AMDGPU/SI: Increase SGPR limit to 96 on Tonga/Iceland

[mareko]

Summary

This is the setting of the Vulkan closed source driver.

It decreases the max wave count from 10 to 8.

26010 shaders in 14650 tests
Totals:
VGPRS: 829593 -> 808440 (-2.55 %)
Spilled SGPRs: 81878 -> 42226 (-48.43 %)
Spilled VGPRs: 367 -> 358 (-2.45 %)
Scratch VGPRs: 1764 -> 1748 (-0.91 %) dwords per thread
Code Size: 36677864 -> 35923932 (-2.06 %) bytes

There is a massive decrease in SGPR spilling in general and -7.4% spilled
VGPRs for DiRT Showdown (= SGPRs spilled to scratch?)

[nhaehnle]

Looks like a good tradeoff.

llvm/trunk/lib/Target/AMDGPU/AMDGPUSubtarget.h

public:
   short getTexVTXClauseSize() const {
     return TexVTXClauseSize;
   }
 };
 
 class SISubtarget final : public AMDGPUSubtarget {
 public:
   enum {
-    FIXED_SGPR_COUNT_FOR_INIT_BUG = 80
+    // The closed Vulkan driver sets 96, which limits the wave count to 8 but
+    // doesn't spill SGPRs as much as when 80 is set.
+    FIXED_SGPR_COUNT_FOR_INIT_BUG = 96
   };
 
 private:
   SIInstrInfo InstrInfo;
   SIFrameLowering FrameLowering;
   SITargetLowering TLInfo;
   std::unique_ptr<GISelAccessor> GISel;
 

llvm/trunk/test/CodeGen/AMDGPU/elf.ll

 
 ; ELF: Symbol {
 ; ELF: Name: test
 ; ELF: Binding: Global
 
 ; CONFIG: .section .AMDGPU.config
 ; CONFIG-NEXT: .long   45096
 ; TYPICAL-NEXT: .long   0
-; TONGA-NEXT: .long   576
+; TONGA-NEXT: .long   704
 ; CONFIG: .p2align 8
 ; CONFIG: test:
 define amdgpu_ps void @test(i32 %p) {
    %i = add i32 %p, 2
    %r = bitcast i32 %i to float
    call void @llvm.SI.export(i32 15, i32 0, i32 1, i32 12, i32 0, float %r, float %r, float %r, float %r)
    ret void
 }
 
 declare void @llvm.SI.export(i32, i32, i32, i32, i32, float, float, float, float)

llvm/trunk/test/CodeGen/AMDGPU/load-constant-i32.ll

define void @constant_zextload_v16i32_to_v16i64(<16 x i64> addrspace(1)* %out, <16 x i32> addrspace(2)* %in) #0 {
   %ext = zext <16 x i32> %ld to <16 x i64>
   store <16 x i64> %ext, <16 x i64> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}constant_sextload_v32i32_to_v32i64:
 
 ; GCN: s_load_dwordx16
-; GCN: s_load_dwordx16
+; GCN-DAG: s_load_dwordx16
 
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
 
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
 
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
 
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
-; GCN-NOHSA: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
+; GCN-NOHSA-DAG: buffer_store_dwordx4
 
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
 
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
 
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
 
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
-; GCN-HSA: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
+; GCN-HSA-DAG: flat_store_dwordx4
 
 define void @constant_sextload_v32i32_to_v32i64(<32 x i64> addrspace(1)* %out, <32 x i32> addrspace(2)* %in) #0 {
   %ld = load <32 x i32>, <32 x i32> addrspace(2)* %in
   %ext = sext <32 x i32> %ld to <32 x i64>
   store <32 x i64> %ext, <32 x i64> addrspace(1)* %out
   ret void
 }