diff --git a/llvm/test/CodeGen/AMDGPU/llvm.ceil.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.ceil.f16.ll
--- a/llvm/test/CodeGen/AMDGPU/llvm.ceil.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.ceil.f16.ll
@@ -1,18 +1,68 @@
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VI %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s
 
 declare half @llvm.ceil.f16(half %a)
 declare <2 x half> @llvm.ceil.v2f16(<2 x half> %a)
 
-; GCN-LABEL: {{^}}ceil_f16:
-; GCN: buffer_load_ushort v[[A_F16:[0-9]+]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32:[0-9]+]], v[[A_F16]]
-; SI:  v_ceil_f32_e32 v[[R_F32:[0-9]+]], v[[A_F32]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16:[0-9]+]], v[[R_F32]]
-; VI:  v_ceil_f16_e32 v[[R_F16:[0-9]+]], v[[A_F16]]
-; GCN: buffer_store_short v[[R_F16]]
-; GCN: s_endpgm
 define amdgpu_kernel void @ceil_f16(
+; SI-LABEL: ceil_f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_ceil_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: ceil_f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_ceil_f16_e32 v0, v0
+; VI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: ceil_f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_u16 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_ceil_f16_e32 v0, v0
+; GFX11-NEXT:    buffer_store_b16 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
@@ -22,27 +72,82 @@
   ret void
 }
 
-; GCN-LABEL: {{^}}ceil_v2f16:
-; GCN: buffer_load_dword v[[A_V2_F16:[0-9]+]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]]
-; SI-DAG:  v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
-; SI-DAG:  v_ceil_f32_e32 v[[R_F32_0:[0-9]+]], v[[A_F32_0]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_0:[0-9]+]], v[[R_F32_0]]
-; SI-DAG:  v_ceil_f32_e32 v[[R_F32_1:[0-9]+]], v[[A_F32_1]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
-; SI:  v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
-; SI-NOT: and
-; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
-
-; VI-DAG:  v_ceil_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
-; VI-DAG:  v_ceil_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
-; VI-NOT: and
-; VI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
-
-; GCN: buffer_store_dword v[[R_V2_F16]]
-; GCN: s_endpgm
+; The original test with manual checks also had these NOT directives:
+; COM: SI: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
+; COM: SI-NOT: and
+; COM: SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
+; COM: VI-DAG: v_ceil_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; COM: VI-NOT: and
+; COM: VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
 define amdgpu_kernel void @ceil_v2f16(
+; SI-LABEL: ceil_v2f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; SI-NEXT:    v_cvt_f32_f16_e32 v1, v1
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_ceil_f32_e32 v1, v1
+; SI-NEXT:    v_cvt_f16_f32_e32 v1, v1
+; SI-NEXT:    v_ceil_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
+; SI-NEXT:    v_or_b32_e32 v0, v0, v1
+; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: ceil_v2f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_ceil_f16_sdwa v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; VI-NEXT:    v_ceil_f16_e32 v0, v0
+; VI-NEXT:    v_or_b32_e32 v0, v0, v1
+; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: ceil_v2f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_b32 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX11-NEXT:    v_ceil_f16_e32 v0, v0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX11-NEXT:    v_ceil_f16_e32 v1, v1
+; GFX11-NEXT:    v_pack_b32_f16 v0, v0, v1
+; GFX11-NEXT:    buffer_store_b32 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.exp2.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.exp2.f16.ll
--- a/llvm/test/CodeGen/AMDGPU/llvm.exp2.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.exp2.f16.ll
@@ -1,18 +1,68 @@
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VI %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s
 
 declare half @llvm.exp2.f16(half %a)
 declare <2 x half> @llvm.exp2.v2f16(<2 x half> %a)
 
-; GCN-LABEL: {{^}}exp2_f16
-; GCN: buffer_load_ushort v[[A_F16:[0-9]+]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32:[0-9]+]], v[[A_F16]]
-; SI:  v_exp_f32_e32 v[[R_F32:[0-9]+]], v[[A_F32]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16:[0-9]+]], v[[R_F32]]
-; VI:  v_exp_f16_e32 v[[R_F16:[0-9]+]], v[[A_F16]]
-; GCN: buffer_store_short v[[R_F16]]
-; GCN: s_endpgm
 define amdgpu_kernel void @exp2_f16(
+; SI-LABEL: exp2_f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_exp_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: exp2_f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_exp_f16_e32 v0, v0
+; VI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: exp2_f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_u16 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_exp_f16_e32 v0, v0
+; GFX11-NEXT:    buffer_store_b16 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
@@ -22,27 +72,83 @@
   ret void
 }
 
-; GCN-LABEL: {{^}}exp2_v2f16
-; GCN: buffer_load_dword v[[A_V2_F16:[0-9]+]]
-; SI:  v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]]
-; SI:  v_exp_f32_e32 v[[R_F32_1:[0-9]+]], v[[A_F32_1]]
-; SI:  v_exp_f32_e32 v[[R_F32_0:[0-9]+]], v[[A_F32_0]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16_0:[0-9]+]], v[[R_F32_0]]
-; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
-; SI-NOT: and
-; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
-
-; VI-DAG: v_exp_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
-; VI-DAG: v_exp_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
-; VI-NOT: and
-; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
-
-; GCN: buffer_store_dword v[[R_V2_F16]]
-; GCN: s_endpgm
+; The original test with manual checks also had these NOT directives:
+; COM: SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
+; COM: SI-NOT: and
+; COM: SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
+; COM: VI-DAG: v_exp_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; COM: VI-NOT: and
+; COM: VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
 define amdgpu_kernel void @exp2_v2f16(
+; SI-LABEL: exp2_v2f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; SI-NEXT:    v_cvt_f32_f16_e32 v1, v1
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_exp_f32_e32 v1, v1
+; SI-NEXT:    v_exp_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v1, v1
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
+; SI-NEXT:    v_or_b32_e32 v0, v0, v1
+; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: exp2_v2f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_exp_f16_sdwa v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; VI-NEXT:    v_exp_f16_e32 v0, v0
+; VI-NEXT:    v_or_b32_e32 v0, v0, v1
+; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: exp2_v2f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_b32 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX11-NEXT:    v_exp_f16_e32 v0, v0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-NEXT:    v_exp_f16_e32 v1, v1
+; GFX11-NEXT:    s_waitcnt_depctr 0xfff
+; GFX11-NEXT:    v_pack_b32_f16 v0, v0, v1
+; GFX11-NEXT:    buffer_store_b32 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.floor.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.floor.f16.ll
--- a/llvm/test/CodeGen/AMDGPU/llvm.floor.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.floor.f16.ll
@@ -1,18 +1,68 @@
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VI %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s
 
 declare half @llvm.floor.f16(half %a)
 declare <2 x half> @llvm.floor.v2f16(<2 x half> %a)
 
-; GCN-LABEL: {{^}}floor_f16
-; GCN: buffer_load_ushort v[[A_F16:[0-9]+]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32:[0-9]+]], v[[A_F16]]
-; SI:  v_floor_f32_e32 v[[R_F32:[0-9]+]], v[[A_F32]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16:[0-9]+]], v[[R_F32]]
-; VI:  v_floor_f16_e32 v[[R_F16:[0-9]+]], v[[A_F16]]
-; GCN: buffer_store_short v[[R_F16]]
-; GCN: s_endpgm
 define amdgpu_kernel void @floor_f16(
+; SI-LABEL: floor_f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_floor_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: floor_f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_floor_f16_e32 v0, v0
+; VI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: floor_f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_u16 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_floor_f16_e32 v0, v0
+; GFX11-NEXT:    buffer_store_b16 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
@@ -22,27 +72,83 @@
   ret void
 }
 
-; GCN-LABEL: {{^}}floor_v2f16
-; GCN: buffer_load_dword v[[A_V2_F16:[0-9]+]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]]
-; SI-DAG:  v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
-; SI-DAG:  v_floor_f32_e32 v[[R_F32_0:[0-9]+]], v[[A_F32_0]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_0:[0-9]+]], v[[R_F32_0]]
-; SI-DAG:  v_floor_f32_e32 v[[R_F32_1:[0-9]+]], v[[A_F32_1]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
-; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
-; SI-NOT: and
-; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
-
-; VI-DAG: v_floor_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
-; VI-DAG: v_floor_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
-; VI-NOT: and
-; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
-
-; GCN: buffer_store_dword v[[R_V2_F16]]
-; GCN: s_endpgm
+; The original test with manual checks also had these NOT directives:
+; COM: SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
+; COM: SI-NOT: and
+; COM: SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
+; COM: VI-DAG: v_floor_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
+; COM: VI-DAG: v_floor_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; COM: VI-NOT: and
+; COM: VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
 define amdgpu_kernel void @floor_v2f16(
+; SI-LABEL: floor_v2f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; SI-NEXT:    v_cvt_f32_f16_e32 v1, v1
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_floor_f32_e32 v1, v1
+; SI-NEXT:    v_cvt_f16_f32_e32 v1, v1
+; SI-NEXT:    v_floor_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
+; SI-NEXT:    v_or_b32_e32 v0, v0, v1
+; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: floor_v2f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_floor_f16_sdwa v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; VI-NEXT:    v_floor_f16_e32 v0, v0
+; VI-NEXT:    v_or_b32_e32 v0, v0, v1
+; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX11-LABEL: floor_v2f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_b32 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX11-NEXT:    v_floor_f16_e32 v0, v0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX11-NEXT:    v_floor_f16_e32 v1, v1
+; GFX11-NEXT:    v_pack_b32_f16 v0, v0, v1
+; GFX11-NEXT:    buffer_store_b32 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.rint.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.rint.f16.ll
--- a/llvm/test/CodeGen/AMDGPU/llvm.rint.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.rint.f16.ll
@@ -1,19 +1,69 @@
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,VI,GFX89 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX89,GFX9 %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck --check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GFX89,VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GFX89,GFX9 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GFX11 %s
 
 declare half @llvm.rint.f16(half %a)
 declare <2 x half> @llvm.rint.v2f16(<2 x half> %a)
 
-; GCN-LABEL: {{^}}rint_f16
-; GCN: buffer_load_ushort v[[A_F16:[0-9]+]]
-; SI:  v_cvt_f32_f16_e32 v[[A_F32:[0-9]+]], v[[A_F16]]
-; SI:  v_rndne_f32_e32 v[[R_F32:[0-9]+]], v[[A_F32]]
-; SI:  v_cvt_f16_f32_e32 v[[R_F16:[0-9]+]], v[[R_F32]]
-; GFX89: v_rndne_f16_e32 v[[R_F16:[0-9]+]], v[[A_F16]]
-; GCN: buffer_store_short v[[R_F16]]
-; GCN: s_endpgm
 define amdgpu_kernel void @rint_f16(
+; SI-LABEL: rint_f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_rndne_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; GFX89-LABEL: rint_f16:
+; GFX89:       ; %bb.0: ; %entry
+; GFX89-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX89-NEXT:    s_mov_b32 s7, 0xf000
+; GFX89-NEXT:    s_mov_b32 s6, -1
+; GFX89-NEXT:    s_mov_b32 s10, s6
+; GFX89-NEXT:    s_mov_b32 s11, s7
+; GFX89-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX89-NEXT:    s_mov_b32 s8, s2
+; GFX89-NEXT:    s_mov_b32 s9, s3
+; GFX89-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; GFX89-NEXT:    s_mov_b32 s4, s0
+; GFX89-NEXT:    s_mov_b32 s5, s1
+; GFX89-NEXT:    s_waitcnt vmcnt(0)
+; GFX89-NEXT:    v_rndne_f16_e32 v0, v0
+; GFX89-NEXT:    buffer_store_short v0, off, s[4:7], 0
+; GFX89-NEXT:    s_endpgm
+;
+; GFX11-LABEL: rint_f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_u16 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_rndne_f16_e32 v0, v0
+; GFX11-NEXT:    buffer_store_b16 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry:
@@ -23,32 +73,103 @@
   ret void
 }
 
-; GCN-LABEL: {{^}}rint_v2f16
-; GCN: buffer_load_dword v[[A_V2_F16:[0-9]+]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]]
-; SI-DAG:  v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
-; SI-DAG:  v_rndne_f32_e32 v[[R_F32_0:[0-9]+]], v[[A_F32_0]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_0:[0-9]+]], v[[R_F32_0]]
-; SI-DAG:  v_rndne_f32_e32 v[[R_F32_1:[0-9]+]], v[[A_F32_1]]
-; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
-; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
-; SI-NOT: v_and_b32
-; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
-
-; VI-DAG: v_rndne_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
-; VI-DAG: v_rndne_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
-; VI-NOT: v_and_b32
-; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
-
-; GFX9: v_rndne_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
-; GFX9: v_rndne_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
-; GFX9: v_pack_b32_f16 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
-
-; GCN: buffer_store_dword v[[R_V2_F16]]
-; GCN: s_endpgm
-
+; The original test with manual checks also had these NOT directives:
+; COM: SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
+; COM: SI-NOT: v_and_b32
+; COM: SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_HI]]
+; COM: VI-DAG: v_rndne_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]]
+; COM: VI-DAG: v_rndne_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; COM: VI-NOT: v_and_b32
+; COM: VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
 define amdgpu_kernel void @rint_v2f16(
+; SI-LABEL: rint_v2f16:
+; SI:       ; %bb.0: ; %entry
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_mov_b32 s10, s6
+; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s8, s2
+; SI-NEXT:    s_mov_b32 s9, s3
+; SI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; SI-NEXT:    s_mov_b32 s4, s0
+; SI-NEXT:    s_mov_b32 s5, s1
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; SI-NEXT:    v_cvt_f32_f16_e32 v1, v1
+; SI-NEXT:    v_cvt_f32_f16_e32 v0, v0
+; SI-NEXT:    v_rndne_f32_e32 v1, v1
+; SI-NEXT:    v_cvt_f16_f32_e32 v1, v1
+; SI-NEXT:    v_rndne_f32_e32 v0, v0
+; SI-NEXT:    v_cvt_f16_f32_e32 v0, v0
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
+; SI-NEXT:    v_or_b32_e32 v0, v0, v1
+; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: rint_v2f16:
+; VI:       ; %bb.0: ; %entry
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s10, s6
+; VI-NEXT:    s_mov_b32 s11, s7
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s8, s2
+; VI-NEXT:    s_mov_b32 s9, s3
+; VI-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; VI-NEXT:    s_mov_b32 s4, s0
+; VI-NEXT:    s_mov_b32 s5, s1
+; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    v_rndne_f16_sdwa v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1
+; VI-NEXT:    v_rndne_f16_e32 v0, v0
+; VI-NEXT:    v_or_b32_e32 v0, v0, v1
+; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; GFX9-LABEL: rint_v2f16:
+; GFX9:       ; %bb.0: ; %entry
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b32 s7, 0xf000
+; GFX9-NEXT:    s_mov_b32 s6, -1
+; GFX9-NEXT:    s_mov_b32 s10, s6
+; GFX9-NEXT:    s_mov_b32 s11, s7
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    s_mov_b32 s8, s2
+; GFX9-NEXT:    s_mov_b32 s9, s3
+; GFX9-NEXT:    buffer_load_dword v0, off, s[8:11], 0
+; GFX9-NEXT:    s_mov_b32 s4, s0
+; GFX9-NEXT:    s_mov_b32 s5, s1
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_rndne_f16_e32 v1, v0
+; GFX9-NEXT:    v_rndne_f16_sdwa v0, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
+; GFX9-NEXT:    v_pack_b32_f16 v0, v1, v0
+; GFX9-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; GFX9-NEXT:    s_endpgm
+;
+; GFX11-LABEL: rint_v2f16:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_load_b128 s[0:3], s[0:1], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_b32 v0, off, s[8:11], 0
+; GFX11-NEXT:    s_mov_b32 s5, s1
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX11-NEXT:    v_rndne_f16_e32 v0, v0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX11-NEXT:    v_rndne_f16_e32 v1, v1
+; GFX11-NEXT:    v_pack_b32_f16 v0, v0, v1
+; GFX11-NEXT:    buffer_store_b32 v0, off, s[4:7], 0
+; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
+; GFX11-NEXT:    s_endpgm
     ptr addrspace(1) %r,
     ptr addrspace(1) %a) {
 entry: