369891b674
In most cases, the type information attached to load and store instructions is meaningless and inconsistently applied. We can usually use ".b" loads and avoid the complexity of trying to assign the correct type. The one expectation is sign-extending load, which will continue to use ".s" to ensure the sign extension into a larger register is done correctly.
32 lines
738 B
LLVM
32 lines
738 B
LLVM
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 | FileCheck %s
|
|
; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 | %ptxas-verify %}
|
|
|
|
; Ensure source scheduling is working
|
|
|
|
define void @foo(ptr %a) {
|
|
; CHECK: .func foo
|
|
; CHECK: ld.b32
|
|
; CHECK-NEXT: ld.b32
|
|
; CHECK-NEXT: ld.b32
|
|
; CHECK-NEXT: ld.b32
|
|
; 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
|
|
}
|
|
|