llvm/test/CodeGen/NVPTX/ldu-reg-plus-offset.ll
Jingyue Wu 87a2b36cf6 [NVPTX] Make the alignment an explicit argument to ldu/ldg
Summary:
Instead of specifying the alignment as metadata which may be destroyed by
transformation passes, make the alignment the second argument to ldu/ldg
intrinsic calls.

Test Plan:
ldu-ldg.ll
ldu-i8.ll
ldu-reg-plus-offset.ll

Reviewers: eliben, meheff, jholewinski

Reviewed By: meheff, jholewinski

Subscribers: jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D5093

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216731 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-29 15:30:20 +00:00

20 lines
757 B
LLVM

; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16-v32:32:32-v64:64:64-v128:128:128-n16:32:64"
define void @reg_plus_offset(i32* %a) {
; CHECK: ldu.global.u32 %r{{[0-9]+}}, [%r{{[0-9]+}}+32];
; CHECK: ldu.global.u32 %r{{[0-9]+}}, [%r{{[0-9]+}}+36];
%p2 = getelementptr i32* %a, i32 8
%t1 = call i32 @llvm.nvvm.ldu.global.i.i32.p0i32(i32* %p2, i32 4)
%p3 = getelementptr i32* %a, i32 9
%t2 = call i32 @llvm.nvvm.ldu.global.i.i32.p0i32(i32* %p3, i32 4)
%t3 = mul i32 %t1, %t2
store i32 %t3, i32* %a
ret void
}
declare i32 @llvm.nvvm.ldu.global.i.i32.p0i32(i32*, i32)
declare i32 @llvm.nvvm.read.ptx.sreg.tid.x()