Commits · main · Kaushik Kulkarni / pycuda

Jan 10, 2023
- Placate flake8-comprehensions · ad519613
  Andreas Klöckner authored Jan 10, 2023
  
  ad519613
- Update compyte · c0c2e10f
  Andreas Klöckner authored Jan 10, 2023
```
https://github.com/inducer/compyte/pull/43
```
  c0c2e10f
Jan 05, 2023
- Preserve strides during copy (closes gh-403) · a679708b
  Andreas Klöckner authored Jan 05, 2023
  
  a679708b
Dec 21, 2022
- Bump version to 2022.2.2 · 41f9119a
  Andreas Klöckner authored Dec 20, 2022
  
  41f9119a
- Fix version ifdef for CU_TARGET_COMPUTE_86 · 54ce4293
  Andreas Klöckner authored Dec 20, 2022
  
  54ce4293
Dec 20, 2022
- Bump version to 2022.2.1 · 62bbf049
  Andreas Klöckner authored Dec 20, 2022
  
  62bbf049
- Drop np.int from curandom for numpy deprecation · f426d482
  Andreas Klöckner authored Dec 20, 2022
  
  f426d482
- Switch np.bool -> np.bool_ to track removal in numpy · 4fca42ec
  Andreas Klöckner authored Dec 20, 2022
  
  4fca42ec
- Update JIT targets (CUDA 12 deprecations, plus additions) (closes gh-402) · 3f67b078
  Andreas Klöckner authored Dec 20, 2022
  
  3f67b078
Nov 21, 2022
- Bump version to 2022.2 · e4122c77
  Andreas Klöckner authored Nov 21, 2022
  
  e4122c77
- Upgrade to bpl-subset with upstream Py3.11 fixes (closes gh-359) · 304138a4
  Andreas Klöckner authored Nov 21, 2022 and Andreas Klöckner committed Nov 21, 2022
  
  304138a4
- Update _pvt_struct for Py3.11 compatibility · 071a5fb9
  Andreas Klöckner authored Nov 21, 2022 and Andreas Klöckner committed Nov 21, 2022
  
  071a5fb9
- Add struct module derivation to LICENSE · 96d64087
  Andreas Klöckner authored Nov 21, 2022 and Andreas Klöckner committed Nov 21, 2022
  
  96d64087
- Drop Py2-supporting _pvt_struct module · 11d7c04a
  Andreas Klöckner authored Nov 21, 2022 and Andreas Klöckner committed Nov 21, 2022
  
  11d7c04a
Nov 01, 2022
- use oldest-supported-numpy · 1504e7a5
  Isuru Fernando authored Oct 31, 2022 and Andreas Klöckner committed Oct 31, 2022
  
  1504e7a5
Oct 17, 2022
- Pin build-time numpy versions for Python 3.10 and 3.11 · af3c860b
  Bruce Merry authored Oct 17, 2022 and Andreas Klöckner committed Oct 17, 2022
```
This avoid errors when trying to use with a numpy version that's older
than the one that was newest when pycuda was built. See
https://github.com/inducer/pycuda/issues/271 for background on why this
is useful.
```
  af3c860b
- Fix noncontiguous transpose (closes gh-385) · 488a25e2
  Andreas Klöckner authored Oct 16, 2022
  
  488a25e2
Oct 14, 2022
- Bump versions of various github actions components · 9c2c982b
  Andreas Klöckner authored Oct 14, 2022 and Andreas Klöckner committed Oct 14, 2022
  
  9c2c982b
Sep 27, 2022
- Switch from size_t to ssize_t · d877812d
  Pierre Paleo authored Sep 26, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  d877812d
- Speed-up test_big_array_elementwise · 8ab39e8d
  Pierre Paleo authored Sep 22, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  8ab39e8d
- Rework tests to use 1-byte data types · a938975e
  Pierre Paleo authored Sep 22, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  a938975e
- Simplify test_big_array_scan · bcf8a150
  Pierre Paleo authored Sep 21, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  bcf8a150
- Add test for scan case · dc3fc53d
  Pierre Paleo authored Sep 21, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  dc3fc53d
- Add test for reduction case · 18ba09d8
  Pierre Paleo authored Sep 21, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  18ba09d8
- Add test for elementwise case · b32e45ed
  Pierre Paleo authored Sep 21, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  b32e45ed
- get_elwise_range_module: use "long" data type for running index, as it should be signed · dd95f7c9
  Pierre Paleo authored Sep 21, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  dd95f7c9
- Details matter · 0acc4f73
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  0acc4f73
- Fix prepared calls, cast to size_t for min() · 3fe9bfcf
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  3fe9bfcf
- Prepare kernel call with size_t for second param · 6a4145b6
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  6a4145b6
- Use size_t only for "n" argument · 8b527553
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  8b527553
- Replace "unsigned (int)" with "size_t" in scan.py · 9c1ad273
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  9c1ad273
- Replace "unsigned (int)" with "size_t" in reduction.py · 6390008a
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  6390008a
- Replace "unsigned (int)" with size_t in elementwise.py · 07beaacf
  Pierre Paleo authored Sep 19, 2022 and Andreas Klöckner committed Sep 27, 2022
  
  07beaacf
Sep 23, 2022

Prevent warning on nvcc_path add · 867ea556

George Gensure authored Sep 23, 2022 and

Andreas Klöckner committed Sep 23, 2022

If pycuda has detected a CUDA runtime via nvcc.exe discovery, return to prevent a misleading warning, consistent with CUDA_PATH usage above.

867ea556

Sep 13, 2022
- Quote type subscripts to avoid py<=3.8 incompatibility (closes gh-382) · 5fbaa4cf
  Andreas Klöckner authored Sep 13, 2022
  
  5fbaa4cf
Sep 11, 2022
- test unequal dtype arith ops (fixes gh-372) · f9c821be
  Kaushik Kulkarni authored Aug 26, 2022 and Andreas Klöckner committed Sep 11, 2022
  
  f9c821be
- handle real casting logic for complex arith ops · 1dcaf262
  Kaushik Kulkarni authored Aug 26, 2022 and Andreas Klöckner committed Sep 11, 2022
  
  1dcaf262
Sep 04, 2022
- Update bpl subset for possible py3.11 compat · 697139f4
  Andreas Klöckner authored Sep 04, 2022
  
  697139f4
Aug 16, 2022
- Add warning about out-of-stream cleanup · 6f60fe4e
  Andreas Klöckner authored Aug 15, 2022 and Andreas Klöckner committed Aug 16, 2022
  
  6f60fe4e
- Use importlib to get the pycuda header location · 4e631349
  Kaushik Kulkarni authored Aug 16, 2022 and Andreas Klöckner committed Aug 16, 2022
  
  4e631349