dpctl
dpctl copied to clipboard
Python SYCL bindings and SYCL-based Python Array API library
```python import dpctl.tensor as dpt a = dpt.ones((8192, 8192), device='cpu', dtype='f4') b = dpt.ones((8192, 8192), device='cpu', dtype=bool) %timeit a[b] #211 ms ± 6.9 ms per loop (mean ± std. dev....
 The exit code indicates "stack buffer overflow" on Windows (see [ref](https://stackoverflow.com/questions/50562192/process-finished-with-exit-code-1073740791-0xc0000409-pycharm-error) ). The crash is sporadic, but persistently recurring. This issue is to reproduce it in a controlled environment...
- From `dpctl` to `pytorch`: ```python import dpctl.tensor as dpt import intel_extension_for_pytorch import torch array_dpctl_cpu = dpt.reshape(dpt.arange(1000, device="cpu", dtype=dpt.float32), (4, 250)) array_torch_cpu = torch.from_dlpack(array_dpctl_cpu) ``` fails with: ``` RuntimeError: Data...
```python import dpctl.tensor as dpt a = dpt.ones((8192, 8192), dtype='i4', device='cpu') b = dpt.ones((8192 + 2, 8192 + 2), dtype='i4', device='cpu') %timeit b[2:, 2:]+=a #209 ms ± 36.8 ms per...
cmake first finds IntelSycl and then fails: ``` ===> Building for py39-dpctl-0.14.2 * Getting build dependencies for wheel... * Building wheel... -------------------------------------------------------------------------------- -- Trying 'Ninja' generator -------------------------------- --------------------------- ---------------------- -----------------...
The C API for `dpctl` and `dpctl.tensor.usm_ndarray` currently is auto-generated using Cython. A problem with using Cython to auto-generate the headers is the complete lack of API documentation. For `dpctl`,...
The API documentation for `dpctl.tensor` is buried under `API Documentation -> dpctl -> Submodules -> dpctl.tensor` on the documentation site. The organization makes it hard to find for the casual...
``` python import dpctl.tensor as dpt #Default type is float64. Xcpu = dpt.ones(4, device="cpu") #GPU without float64 support. Should have been cast to dtype float32, but failed. dpt.asarray(Xcpu, device="gpu") ```
The #1051 modified behavior of `linspace` outside of scope of array-API spec (presumable less common/useful combination of parameters). It is an improvement towards aligning with NumPy's behavior, but it only...