Setting the ptx version manually

[tdavidcl]

Hi,
I want to run my SYCL code on a GPU node from my university. But due to a non standard CUDA installation the PTX version seem's to not be correctly detected. It is set to ptx75 instead of ptx72 (beacause we have CUDA 11.2 which support ptx up to .version 7.2) , giving me the following output :

/local/tdavidcl/sycl_workspace/dpcpp/bin/clang++ -fsycl -std=c++17 -pedantic-errors -fsycl-targets=nvptx64-nvidia-cuda -Xsycl-target-backend=nvptx64-nvidia-cuda main.cpp.o -o main
clang-14: warning: CUDA version is newer than the latest supported version 11.5 [-Wunknown-cuda-version]
ptxas /tmp/main-sm_50-227eb5.s, line 5; fatal   : Unsupported .version 7.5; current version is '7.2'
ptxas fatal   : Ptx assembly aborted due to errors
llvm-foreach:
clang-14: error: ptxas command failed with exit code 255 (use -v to see invocation)
clang version 15.0.0 (https://github.com/intel/llvm 8e30ff36465387ee7cdd7c98f0a00c6aea6de505)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /local/tdavidcl/sycl_workspace/dpcpp/bin
clang-14: note: diagnostic msg: Error generating preprocessed source(s) - no preprocessable inputs.

indead looking at the begining of /tmp/main-sm_50-227eb5.s the version is set to 7.5 instead of 7.2 :

//
// Generated by LLVM NVPTX Back-End
//

.version 7.5
.target sm_50
.address_size 64

output of sycl-ls :

../../dpcpp/bin/sycl-ls
[opencl:cpu:0] Portable Computing Language, pthread-AMD EPYC 7742 64-Core Processor 1.2 [1.6]
[opencl:cpu:1] AMD Accelerated Parallel Processing, AMD EPYC 7742 64-Core Processor 1.2 [1912.5 (sse2,avx)]
[opencl:cpu:2] Intel(R) OpenCL, AMD EPYC 7742 64-Core Processor                 1.2 [1.2.0.25]
[ext_oneapi_cuda:gpu:0] NVIDIA CUDA BACKEND, GeForce RTX 2080 SUPER 0.0 [CUDA 11.2]
[ext_oneapi_cuda:gpu:1] NVIDIA CUDA BACKEND, GeForce RTX 2080 SUPER 0.0 [CUDA 11.2]
[host:host:0] SYCL host platform, SYCL host device 1.2 [1.2]

output of clang++ --version :

clang version 15.0.0 (https://github.com/intel/llvm 8e30ff36465387ee7cdd7c98f0a00c6aea6de505)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /local/tdavidcl/sycl_workspace/shamrock/build/../../dpcpp/bin

since --cuda-gpu-arch=sm72 doesn't set the ptx version how can I change it ?