ef8655adc8
CUDA-12 no longer supports 32-bit compilation. Tests agnostic to 32/64 compilation mode are switched to use nvptx64. Tests that do care about it have 32-bit ptxas compilation disabled with cuda-12+. Differential Revision: https://reviews.llvm.org/D152199
32 lines
734 B
LLVM
32 lines
734 B
LLVM
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 | FileCheck %s
|
|
; RUN: %if ptxas %{ llc < %s -march=nvptx64 -mcpu=sm_20 | %ptxas-verify %}
|
|
|
|
; Ensure source scheduling is working
|
|
|
|
define void @foo(ptr %a) {
|
|
; CHECK: .func foo
|
|
; CHECK: ld.u32
|
|
; CHECK-NEXT: ld.u32
|
|
; CHECK-NEXT: ld.u32
|
|
; CHECK-NEXT: ld.u32
|
|
; CHECK-NEXT: add.s32
|
|
; CHECK-NEXT: add.s32
|
|
; CHECK-NEXT: add.s32
|
|
%val0 = load i32, ptr %a
|
|
%ptr1 = getelementptr i32, ptr %a, i32 1
|
|
%val1 = load i32, ptr %ptr1
|
|
%ptr2 = getelementptr i32, ptr %a, i32 2
|
|
%val2 = load i32, ptr %ptr2
|
|
%ptr3 = getelementptr i32, ptr %a, i32 3
|
|
%val3 = load i32, ptr %ptr3
|
|
|
|
%t0 = add i32 %val0, %val1
|
|
%t1 = add i32 %t0, %val2
|
|
%t2 = add i32 %t1, %val3
|
|
|
|
store i32 %t2, ptr %a
|
|
|
|
ret void
|
|
}
|
|
|