0f1b5f115a
ptxas is a proprietary compiler from Nvidia that can compile PTX to machine code (SASS). It has a lot of diagnostics to catch errors in PTX, which can be used to verify PTX output from llc. Set -DPXTAS_EXECUTABLE=/path/to/ptxas CMake option to enable it. If this option is not set, then ptxas is substituted to true which effectively disables all ptxas RUN lines. LLVM_PTXAS_EXECUTABLE environment variable takes precedence over the CMake option, and allows to override ptxas executable that is used for LIT without complete re-configuration. Differential Revision: https://reviews.llvm.org/D121727
17 lines
797 B
LLVM
17 lines
797 B
LLVM
; RUN: llc < %s -mtriple=nvptx -mcpu=sm_20 -relocation-model=static | FileCheck %s --check-prefix CHK
|
|
; RUN: %if ptxas %{ llc < %s -mtriple=nvptx -mcpu=sm_20 -relocation-model=static | %ptxas-verify %}
|
|
|
|
%MyStruct = type { i32, i32, float }
|
|
@Gbl = internal addrspace(3) global [1024 x %MyStruct] zeroinitializer
|
|
|
|
; CHK-LABEL: foo
|
|
define void @foo(float %f) {
|
|
entry:
|
|
; CHK: ld.shared.f32 %{{[a-zA-Z0-9]+}}, [Gbl+8];
|
|
%0 = load float, float addrspace(3)* getelementptr inbounds ([1024 x %MyStruct], [1024 x %MyStruct] addrspace(3)* @Gbl, i32 0, i32 0, i32 2)
|
|
%add = fadd float %0, %f
|
|
; CHK: st.shared.f32 [Gbl+8], %{{[a-zA-Z0-9]+}};
|
|
store float %add, float addrspace(3)* getelementptr inbounds ([1024 x %MyStruct], [1024 x %MyStruct] addrspace(3)* @Gbl, i32 0, i32 0, i32 2)
|
|
ret void
|
|
}
|