Skip to content

Commit 83aa119

Browse files
committed
update tests to use CUDA sm_75 / ptx63 as default
1 parent 2e56f58 commit 83aa119

15 files changed

+169
-160
lines changed

clang/test/Driver/cuda-detect.cu

Lines changed: 21 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -60,7 +60,7 @@
6060
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_50 \
6161
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
6262
// RUN: | FileCheck %s -check-prefix COMMON \
63-
// RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE30
63+
// RUN: -check-prefixes PTX63,LIBDEVICE,LIBDEVICE10
6464
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_60 \
6565
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
6666
// RUN: | FileCheck %s -check-prefix COMMON \
@@ -96,14 +96,14 @@
9696

9797

9898
// Verify that -nocudainc prevents adding include path to CUDA headers.
99-
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
99+
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_75 \
100100
// RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
101101
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \
102-
// RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
103-
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \
102+
// RUN: -check-prefixes PTX63,LIBDEVICE,LIBDEVICE10
103+
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_75 \
104104
// RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
105105
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \
106-
// RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
106+
// RUN: -check-prefixes PTX63,LIBDEVICE,LIBDEVICE10
107107

108108
// We should not add any CUDA include paths if there's no valid CUDA installation
109109
// RUN: not %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
@@ -123,10 +123,10 @@
123123
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix MISSINGLIBDEVICE
124124

125125
// Verify that -nocudalib prevents linking libdevice bitcode in.
126-
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
126+
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_75 \
127127
// RUN: -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
128128
// RUN: | FileCheck %s -check-prefix COMMON
129-
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \
129+
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_75 \
130130
// RUN: -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
131131
// RUN: | FileCheck %s -check-prefix COMMON
132132

@@ -152,10 +152,10 @@
152152
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
153153
// RUN: | FileCheck %s -check-prefix CUDA80
154154

155-
// Verify that if no version file is found, we report the default of 7.0.
155+
// Verify that if no version file is found, we report the default of 10.0.
156156
// RUN: %clang -### -v --target=x86_64-linux-gnu --cuda-gpu-arch=sm_50 \
157157
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
158-
// RUN: | FileCheck %s -check-prefix CUDA70
158+
// RUN: | FileCheck %s -check-prefix CUDA100
159159

160160
// CHECK: Found CUDA installation: {{.*}}/Inputs/CUDA/usr/local/cuda
161161
// NO-LIBDEVICE: Found CUDA installation: {{.*}}/Inputs/CUDA-nolibdevice/usr/local/cuda
@@ -174,6 +174,7 @@
174174
// LIBDEVICE50-SAME: libdevice.compute_50.10.bc
175175
// PTX42-SAME: "-target-feature" "+ptx42"
176176
// PTX60-SAME: "-target-feature" "+ptx60"
177+
// PTX63-SAME: "-target-feature" "+ptx63"
177178
// CUDAINC-SAME: "-include" "__clang_cuda_runtime_wrapper.h"
178179
// NOCUDAINC-NOT: "-include" "__clang_cuda_runtime_wrapper.h"
179180
// CUDAINC-SAME: "-internal-isystem" "{{.*}}/Inputs/CUDA{{[_0-9]+}}/usr/local/cuda/include"
@@ -188,14 +189,20 @@
188189
// CHECK-CXXINCLUDE-SAME: {{.*}}"-internal-isystem" "{{.+}}/include/c++/4.8"
189190
// CHECK-CXXINCLUDE: ld{{.*}}"
190191

192+
// CUDA70: "-cc1" "-triple" "nvptx64-nvidia-cuda"
193+
// CUDA70-SAME: -target-sdk-version=7.0
194+
// CUDA70: "-cc1" "-triple" "x86_64-unknown-linux-gnu"
195+
// CUDA70-SAME: -target-sdk-version=7.0
196+
// CUDA70: ld{{.*}}"
197+
191198
// CUDA80: "-cc1" "-triple" "nvptx64-nvidia-cuda"
192199
// CUDA80-SAME: -target-sdk-version=8.0
193200
// CUDA80: "-cc1" "-triple" "x86_64-unknown-linux-gnu"
194201
// CUDA80-SAME: -target-sdk-version=8.0
195202
// CUDA80: ld{{.*}}"
196203

197-
// CUDA70: "-cc1" "-triple" "nvptx64-nvidia-cuda"
198-
// CUDA70-SAME: -target-sdk-version=7.0
199-
// CUDA70: "-cc1" "-triple" "x86_64-unknown-linux-gnu"
200-
// CUDA70-SAME: -target-sdk-version=7.0
201-
// CUDA70: ld{{.*}}"
204+
// CUDA100: "-cc1" "-triple" "nvptx64-nvidia-cuda"
205+
// CUDA100-SAME: -target-sdk-version=10.0
206+
// CUDA100: "-cc1" "-triple" "x86_64-unknown-linux-gnu"
207+
// CUDA100-SAME: -target-sdk-version=10.0
208+
// CUDA100: ld{{.*}}"

clang/test/Driver/cuda-flush-denormals-to-zero.cu

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -2,14 +2,14 @@
22
// -fgpu-flush-denormals-to-zero. This should be translated to
33
// -fdenormal-fp-math-f32=preserve-sign
44

5-
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fgpu-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=FTZ %s
6-
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fno-gpu-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=NOFTZ %s
5+
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fgpu-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=FTZ %s
6+
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fno-gpu-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=NOFTZ %s
77
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_70 -fgpu-flush-denormals-to-zero -nocudainc -nocudalib %s 2>&1 | FileCheck -check-prefix=FTZ %s
88
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_70 -fno-gpu-flush-denormals-to-zero -nocudainc -nocudalib %s 2>&1 | FileCheck -check-prefix=NOFTZ %s
99

1010
// Test alias options -f[no-]cuda-flush-denormals-to-zero
11-
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fcuda-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=FTZ %s
12-
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fno-cuda-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=NOFTZ %s
11+
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fcuda-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=FTZ %s
12+
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fno-cuda-flush-denormals-to-zero -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda %s 2>&1 | FileCheck -check-prefix=NOFTZ %s
1313

1414
// Test explicit argument, with CUDA offload kind
1515
// RUN: %clang -x hip -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=gfx803 -fgpu-flush-denormals-to-zero -nocudainc -nogpulib %s 2>&1 | FileCheck -check-prefix=FTZ %s

clang/test/Driver/cuda-march.cu

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -5,12 +5,12 @@
55
// RUN: %clang -### --target=x86_64-linux-gnu -c \
66
// RUN: -nogpulib -nogpuinc -march=haswell %s 2>&1 | FileCheck %s
77
// RUN: %clang -### --target=x86_64-linux-gnu -c \
8-
// RUN: -nogpulib -nogpuinc -march=haswell --cuda-gpu-arch=sm_52 %s 2>&1 | FileCheck %s
8+
// RUN: -nogpulib -nogpuinc -march=haswell --cuda-gpu-arch=sm_75 %s 2>&1 | FileCheck %s
99

1010
// CHECK: "-cc1"{{.*}} "-triple" "nvptx
11-
// CHECK-SAME: "-target-cpu" "sm_52"
11+
// CHECK-SAME: "-target-cpu" "sm_75"
1212

1313
// CHECK: ptxas
14-
// CHECK-SAME: "--gpu-name" "sm_52"
14+
// CHECK-SAME: "--gpu-name" "sm_75"
1515

1616
// CHECK: "-cc1"{{.*}} "-target-cpu" "haswell"

clang/test/Driver/cuda-options.cu

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -104,12 +104,12 @@
104104
// RUN: | FileCheck -check-prefixes ARCH-SM52,NOARCH-SM60,NOARCH-SM70 %s
105105

106106
// c) if --no-cuda-gpu-arch=X negates all preceding --cuda-gpu-arch=X
107-
// we default to sm_52 -- same as if no --cuda-gpu-arch were passed.
107+
// we default to sm_75 -- same as if no --cuda-gpu-arch were passed.
108108
// RUN: %clang -### --target=x86_64-linux-gnu --cuda-device-only \
109109
// RUN: -nogpulib -nogpuinc --cuda-gpu-arch=sm_70 --cuda-gpu-arch=sm_60 \
110110
// RUN: --no-cuda-gpu-arch=sm_70 --no-cuda-gpu-arch=sm_60 \
111111
// RUN: -c %s 2>&1 \
112-
// RUN: | FileCheck -check-prefixes ARCH-SM52,NOARCH-SM60,NOARCH-SM70 %s
112+
// RUN: | FileCheck -check-prefixes ARCH-SM75,NOARCH-SM60,NOARCH-SM70 %s
113113

114114
// d) --no-cuda-gpu-arch=X is a no-op if there's no preceding --cuda-gpu-arch=X
115115
// RUN: %clang -### --target=x86_64-linux-gnu --cuda-device-only \
@@ -193,6 +193,8 @@
193193
// NOARCH-SM60-NOT: "-cc1"{{.*}}"-target-cpu" "sm_60"
194194
// ARCH-SM70: "-cc1"{{.*}}"-target-cpu" "sm_70"
195195
// NOARCH-SM70-NOT: "-cc1"{{.*}}"-target-cpu" "sm_70"
196+
// ARCH-SM75: "-cc1"{{.*}}"-target-cpu" "sm_75"
197+
// NOARCH-SM75-NOT: "-cc1"{{.*}}"-target-cpu" "sm_75"
196198
// ARCHALLERROR: error: unsupported CUDA gpu architecture: all
197199

198200
// Match device-side preprocessor and compiler phases with -save-temps.
Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,8 @@
11
// RUN: %clang -### --target=i386-unknown-linux \
2-
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda \
2+
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda \
33
// RUN: --ptxas-path=/some/path/to/ptxas %s 2>&1 \
44
// RUN: | FileCheck %s
55

66
// CHECK-NOT: "ptxas"
77
// CHECK: "/some/path/to/ptxas"
8-
// CHECK-SAME: "--gpu-name" "sm_52"
8+
// CHECK-SAME: "--gpu-name" "sm_75"
Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
// Checks that cuda compilation does the right thing when passed -fcuda-short-ptr
22

3-
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fcuda-short-ptr -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 | FileCheck %s
3+
// RUN: %clang -### --target=x86_64-linux-gnu -c -march=haswell --cuda-gpu-arch=sm_20 -fcuda-short-ptr -nocudainc -nocudalib --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda %s 2>&1 | FileCheck %s
44

55
// CHECK: "-mllvm" "--nvptx-short-ptr"
66
// CHECK-SAME: "-fcuda-short-ptr"

clang/test/Driver/cuda-version-check.cu

Lines changed: 10 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_20 --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
1+
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_20 --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
22
// RUN: FileCheck %s --check-prefix=OK
33
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_20 --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda 2>&1 %s | \
44
// RUN: FileCheck %s --check-prefix=OK
@@ -15,43 +15,43 @@
1515
// RUN: --cuda-path=%S/Inputs/CUDA-unknown/usr/local/cuda 2>&1 %s | \
1616
// RUN: FileCheck %s --check-prefix=UNKNOWN_VERSION_CXX
1717

18-
// The installation at Inputs/CUDA is CUDA 7.0, which doesn't support sm_60.
19-
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
18+
// The installation at Inputs/CUDA_70 is CUDA 7.0, which doesn't support sm_60.
19+
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
2020
// RUN: FileCheck %s --check-prefix=ERR_SM60
2121

2222
// This should only complain about sm_60, not sm_35.
2323
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-gpu-arch=sm_35 \
24-
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
24+
// RUN: --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
2525
// RUN: FileCheck %s --check-prefix=ERR_SM60 --check-prefix=OK_SM35
2626

2727
// We should get two errors here, one for sm_60 and one for sm_61.
2828
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-gpu-arch=sm_61 \
29-
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
29+
// RUN: --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
3030
// RUN: FileCheck %s --check-prefix=ERR_SM60 --check-prefix=ERR_SM61
3131

3232
// We should still get an error if we pass -nocudainc, because this compilation
3333
// would invoke ptxas, and we do a version check on that, too.
34-
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
34+
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 -nocudainc --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
3535
// RUN: FileCheck %s --check-prefix=ERR_SM60
3636

3737
// If with -nocudainc and -E, we don't touch the CUDA install, so we
3838
// shouldn't get an error.
3939
// RUN: %clang --target=x86_64-linux -v -### -E --cuda-device-only --cuda-gpu-arch=sm_60 -nocudainc \
40-
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 %s | \
40+
// RUN: --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 %s | \
4141
// RUN: FileCheck %s --check-prefix=OK
4242

4343
// --no-cuda-version-check should suppress all of these errors.
44-
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-path=%S/Inputs/CUDA/usr/local/cuda 2>&1 \
44+
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda 2>&1 \
4545
// RUN: --no-cuda-version-check %s | \
4646
// RUN: FileCheck %s --check-prefix=OK
4747

4848
// We need to make sure the version check is done only for the device toolchain,
4949
// therefore we should not get an error in host-only mode. We use the -S here
5050
// to avoid the error being produced in case by the assembler tool, which does
5151
// the same check.
52-
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-host-only --cuda-path=%S/Inputs/CUDA/usr/local/cuda -S 2>&1 %s | \
52+
// RUN: %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-host-only --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda -S 2>&1 %s | \
5353
// RUN: FileCheck %s --check-prefix=OK
54-
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-device-only --cuda-path=%S/Inputs/CUDA/usr/local/cuda -S 2>&1 %s | \
54+
// RUN: not %clang --target=x86_64-linux -v -### --cuda-gpu-arch=sm_60 --cuda-device-only --cuda-path=%S/Inputs/CUDA_70/usr/local/cuda -S 2>&1 %s | \
5555
// RUN: FileCheck %s --check-prefix=ERR_SM60
5656

5757
// OK-NOT: error: GPU arch

clang/test/Driver/cuda-windows.cu

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -3,7 +3,7 @@
33
// RUN: %clang -v --target=i386-pc-windows-mingw32 \
44
// RUN: --sysroot=%S/Inputs/CUDA-windows 2>&1 %s -### | FileCheck %s
55

6-
// CHECK: Found CUDA installation: {{.*}}/Inputs/CUDA-windows/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v8.0
6+
// CHECK: Found CUDA installation: {{.*}}/Inputs/CUDA-windows/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0
77
// CHECK: "-cc1" "-triple" "nvptx-nvidia-cuda"
88
// CHECK-SAME: "-fms-extensions"
99
// CHECK-SAME: "-fms-compatibility"

clang/test/Driver/lto.cu

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -26,7 +26,7 @@
2626

2727
// llvm-bc and llvm-ll outputs need to match regular suffixes
2828
// (unfortunately).
29-
// RUN: %clangxx %s --target=x86_64-unknown-linux-gnu --no-offload-new-driver -nocudainc -nocudalib -flto -save-temps --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda -### 2> %t
29+
// RUN: %clangxx %s --target=x86_64-unknown-linux-gnu --no-offload-new-driver -nocudainc -nocudalib -flto -save-temps --cuda-path=%S/Inputs/CUDA/usr/local/cuda -### 2> %t
3030
// RUN: FileCheck -check-prefix=CHECK-COMPILELINK-SUFFIXES < %t %s
3131
//
3232
// CHECK-COMPILELINK-SUFFIXES: "-o" "[[CPP:.*lto-host.*\.cui]]" "-x" "cuda" "{{.*}}lto.cu"

clang/test/Driver/sycl-offload-new-driver.cpp

Lines changed: 17 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -5,23 +5,23 @@
55
// OFFLOAD-NEW-DRIVER: 0: input, "[[INPUT:.+\.cpp]]", c++, (host-sycl)
66
// OFFLOAD-NEW_DRIVER: 1: preprocessor, {0}, c++-cpp-output, (host-sycl)
77
// OFFLOAD-NEW_DRIVER: 2: compiler, {1}, ir, (host-sycl)
8-
// OFFLOAD-NEW_DRIVER: 3: input, "[[INPUT]]", c++, (device-sycl)
9-
// OFFLOAD-NEW_DRIVER: 4: preprocessor, {3}, c++-cpp-output, (device-sycl)
10-
// OFFLOAD-NEW_DRIVER: 5: compiler, {4}, ir, (device-sycl)
11-
// OFFLOAD-NEW_DRIVER: 6: backend, {5}, assembler, (device-sycl)
12-
// OFFLOAD-NEW_DRIVER: 7: assembler, {6}, object, (device-sycl)
13-
// OFFLOAD-NEW_DRIVER: 8: offload, "device-sycl (nvptx64-nvidia-cuda)" {7}, object
8+
// OFFLOAD-NEW_DRIVER: 3: input, "[[INPUT]]", c++, (device-sycl, sm_75)
9+
// OFFLOAD-NEW_DRIVER: 4: preprocessor, {3}, c++-cpp-output, (device-sycl, sm_75)
10+
// OFFLOAD-NEW_DRIVER: 5: compiler, {4}, ir, (device-sycl, sm_75)
11+
// OFFLOAD-NEW_DRIVER: 6: backend, {5}, ir, (device-sycl, sm_75)
12+
// OFFLOAD-NEW_DRIVER: 7: offload, "device-sycl (nvptx64-nvidia-cuda:sm_75)" {6}, ir
13+
// OFFLOAD-NEW_DRIVER: 8: input, "[[INPUT]]", c++, (device-sycl)
14+
// OFFLOAD-NEW_DRIVER: 9: input, "[[INPUT]]", c++, (device-sycl)
1415
// OFFLOAD-NEW_DRIVER: 9: input, "[[INPUT]]", c++, (device-sycl)
1516
// OFFLOAD-NEW_DRIVER: 10: preprocessor, {9}, c++-cpp-output, (device-sycl)
1617
// OFFLOAD-NEW_DRIVER: 11: compiler, {10}, ir, (device-sycl)
17-
// OFFLOAD-NEW_DRIVER: 12: backend, {11}, assembler, (device-sycl)
18-
// OFFLOAD-NEW_DRIVER: 13: assembler, {12}, object, (device-sycl)
19-
// OFFLOAD-NEW_DRIVER: 14: offload, "device-sycl (spir64-unknown-unknown)" {13}, object
20-
// OFFLOAD-NEW_DRIVER: 15: llvm-offload-binary, {8, 14}, image, (device-sycl)
21-
// OFFLOAD-NEW_DRIVER: 16: offload, "host-sycl (x86_64-unknown-linux-gnu)" {2}, "device-sycl (x86_64-unknown-linux-gnu)" {15}, ir
22-
// OFFLOAD-NEW_DRIVER: 17: backend, {16}, assembler, (host-sycl)
23-
// OFFLOAD-NEW_DRIVER: 18: assembler, {17}, object, (host-sycl)
24-
// OFFLOAD-NEW_DRIVER: 19: clang-linker-wrapper, {18}, image, (host-sycl)
18+
// OFFLOAD-NEW_DRIVER: 12: backend, {11}, ir, (device-sycl)
19+
// OFFLOAD-NEW_DRIVER: 13: offload, "device-sycl (spir64-unknown-unknown)" {12}, ir
20+
// OFFLOAD-NEW_DRIVER: 14: llvm-offload-binary, {7, 13}, image, (device-sycl)
21+
// OFFLOAD-NEW_DRIVER: 15: offload, "host-sycl (x86_64-unknown-linux-gnu)" {2}, "device-sycl (x86_64-unknown-linux-gnu)" {14}, ir
22+
// OFFLOAD-NEW_DRIVER: 16: backend, {15}, assembler, (host-sycl)
23+
// OFFLOAD-NEW_DRIVER: 17: assembler, {16}, object, (host-sycl)
24+
// OFFLOAD-NEW_DRIVER: 18: clang-linker-wrapper, {17}, image, (host-sycl)
2525

2626
/// Check the toolflow for SYCL compilation using new offload model
2727
// RUN: %clangxx -### --target=x86_64-unknown-linux-gnu -fsycl -fsycl-targets=spir64 --offload-new-driver %s 2>&1 | FileCheck -check-prefix=CHK-FLOW %s
@@ -96,10 +96,10 @@
9696
// RUN: | FileCheck -check-prefix=CHK_ARCH \
9797
// RUN: -DTRIPLE=amdgcn-amd-amdhsa -DARCH=gfx900 %s
9898
// RUN: %clangxx -### --target=x86_64-unknown-linux-gnu -fsycl \
99-
// RUN: -fno-sycl-libspirv -fsycl-targets=nvidia_gpu_sm_50 \
99+
// RUN: -fno-sycl-libspirv -fsycl-targets=nvidia_gpu_sm_75 \
100100
// RUN: -nogpulib --offload-new-driver %s 2>&1 \
101101
// RUN: | FileCheck -check-prefix=CHK_ARCH \
102-
// RUN: -DTRIPLE=nvptx64-nvidia-cuda -DARCH=sm_50 %s
102+
// RUN: -DTRIPLE=nvptx64-nvidia-cuda -DARCH=sm_75 %s
103103
// CHK_ARCH: clang{{.*}} "-triple" "[[TRIPLE]]"
104104
// CHK_ARCH-SAME: "-fsycl-is-device" {{.*}} "--offload-new-driver"{{.*}} "-o" "[[CC1DEVOUT:.+\.bc]]"
105105
// CHK_ARCH-NEXT: llvm-offload-binary{{.*}} "--image=file=[[CC1DEVOUT]],triple=[[TRIPLE]],arch=[[ARCH]]{{.*}},kind=sycl{{.*}}"
@@ -174,7 +174,7 @@
174174
// RUN: %clangxx -fsycl -### -fsycl-targets=nvptx64-nvidia-cuda \
175175
// RUN: -fno-sycl-libspirv -nocudalib --offload-new-driver %s 2>&1 \
176176
// RUN: | FileCheck -check-prefix NVPTX_DEF_ARCH %s
177-
// NVPTX_DEF_ARCH: llvm-offload-binary{{.*}} "--image=file={{.*}},triple=nvptx64-nvidia-cuda,arch=sm_50,kind=sycl"
177+
// NVPTX_DEF_ARCH: llvm-offload-binary{{.*}} "--image=file={{.*}},triple=nvptx64-nvidia-cuda,arch=sm_75,kind=sycl"
178178

179179
/// check for -sycl-embed-ir transmission to clang-linker-wrapper tool
180180
// RUN: %clangxx -fsycl -### -fsycl-targets=nvptx64-nvidia-cuda \

0 commit comments

Comments
 (0)