[llvm] Add register bank legalization rules for amdgcn_cvt_sr_f16_f32. (PR #203253)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Jun 11 05:25:17 PDT 2026
https://github.com/anjenner created https://github.com/llvm/llvm-project/pull/203253
None
>From 7e6eee6a48c37847ef32074f8433b0a1c3e7e8bb Mon Sep 17 00:00:00 2001
From: Andrew Jenner <Andrew.Jenner at amd.com>
Date: Thu, 11 Jun 2026 08:32:00 -0400
Subject: [PATCH] Add register bank legalization rules for
amdgcn_cvt_sr_f16_f32.
---
llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp | 3 +++
llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.sr.ll | 2 +-
2 files changed, 4 insertions(+), 1 deletion(-)
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
index f622f703b8a4e..c472a3eb2eb5c 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
@@ -2263,6 +2263,9 @@ RegBankLegalizeRules::RegBankLegalizeRules(const GCNSubtarget &_ST,
addRulesForIOpcs({amdgcn_kill, amdgcn_wqm_demote})
.Any({{}, {{}, {IntrId, Vcc}}});
+ addRulesForIOpcs({amdgcn_cvt_sr_f16_f32}, Standard)
+ .Div(V2S16, {{VgprV2S16}, {IntrId, VgprV2S16, Vgpr32, Vgpr32}});
+
addRulesForIOpcs({amdgcn_ballot}, Standard)
.Uni(S64, {{Sgpr64}, {IntrId, Vcc}})
.Uni(S32, {{Sgpr32}, {IntrId, Vcc}});
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.sr.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.sr.ll
index 64a15bc102759..13cf8f9917ee6 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.sr.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.sr.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx950 -o - %s | FileCheck -check-prefix=GFX950 %s
-; RUN: llc -global-isel=1 -global-isel-abort=2 -mtriple=amdgcn -mcpu=gfx950 -o - %s | FileCheck -check-prefix=GFX950 %s
+; RUN: llc -global-isel=1 -global-isel-abort=2 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx950 -o - %s | FileCheck -check-prefix=GFX950 %s
declare <2 x half> @llvm.amdgcn.cvt.sr.f16.f32(<2 x half>, float, i32, i1)
declare <2 x bfloat> @llvm.amdgcn.cvt.sr.bf16.f32(<2 x bfloat>, float, i32, i1)
More information about the llvm-commits
mailing list