From ba92059ca9ffce8b9eb0c496638dec62a2975ae3 Mon Sep 17 00:00:00 2001 From: Changpeng Fang Date: Fri, 16 Feb 2018 19:14:17 +0000 Subject: AMDGPU/SI: Extend promoting alloca to vector to arrays of up to 16 elements Summary: This patch extends the promotion of alloca to vector to the arrays of up to 16 elements. Also we introduce an option, -disable-promote-alloca-to-vector, to switch promotion to vector off, if needed. Reviewers: arsenm Differential Revision: https://reviews.llvm.org/D33559 llvm-svn: 325372 --- llvm/test/CodeGen/AMDGPU/private-memory-r600.ll | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'llvm/test/CodeGen/AMDGPU/private-memory-r600.ll') diff --git a/llvm/test/CodeGen/AMDGPU/private-memory-r600.ll b/llvm/test/CodeGen/AMDGPU/private-memory-r600.ll index 65e72817429..5e03b8a6e41 100644 --- a/llvm/test/CodeGen/AMDGPU/private-memory-r600.ll +++ b/llvm/test/CodeGen/AMDGPU/private-memory-r600.ll @@ -1,5 +1,6 @@ -; RUN: llc -march=r600 -mtriple=r600---amdgiz -mcpu=redwood < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC -; RUN: opt -S -mtriple=r600-unknown-unknown-amdgiz -mcpu=redwood -amdgpu-promote-alloca < %s | FileCheck -check-prefix=OPT %s +; RUN: llc -march=r600 -mcpu=redwood -disable-promote-alloca-to-vector < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC +; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck %s -check-prefix=R600-VECT -check-prefix=FUNC +; RUN: opt -S -mtriple=r600-unknown-unknown -mcpu=redwood -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=OPT %s target datalayout = "A5" declare i32 @llvm.r600.read.tidig.x() nounwind readnone @@ -110,7 +111,7 @@ for.end: ; FUNC-LABEL: {{^}}short_array: -; R600: MOVA_INT +; R600-VECT: MOVA_INT define amdgpu_kernel void @short_array(i32 addrspace(1)* %out, i32 %index) #0 { entry: %0 = alloca [2 x i16], addrspace(5) @@ -127,7 +128,7 @@ entry: ; FUNC-LABEL: {{^}}char_array: -; R600: MOVA_INT +; R600-VECT: MOVA_INT define amdgpu_kernel void @char_array(i32 addrspace(1)* %out, i32 %index) #0 { entry: %0 = alloca [2 x i8], addrspace(5) -- cgit v1.2.3