SYCL: feature test compile failed!!

[evstratios]

Don't delete my question.

I've been trying to get Intel Extension For PyTorch

to compile since August 2024..

Intel's script first tries to compile PyTorch for compatibility.

I can't get past PyTorch's compilation to get to compile IPEX.

I actually did get through PyTorch once but it had some error and wouldn't actually load in python.

Could you people please help me!!!

I don't understand why you only provide binaries

for your thousand dollar data center gpu cards

and you leave your enormous numbers of people

who have qualified gpus to fend for ourselves!?!!?

I am stuck. The AIs I've been using are stuck.

I need help.

I have ADL-P - I paid my money too.

Please don't give me beginner advice for idiots.

I've been at this all day every day for seven months.

I need actual help that gets me finished with this successfully.

I'm watching the AI industry pass me by because

Intel refuses to give us binaries! I'm lost. I need help.

-- Found SYCL: /opt/intel/oneapi/compiler/2025.0/include;/opt/intel/oneapi/compiler/2025.0/include/sycl;/opt/intel/oneapi/compiler/2025.0/include/sycl (found version "20250004")
SYCL: feature test compile failed!!
compile output is:
CMake Error at third_party/torch-xpu-ops/cmake/Modules/FindSYCLToolkit.cmake:237 (message):
Can not build SYCL_CMPLR_TEST
Call Stack (most recent call first):
third_party/torch-xpu-ops/cmake/SYCL.cmake:5 (include)
third_party/torch-xpu-ops/CMakeLists.txt:38 (include)

-- Configuring incomplete, errors occurred!
Building wheel torch-2.6.0a0+git1eba9b3
-- Building version 2.6.0a0+git1eba9b3
cmake -GNinja -DBUILD_BINARY=0 -DBUILD_BY_PER_KERNEL=0 -DBUILD_CAFFE2=0 -DBUILD_CAFFE2_OPS=0 -DBUILD_PYTHON=True -DBUILD_STATS=1 -DBUILD_TEST=True -DBUILD_WITH_SANITIZER=0 -DBUILD_WITH_XPU=1 -DCMAKE_ARGS=-DCMAKE_AR=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-ar -DCMAKE_CXX_COMPILER_AR=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-gcc-ar -DCMAKE_C_COMPILER_AR=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-gcc-ar -DCMAKE_RANLIB=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-ranlib -DCMAKE_CXX_COMPILER_RANLIB=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-gcc-ranlib -DCMAKE_C_COMPILER_RANLIB=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-gcc-ranlib -DCMAKE_LINKER=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-ld -DCMAKE_STRIP=/home/evstratios/miniforge3/envs/ipex-xe/bin/x86_64-conda-linux-gnu-strip -DCMAKE_BUILD_TYPE=Release -DBUILD_PYTHON=True -DCMAKE_BUILD_TYPE=Release -DCMAKE_SYCL_DEVICE_OBJECT_OUTPUT_DIR=./sycl_objs -DCMAKE_INSTALL_PREFIX=/home/evstratios/ipex/pytorch/torch -DONEDNN_GRAPH_BUILD_COMPILER_BACKEND=1 -DSYCL_OCLOC_EXECUTABLE=/usr/bin/ocloc -DSYCL_PI_LEVEL_ZERO_DEVICE_SCOPE_EVENTS=1 -DSYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1 -DTORCH_XPU_ARCH_LIST=adl-p -DUSE_XPU=1 -GNinja -Wno-dev -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/home/evstratios/ipex/pytorch/torch -DCMAKE_PREFIX_PATH=/home/evstratios/miniforge3/envs/ipex-xe/lib/python3.10/site-packages;/home/evstratios/miniforge3/envs/ipex-xe:/opt/intel/oneapi/pti/0.10/lib/cmake/pti:/opt/intel/oneapi/mkl/2025.0/lib/cmake:/opt/intel/oneapi/compiler/2025.0:/opt/intel/oneapi/vpl/2023.1.0:/opt/intel/oneapi/tbb/2022.1/env/..:/opt/intel/oneapi/pti/0.11/lib/cmake/pti:/opt/intel/oneapi/mkl/2025.1/lib/cmake:/opt/intel/oneapi/ispc/1.23/lib/cmake/ispcrt:/opt/intel/oneapi/ipp/2022.1/lib/cmake/ipp:/opt/intel/oneapi/dpl/2022.8/lib/cmake/oneDPL:/opt/intel/oneapi/dnnl/2025.1/lib/cmake:/opt/intel/oneapi/dal/2025.4:/opt/intel/oneapi/compiler/2025.1:/opt/intel/oneapi/ccl/2021.15/lib/cmake/oneCCL:/home/evstratios/miniforge3/envs/ipex-xe:/home/evstratios/miniforge3/envs/ipex-xe/x86_64-conda-linux-gnu/sysroot/usr -DGLIBCXX_USE_CXX11_ABI=1 -DJAVA_HOME=/home/evstratios/miniforge3/envs/ipex-xe/lib/jvm -DPython_EXECUTABLE=/home/evstratios/miniforge3/envs/ipex-xe/bin/python -DTORCH_BUILD_VERSION=2.6.0a0+git1eba9b3 -DTORCH_XPU_ARCH_LIST=adl-p -DUSE_ABI=1 -DUSE_AOT_DEVLIST=adl-p -DUSE_CCACHE=1 -DUSE_CHANNELS_LAST_1D=0 -DUSE_COREML_DELEGATE=1 -DUSE_CPP_CODE_COVERAGE=1 -DUSE_CUDA=0 -DUSE_CUDNN=0 -DUSE_DISTRIBUTED=0 -DUSE_EIGEN_FOR_BLAS=1 -DUSE_FAKELOWP=0 -DUSE_FBGEMM=0 -DUSE_FFMPEG=1 -DUSE_GLOO=1 -DUSE_GOLD_LINKER=1 -DUSE_IBVERB=0 -DUSE_ITT_ANNOTATION=1 -DUSE_KINETO=0 -DUSE_LEVEL_ZERO=1 -DUSE_MKLDNN=1 -DUSE_MKLDNN_ACL=0 -DUSE_MPI=0 -DUSE_NNPACK=1 -DUSE_NUMA=0 -DUSE_NUMPY=True -DUSE_ONEMKL=1 -DUSE_OPENCL=1 -DUSE_OPENCV=1 -DUSE_OPENMP=1 -DUSE_PYTORCH_QNNPACK=1 -DUSE_QNNPACK=0 -DUSE_RPC=1 -DUSE_SPLIT_FP64_LOOPS=1 -DUSE_STATIC_MKL=1 -DUSE_SYCL_ASSERT=0 -DUSE_SYSTEM_LIBS=0 -DUSE_TENSORPIPE=1 -DUSE_TORCH_XPU_OPS=0 -DUSE_UCC=0 -DUSE_VULKAN=0 -DUSE_XETLA=0 -DUSE_XNNPACK=1 -DUSE_XPU=1 /home/evstratios/ipex/pytorch