[code-reflection] Integrated: cuda fixes for violajones
Gary Frost
gfrost at openjdk.org
Thu May 30 11:29:23 UTC 2024
Found source of access violation in CUDA backend.
Also a bug in the algorithm itself (causing to walk off array).
Violajones still does not work for larger Nasa image. But does for other three image types.
-------------
Commit messages:
- cuda fixes for violajones
Changes: https://git.openjdk.org/babylon/pull/105/files
Webrev: https://webrevs.openjdk.org/?repo=babylon&pr=105&range=00
Stats: 315 lines in 8 files changed: 110 ins; 94 del; 111 mod
Patch: https://git.openjdk.org/babylon/pull/105.diff
Fetch: git fetch https://git.openjdk.org/babylon.git pull/105/head:pull/105
PR: https://git.openjdk.org/babylon/pull/105
More information about the babylon-dev
mailing list