Block a user
Overlay-Next: grid added to ShapeCache
So FYI. This allocate resolution * resolution * 6
amount of vertices with uninitialized memory. It does change the size of the container. So when you append to it you just increase this size without setting the already allocated data.
Clément Foucault
pushed to overlay-next-attribute-viewer at laurynas/blender
2024-09-30 12:27:36 +02:00
e30fca8abe
Merge branch 'main' into overlay-next-attribute-viewer
ce7667f3b1
GPv3: Join Operator
91eac6c771
Cleanup: Make format
ac24912736
Fix #128315: EEVEE: Uneven cryptomatte levels
2d5591b63f
Cleanup: mark unused variables as such
Clément Foucault
pushed to overlay-next-attribute-viewer at laurynas/blender
2024-09-30 12:19:47 +02:00
60a2e6253b
Fix sample overlap
c2b565694e
Fix wrong additional infos.
Metal: Add support for parallel compilation and precompilation specialisation
Fix #128164: Race condition in ShaderCreateInfo::finalize()
Ok then, here is another approach I can see working:
- Make all static create infos call
finalize
on startup. - Make a special variant of the ShaderCreateInfo that does not do any recursive…
Clément Foucault
pushed to Metal_parallel_compilation_support at Jason-Fielder/blender
2024-09-30 11:13:04 +02:00
e2395f3f0d
Apply clang format
Clément Foucault
pushed to Metal_parallel_compilation_support at fclem/blender
2024-09-30 11:12:45 +02:00
e2395f3f0d
Apply clang format
Fix #128164: Race condition in ShaderCreateInfo::finalize()
This could deadlock in the case of recursive dependencies.
Fix #128164: Race condition in ShaderCreateInfo::finalize()
@Sergey I am curious if that's the best we can do?
5844f3b5b5
Cleanup: Metal: Make shader defines more readable
a9d1ddee9b
GPU: Remove old driver derivative sign workaround
0cc7319bd3
GPU: Add matrix constructor linter
fbb96f8b0f
GPU: Change to explicit matrix reshaping functions
2d64e0d6fb
Metal: Move decorator macro injection to glsl_preprocess