How to Install, Compile and Run SYCL using NVIDIA GPUs


Build a C++ Complier for SYCL

#Create folder; Assume is as $DPCPP_HOME
mkdir $HOME/install
cd $HOME/install

# Build INTEL's LLVM
# get sycl branch
git clone -b sycl
cd llvm

# configure
python ./buildbot/ --cuda -t release --cmake-gen "Unix Makefiles"

# Assuming no error. Build takes 20mins. Get a coffee!
cd build
make sycl-toolchain -j `nproc`
make install

# assumes $HOME/install/llvm installation

HelloWorld SYCL - Compile and Run

#include <sycl/sycl.hpp>

int main() {
  // Creating buffer of 4 ints to be used inside the kernel code
  sycl::buffer<sycl::cl_int, 1> Buffer(4);

  // Creating SYCL queue
  sycl::queue Queue;

  // Size of index space for kernel
  sycl::range<1> NumOfWorkItems{Buffer.size()};

  // Submitting command group(work) to queue
  Queue.submit([&](sycl::handler &cgh) {
    // Getting write only access to the buffer on a device
    auto Accessor = Buffer.get_access<sycl::access::mode::write>(cgh);
    // Executing kernel
    cgh.parallel_for<class FillBuffer>(
        NumOfWorkItems, [=](sycl::id<1> WIid) {
          // Fill buffer with indexes
          Accessor[WIid] = (sycl::cl_int)WIid.get(0);

  // Getting read only access to the buffer on the host.
  // Implicit barrier waiting for queue to complete the work.
  const auto HostAccessor = Buffer.get_access<sycl::access::mode::read>();

  // Check the results
  bool MismatchFound = false;
  for (size_t I = 0; I < Buffer.size(); ++I) {
    if (HostAccessor[I] != I) {
      std::cout << "The result is incorrect for element: " << I
                << " , expected: " << I << " , got: " << HostAccessor[I]
                << std::endl;
      MismatchFound = true;

  if (!MismatchFound) {
    std::cout << "The results are correct!" << std::endl;

  return MismatchFound;

# do it
export PATH=$HOME/install/llvm/build/bin:$PATH
export LD_LIBRARY_PATH=$HOME/install/llvm/build/lib:$LD_LIBRARY_PATH

# verify
$ clang++ --version
clang version 16.0.0
clang++ -fsycl -fsycl-targets=nvptx64-nvidia-cuda simple-sycl-app.cpp -o simple-sycl-app-cuda

The results are correct!

Non-default CUDA installation

If CUDA installation folder is different or default CUDA is not the required version. Modify only two steps from above rest is the same. Assuming SYCL to use this /usr/local/cuda-11.5 version.

# configure
python3 ./buildbot/ --cuda -t release --cmake-gen "Unix Makefiles" --cmake-opt="-DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-11.5"


# compile
clang++ -fsycl -fsycl-targets=nvptx64-nvidia-cuda simple-sycl-app.cpp -o simple-sycl-app-cuda --cuda-path=/usr/local/cuda-11.5/

That’s it. TATA.


★ 5 min read · Rajesh Pandian M · sycl , cuda , nvidia , intel , oneapi , dpcpp