Skip to content

23 SYCL e2e tests failed as "cannot select nvvm intrinsic" error #20109

@fzou1

Description

@fzou1

The failed tests are:

SYCL :: AtomicRef/compare_exchange.cpp 
SYCL :: AtomicRef/compare_exchange_local.cpp 
SYCL :: AtomicRef/max.cpp 
SYCL :: AtomicRef/max_local.cpp 
SYCL :: AtomicRef/min.cpp 
SYCL :: AtomicRef/min_local.cpp 
SYCL :: BFloat16/bfloat16_vec.cpp 
SYCL :: Basic/code_location_e2e.cpp 
SYCL :: Basic/compare_exchange_strong.cpp 
SYCL :: Basic/stream/auto_flush.cpp 
SYCL :: Basic/stream/release_resources_test.cpp 
SYCL :: Basic/stream/stream.cpp 
SYCL :: Basic/stream/stream_copies_buffer_sync.cpp 
SYCL :: Basic/stream/stream_max_stmt_exceed.cpp 
SYCL :: Basic/stream/zero_buffer_size.cpp 
SYCL :: Basic/vector/scalar_access.cpp 
SYCL :: Complex/sycl_complex_stream_test.cpp 
SYCL :: Reduction/reduction_big_data.cpp 
SYCL :: Reduction/reduction_nd_ext_double.cpp 
SYCL :: Reduction/reduction_usm.cpp 
SYCL :: Reduction/reduction_usm_dw.cpp 
SYCL :: Regression/unoptimized_stream.cpp 
SYCL :: syclcompat/atomic/atomic_class.cpp

The failure log:

FAIL: SYCL :: AtomicRef/min_local.cpp (175 of 2479)
 ******************** TEST 'SYCL :: AtomicRef/min_local.cpp' FAILED ********************
 Exit Code: 1
 
 Command Output (stdout):
 --
 # RUN: at line 1
 /__w/llvm/llvm/toolchain/bin/clang++ -Werror -fpreview-breaking-changes  -fsycl -fsycl-targets=amdgcn-amd-amdhsa,spir64,nvptx64-nvidia-cuda -Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch=gfx1030  /__w/llvm/llvm/llvm/sycl/test-e2e/AtomicRef/min_local.cpp -o /__w/llvm/llvm/build-e2e/AtomicRef/Output/min_local.cpp.tmp.out
 # executed command: /__w/llvm/llvm/toolchain/bin/clang++ -Werror -fpreview-breaking-changes -fsycl -fsycl-targets=amdgcn-amd-amdhsa,spir64,nvptx64-nvidia-cuda -Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch=gfx1030 /__w/llvm/llvm/llvm/sycl/test-e2e/AtomicRef/min_local.cpp -o /__w/llvm/llvm/build-e2e/AtomicRef/Output/min_local.cpp.tmp.out
 # .---command stderr------------
 # | fatal error: error in backend: Cannot select: intrinsic %llvm.nvvm.atomic.cas.shared.i
 # | llvm-foreach: 
 # | clang++: error: clang frontend command failed with exit code 70 (use -v to see invocation)
 # | clang version 22.0.0git (https://github.com/intel/llvm 13197966d75997046e070e13f5ff8bb836f5e01d)
 # | Target: x86_64-unknown-linux-gnu
 # | Thread model: posix
 # | InstalledDir: /__w/llvm/llvm/toolchain/bin
 # | Build config: +assertions
 # | clang++: note: diagnostic msg: Error generating preprocessed source(s).
 # `-----------------------------
 # error: command failed with exit status: 1

Test job link: https://github.com/intel/llvm/actions/runs/17654366077/job/50173192052?pr=19997

To reproduce

Environment

Additional context

No response

Metadata

Metadata

Assignees

Labels

bugSomething isn't workingconfirmedcudaCUDA back-end

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions