Aurora Perego

Results 13 issues of Aurora Perego

That is a tough topic.. first a quick look into the usage of device global variables in SYCL (since [oneAPI 2023.2](https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/experimental/sycl_ext_oneapi_device_global.asciidoc)): ```cpp // declaration of the variable sycl::ext::oneapi::experimental::device_global myGlobVar; int...

Type:Enhancement
Backend:SYCL

As discussed this morning, these are the steps to compile [pixeltrack-standalone](https://github.com/cms-patatrack/pixeltrack-standalone): ```bash git clone https://github.com/AuroraPerego/pixeltrack-standalone.git -b alpaka_SYCL cd pixeltrack-standalone make environment . env.sh make -j alpaka ``` The version that...

The requirements / dependencies were needed for the buffers, so now can be removed. SYCL `queue`s and `event`s are thread-safe, the mutex in `QueueGenericSyclBase.hpp` has been removed as well. When...

Type:Refactoring
Backend:SYCL

The shared memory in SYCL currently is allocated with `sycl::local_accessor` that must be instantiated before the kernel and then their pointer is passed as an argument to the kernel. The...

Type:Bug
Type:Enhancement
Backend:SYCL

In CUDA/ROCm timing is disabled by default when [creating events](https://github.com/alpaka-group/alpaka/blob/a9f5b59da076e0371a54cb7c4158b50f116e13f5/include/alpaka/event/EventUniformCudaHipRt.hpp#L58), while in SYCL the profiling on the queue is [enabled](https://github.com/alpaka-group/alpaka/blob/a9f5b59da076e0371a54cb7c4158b50f116e13f5/include/alpaka/queue/sycl/QueueGenericSyclBase.hpp#L46). It would be nice to align all the backends to...

Type:Question
Type:Refactoring
Backend:CUDA
Backend:SYCL
Backend:HIP

**Bug summary** A simple application compiles but fails to run with the JIT compiler. AOT compilation with CUDA results in a working executable. I'm afraid I've missed something when building...

bug

following the request in #2248, this adds the same `trait` introduced previously for kernel arguments also for the kernel itself, with the same approach used in #2198.

As reported in #2259, with CUDA 12.4 there is a compile time error due to device global variables, because `inline __device__` is not allowed by CUDA. This removes the `inline`...

should fix #2285 The example provided in the issue does not compile now, it says that there is no `sin` function in the `alpaka` namespace. I can squash the commits...

The `allocMappedBuf` method was already there, but when calling `allocMappedBufIfSupported` it was not used because the trait `hasMappedBufSupport` has never been specialized to be `true`. This led to spotting a...