dpnp
dpnp copied to clipboard
IntelPython
Data Parallel Extension for NumPy
When we have dpnp installed in a Conda environment, how can we test the package?
## Description All `fft.fft` tests (third_party(cupy) and dpnp own) are failed on both devices (GPU and CPU). On GPU dpnp.fft.fft call occurs Fatal error. ## Reproduce See #819
Original Numpy is able to work with negative inputs. It should be easy enough to support at least some cases. Please look into changes of PR #773 for more details.
dpnp/backend/examples/example_bs.cpp can not be compiled because interface of dpnp_divide_c has changed
Related to https://github.com/IntelPython/dpctl/pull/184. It is a part of movement dparray from dpNP to dpCtl. Can I propose PR or dpNP team will do that? Are there in dpNP some places...
run by ``` export DPNP_BACKEND_TESTS_ENABLE=1 ./0.build.sh ``` main backend files compile options looks like: ``` dpcpp -DDPNP_LOCAL_QUEUE=1 -DMKL_ILP64=1 -DONEDPL_USE_PREDEFINED_POLICIES=0 -DPSTL_USE_PARALLEL_POLICIES=0 -D_FORTIFY_SOURCE=2 -D_GLIBCXX_USE_TBB_PAR_BACKEND=0 -O3 -DNDEBUG -fPIC -W -Wextra -Wshadow -Wall -Wstrict-prototypes...
Need to implement `random` module in the project. Currently, it implemented in Python. Need to implement it in Cython and cover following functionality https://docs.scipy.org/doc/numpy-1.13.0/reference/routines.random.html ### Simple random data - [x]...
The following program is resulting in error: dpctl: [fd63a3e2b15942e0b93f98e7dbbb14ae49e5a724](https://github.com/IntelPython/dpctl/commit/fd63a3e2b15942e0b93f98e7dbbb14ae49e5a724) dpnp: [43a4b538722b26ca00b0a475e44a4a5f8376267c](https://github.com/IntelPython/dpnp/commit/43a4b538722b26ca00b0a475e44a4a5f8376267c) ``` import dpctl import dpnp with dpctl.device_context("level0:gpu:0"): gpu_result = dpnp.random.random_sample(9) with dpctl.device_context("opencl:gpu:0"): gpu_result = dpnp.random.random_sample(9) ``` This is the...
linalg.cholesky tests fail on GPU. I marked these tests as skipped in this PR https://github.com/IntelPython/dpnp/pull/553. Need to be fixed, then return tests.
`eig`, `eigvals`, `svd` functions give SegFault in the end of program execution (freeing resources)
``` $ DPNP_QUEUE_GPU=1 pytest tests/test_linalg.py ============================================================================================================ test session starts ============================================================================================================ platform linux -- Python 3.7.7, pytest-5.4.3, py-1.9.0, pluggy-0.13.1 rootdir: /localdisk/work/amakarye/git/dpnp, inifile: setup.cfg plugins: hypothesis-6.1.1 collecting ... DPNP: current queue is...