Ronan Keryell

Results 101 issues of Ronan Keryell

Some examples following some discussions inside the SYCL committee about kernel language restrictions. https://gitlab.khronos.org/sycl/Specification/issues/168

Clarification

The current accessor implementation relies on the usual easy detail::shared_ptr_implementation but requires some painful magical Clang rewriting of lambda captures to use drt::accessor instead. Instead, make a new accessor implementation...

enhancement

During San Diego C++ SG1 committee, Pablo Halpern (Intel) presented "Context Tokens for Parallel Algorithms" www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0335r1.pdf which is pretty close to the SYCL command-group handler when we have kernels with...

extension

While SYCL is a standard, some implementations are better suited for some devices and it makes sense in a full platform to take advantage of multiple SYCL implementation in the...

extension

I was stunned by the presentation from David Hollman at the last ISO C++ meeting about how to manage properties in the context of the executor mechanism http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1393r0.html It is...

extension

SYCL 1.2.1 provides OpenCL interoperability only with the OpenCL C host API. triSYCL also provides interoperability through `boost::compute` https://github.com/triSYCL/triSYCL/issues/210 It would be interesting to get some interoperability with some other...

extension

While mentioned in `doc/architecture.rst` it is not described in `doc/macros.rst`.

Clarification

Right now the Makefile has the name of 1 kernel hard-coded...

bug
enhancement

Since ranges are coming with C++20, add some range interfaces to SYCL instead of using pair of iterators, such as: ``` std::vector v { 1, 2, 3}; cl::sycl:buffer b {...

extension

OpenMP is used now for work-group parallelism, for SIMD parallelism inside a work-group and to implement intra-work-group barriers. It would be interesting to have some benchmarking results on a multicore...

question