alpaka
alpaka copied to clipboard
Abstraction Library for Parallel Kernel Acceleration :llama:
I checked the following CI job from our dev branch: https://gitlab.com/hzdr/crp/alpaka/-/jobs/7619007751 All tests together takes 8 minutes to run. There are a few tests, which takes a big amount of...
A compute bound benchmark depends on A hydrodynamics problem to solve the compressible Euler equations in 3D, using an explicit, second-order method. (Currently it is 3D but could be 2D...
It looks like checking the maximum grid size with `alpaka::getAccDevProps(dev)` and launching a kernel with twice as many blocks _does not_ fail, at least on some CPU backends. See the...
I am not sure if this is a real issue. Feel free to close it, if it's not of interest. We had a nasty situation where in our github CI...
A common pattern in the CMS software has been to fix the block size at compile time and determine the number of blocks at runtime based on the problem size,...
I've building the alpaka tests with gcc 13 (tested also 12) and CUDA 12.5 (tested also 12.4) on Ubuntu 22.04, with ```bash cmake /home/fwyzard/src/alpaka-group/alpaka \ -G 'Unix Makefiles' \ --log-level=VERBOSE...
Babelstream benchmark results are in txt format. They are converted to json by this script. Commands : - To convert txt output to Json for Nocodb input: `python3 benchmarkXml2Json.py BabelStreamBenchmarkResults.txt...
Refactor bash CI scripts and add guards to guaranty, that every configuration environment variable is used in the cmake configuration. The PR is motivated by PR #2308, where the `alpaka_CXX_STANDARD`...