v4
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[20:21]
+; GFX9-W64-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_strict_wwm_strict_wqm_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s20, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s15, s13
+; GFX10-W32-NEXT:    s_mov_b32 s21, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s19, s5
+; GFX10-W32-NEXT:    s_mov_b32 s18, s4
+; GFX10-W32-NEXT:    s_mov_b32 s17, s3
+; GFX10-W32-NEXT:    s_mov_b32 s16, s2
+; GFX10-W32-NEXT:    s_mov_b32 s14, s12
+; GFX10-W32-NEXT:    s_mov_b32 s13, s11
+; GFX10-W32-NEXT:    s_mov_b32 s12, s10
+; GFX10-W32-NEXT:    s_mov_b32 s11, s9
+; GFX10-W32-NEXT:    s_mov_b32 s10, s8
+; GFX10-W32-NEXT:    s_mov_b32 s9, s7
+; GFX10-W32-NEXT:    s_mov_b32 s8, s6
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s21
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    v_mov_b32_e32 v2, s1
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    buffer_load_dword v2, v2, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v3, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_or_saveexec_b32 s0, -1
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(1)
+; GFX10-W32-NEXT:    v_add_f32_e32 v2, v2, v2
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v0, v2
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v4, v3
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v4
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s20
+; GFX10-W32-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   call void @llvm.amdgcn.struct.buffer.store.f32(float %inp, <4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   %reload = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %reload, %reload
   %temp2 = call float @llvm.amdgcn.strict.wwm.f32(float %temp)
   %temp3 = fadd float %temp2, %temp2
   %reload_wwm = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   %temp4 = call float @llvm.amdgcn.strict.wqm.f32(float %reload_wwm)
   %temp5 = fadd float %temp3, %temp4
   %tex = call float @llvm.amdgcn.image.sample.1d.f32.f32(i32 1, float %temp5, <8 x i32> %res2, <4 x i32> %res, i1 false, i32 0, i32 0)
   call void @llvm.amdgcn.struct.buffer.store.f32(float %tex, <4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   %out = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   ret float %out
 }
 
-;CHECK-LABEL: {{^}}test_wqm_strict_wqm_wqm:
-;CHECK: buffer_store_dword
-
-;CHECK: s_wqm_b64 exec, exec
-
 ;TODO: WQM -> StrictWQM transition could be improved. StrictWQM could use the exec from the previous state instead of calling s_wqm again.
-;CHECK: s_mov_b64 [[ORIG2:s\[[0-9]+:[0-9]+\]]], exec
-;CHECK: s_wqm_b64 exec, exec
-;CHECK: buffer_load_dword
-;CHECK: s_mov_b64 exec, [[ORIG2]]
-
-;CHECK: image_sample
-
 define amdgpu_ps float @test_wqm_strict_wqm_wqm(i32 inreg %idx0, i32 inreg %idx1, <4 x i32> inreg %res, float %inp, <8 x i32> inreg %res2) {
+; GFX9-W64-LABEL: test_wqm_strict_wqm_wqm:
+; GFX9-W64:       ; %bb.0: ; %main_body
+; GFX9-W64-NEXT:    s_mov_b64 s[20:21], exec
+; GFX9-W64-NEXT:    s_mov_b32 s15, s13
+; GFX9-W64-NEXT:    s_mov_b64 s[22:23], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    s_mov_b32 s19, s5
+; GFX9-W64-NEXT:    s_mov_b32 s18, s4
+; GFX9-W64-NEXT:    s_mov_b32 s17, s3
+; GFX9-W64-NEXT:    s_mov_b32 s16, s2
+; GFX9-W64-NEXT:    s_mov_b32 s14, s12
+; GFX9-W64-NEXT:    s_mov_b32 s13, s11
+; GFX9-W64-NEXT:    s_mov_b32 s12, s10
+; GFX9-W64-NEXT:    s_mov_b32 s11, s9
+; GFX9-W64-NEXT:    s_mov_b32 s10, s8
+; GFX9-W64-NEXT:    s_mov_b32 s9, s7
+; GFX9-W64-NEXT:    s_mov_b32 s8, s6
+; GFX9-W64-NEXT:    v_mov_b32_e32 v1, s0
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[22:23]
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    v_mov_b32_e32 v0, s1
+; GFX9-W64-NEXT:    buffer_load_dword v0, v0, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 s[0:1], exec
+; GFX9-W64-NEXT:    s_wqm_b64 exec, exec
+; GFX9-W64-NEXT:    buffer_load_dword v2, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-W64-NEXT:    v_mov_b32_e32 v3, v2
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX9-W64-NEXT:    v_add_f32_e32 v0, v0, v3
+; GFX9-W64-NEXT:    s_and_b64 exec, exec, s[20:21]
+; GFX9-W64-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    buffer_load_dword v0, v1, s[16:19], 0 idxen
+; GFX9-W64-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-W64-NEXT:    ; return to shader part epilog
+;
+; GFX10-W32-LABEL: test_wqm_strict_wqm_wqm:
+; GFX10-W32:       ; %bb.0: ; %main_body
+; GFX10-W32-NEXT:    s_mov_b32 s20, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s15, s13
+; GFX10-W32-NEXT:    s_mov_b32 s21, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    s_mov_b32 s19, s5
+; GFX10-W32-NEXT:    s_mov_b32 s18, s4
+; GFX10-W32-NEXT:    s_mov_b32 s17, s3
+; GFX10-W32-NEXT:    s_mov_b32 s16, s2
+; GFX10-W32-NEXT:    s_mov_b32 s14, s12
+; GFX10-W32-NEXT:    s_mov_b32 s13, s11
+; GFX10-W32-NEXT:    s_mov_b32 s12, s10
+; GFX10-W32-NEXT:    s_mov_b32 s11, s9
+; GFX10-W32-NEXT:    s_mov_b32 s10, s8
+; GFX10-W32-NEXT:    s_mov_b32 s9, s7
+; GFX10-W32-NEXT:    s_mov_b32 s8, s6
+; GFX10-W32-NEXT:    v_mov_b32_e32 v1, s0
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s21
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, s1
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s20
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v0, v3, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 s0, exec_lo
+; GFX10-W32-NEXT:    s_wqm_b32 exec_lo, exec_lo
+; GFX10-W32-NEXT:    buffer_load_dword v2, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_mov_b32 exec_lo, s0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(1)
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_mov_b32_e32 v3, v2
+; GFX10-W32-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v0
+; GFX10-W32-NEXT:    v_add_f32_e32 v0, v0, v3
+; GFX10-W32-NEXT:    s_and_b32 exec_lo, exec_lo, s20
+; GFX10-W32-NEXT:    image_sample v0, v0, s[8:15], s[16:19] dmask:0x1 dim:SQ_RSRC_IMG_1D
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    buffer_store_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    buffer_load_dword v0, v1, s[16:19], 0 idxen
+; GFX10-W32-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-W32-NEXT:    s_waitcnt_vscnt null, 0x0
+; GFX10-W32-NEXT:    ; return to shader part epilog
 main_body:
   call void @llvm.amdgcn.struct.buffer.store.f32(float %inp, <4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   %reload = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx1, i32 0, i32 0, i32 0)
   %temp = fadd float %reload, %reload
   %tex = call float @llvm.amdgcn.image.sample.1d.f32.f32(i32 1, float %temp, <8 x i32> %res2, <4 x i32> %res, i1 false, i32 0, i32 0)
   %temp2 = fadd float %tex, %tex
   %reload_wwm = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %res, i32 %idx0, i32 0, i32 0, i32 0)
   %temp3 = call float @llvm.amdgcn.strict.wqm.f32(float %reload_wwm)