[llvm] AMDGPU/GlobalISel: Fix regBankLegalize rules for uniform cvt_pkrtz (PR #203283)

[via llvm-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-amdgpu

Author: Petar Avramovic (petar-avramovic)

<details>
<summary>Changes</summary>



---
Full diff: https://github.com/llvm/llvm-project/pull/203283.diff


2 Files Affected:

- (modified) llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp (+6-1) 
- (modified) llvm/test/CodeGen/AMDGPU/scalar-float-sop2.ll (+2-2) 


``````````diff
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
index 838951bcd8d26..eb343cf712fdc 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
@@ -1949,11 +1949,16 @@ RegBankLegalizeRules::RegBankLegalizeRules(const GCNSubtarget &_ST,
       .Div(S64, {{Vgpr64}, {IntrId, Vgpr64, Vgpr32, Vgpr32}, V_BFE});
 
   addRulesForIOpcs({amdgcn_cvt_pk_i16, amdgcn_cvt_pk_u16, amdgcn_cvt_pknorm_i16,
-                    amdgcn_cvt_pknorm_u16, amdgcn_cvt_pkrtz},
+                    amdgcn_cvt_pknorm_u16},
                    Standard)
       .Uni(V2S16, {{UniInVgprV2S16}, {IntrId, Vgpr32, Vgpr32}})
       .Div(V2S16, {{VgprV2S16}, {IntrId, Vgpr32, Vgpr32}});
 
+  addRulesForIOpcs({amdgcn_cvt_pkrtz}, Standard)
+      .Uni(V2S16, {{SgprV2S16}, {IntrId, Sgpr32, Sgpr32}}, hasSALUFloat)
+      .Uni(V2S16, {{UniInVgprV2S16}, {IntrId, Vgpr32, Vgpr32}}, !hasSALUFloat)
+      .Div(V2S16, {{VgprV2S16}, {IntrId, Vgpr32, Vgpr32}});
+
   addRulesForIOpcs({amdgcn_cvt_scalef32_sr_pk32_bf6_f16,
                     amdgcn_cvt_scalef32_sr_pk32_fp6_f16,
                     amdgcn_cvt_scalef32_sr_pk32_bf6_bf16,
diff --git a/llvm/test/CodeGen/AMDGPU/scalar-float-sop2.ll b/llvm/test/CodeGen/AMDGPU/scalar-float-sop2.ll
index b2770f337fdb4..d134f67213d02 100644
--- a/llvm/test/CodeGen/AMDGPU/scalar-float-sop2.ll
+++ b/llvm/test/CodeGen/AMDGPU/scalar-float-sop2.ll
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=amdgcn -mcpu=gfx1150 < %s | FileCheck -check-prefixes=CHECK,GFX1150 %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1150 -global-isel < %s | FileCheck -check-prefixes=CHECK,GFX1150 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1150 -global-isel -new-reg-bank-select < %s | FileCheck -check-prefixes=CHECK,GFX1150 %s
 ; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck -check-prefixes=CHECK,GFX12 %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel < %s | FileCheck -check-prefixes=CHECK,GFX12 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel -new-reg-bank-select < %s | FileCheck -check-prefixes=CHECK,GFX12 %s
 
 define amdgpu_vs float @fadd_f32(float inreg %a, float inreg %b) {
 ; CHECK-LABEL: fadd_f32:

``````````

</details>


https://github.com/llvm/llvm-project/pull/203283