Pierre Paleo

Hi, Slightly related to #2399 : in `seafile-server-8.0.7`, it seems that the python interpreter is hardcoded (to `python3.6`) in the `seahub/thirdpart/bin/` files: ```bash $ head -1 seahub/thirdpart/bin/* ``` gives ```...

Enumerating devices creates a context not released

2

Hi, I'm using pocl through [pyopencl](https://documen.tician.de/pyopencl). When enumerating the devices on the POCL platform, contexts are created on each GPU. I only tried with Nvidia GPU though, not with AMD...

This PR adds a small utility for easily importing/exporting FFTW plans, called [wisdom](https://www.fftw.org/fftw3_doc/Words-of-Wisdom_002dSaving-Plans.html). The wisdom data structure is saved in a numpy `npz` file, along with some "metadata" (eg. machine...

`Histogramnd` crashes on arrays with more than `2**31 - 1` elements: ```python import numpy as np from silx.math import Histogramnd def test_histogram(n_samples, n_bins): data = np.random.randint(0, high=1000, size=(n_samples,), dtype=np.int32) dmin,...

This PR addresses #2630. WIP: - [ ] Convolution (`opencl.convolution`) - [ ] Fast Fourier Transform (`math.fft`) - [ ] Filtered Back-projection (`opencl.backprojection`)

This is an internal refactoring of the OpenCL projector (Radon transform), see #2493. The PR #2454 should be merged beforehand. Work in progress: - [x] Refactoring - [x] Input and...

[WIP] Fix operations issues on large arrays

9

## About This MR fixes a "pycuda hanging forever" issue when array sizes exceed `2**34` bytes. It's done by replacing some occurrences of `unsigned (int)` with `size_t` in template kernels...

pycuda issue when using arrays bigger than 17 GB

3

**Describe the bug** It seems that pycuda is not able to use arrays bigger than 17 GB. Allocating (with `gpuarray.empty` or `gpuarray.zeros`) works, but any subsequent operation on the array...

[OpenCL] "Cannot allocate memory" issues on PPC64

1

Any OpenCL test run on our Power9 machine results in the following error (with my environment): ```bash PYOPENCL_CTX="0:1" ./run_tests.py silx.opencl.test.test_addition.suite [...] OSError: [Errno 12] Cannot allocate memory ``` The reason...

Follow-up of #2696/#2698. Previously, pyopencl code used ```python d_array = pyopencl.array.zeros(queue, shape, dtype) ``` Because of the issues with `pocl`, this was replaced with ```python d_array = pyopencl.array.empty(queue, shape, dtype)...