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
66 lines
1.7 KiB
LLVM
66 lines
1.7 KiB
LLVM
; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s --check-prefix=PTX32
|
|
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 | FileCheck %s --check-prefix=PTX64
|
|
; RUN: %if ptxas %{ llc < %s -march=nvptx -mcpu=sm_20 | %ptxas-verify %}
|
|
; RUN: %if ptxas %{ llc < %s -march=nvptx64 -mcpu=sm_20 | %ptxas-verify %}
|
|
|
|
|
|
;; i8
|
|
define i8 @ld_global_i8(i8 addrspace(0)* %ptr) {
|
|
; PTX32: ld.u8 %r{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.u8 %r{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load i8, i8 addrspace(0)* %ptr
|
|
ret i8 %a
|
|
}
|
|
|
|
;; i16
|
|
define i16 @ld_global_i16(i16 addrspace(0)* %ptr) {
|
|
; PTX32: ld.u16 %r{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.u16 %r{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load i16, i16 addrspace(0)* %ptr
|
|
ret i16 %a
|
|
}
|
|
|
|
;; i32
|
|
define i32 @ld_global_i32(i32 addrspace(0)* %ptr) {
|
|
; PTX32: ld.u32 %r{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.u32 %r{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load i32, i32 addrspace(0)* %ptr
|
|
ret i32 %a
|
|
}
|
|
|
|
;; i64
|
|
define i64 @ld_global_i64(i64 addrspace(0)* %ptr) {
|
|
; PTX32: ld.u64 %rd{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.u64 %rd{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load i64, i64 addrspace(0)* %ptr
|
|
ret i64 %a
|
|
}
|
|
|
|
;; f32
|
|
define float @ld_global_f32(float addrspace(0)* %ptr) {
|
|
; PTX32: ld.f32 %f{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.f32 %f{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load float, float addrspace(0)* %ptr
|
|
ret float %a
|
|
}
|
|
|
|
;; f64
|
|
define double @ld_global_f64(double addrspace(0)* %ptr) {
|
|
; PTX32: ld.f64 %fd{{[0-9]+}}, [%r{{[0-9]+}}]
|
|
; PTX32: ret
|
|
; PTX64: ld.f64 %fd{{[0-9]+}}, [%rd{{[0-9]+}}]
|
|
; PTX64: ret
|
|
%a = load double, double addrspace(0)* %ptr
|
|
ret double %a
|
|
}
|