Michał Zientkiewicz
Michał Zientkiewicz
Design: #129 (proposal from a comment) This work builds on #105 The resource is defined in terms of memory kind. `resource_view` (a glorifieid pointer with elaborate conversion logic) is defined...
The file `include/cuda/std/detail/__config` unconditionally disables multiple features, including RTTI and exceptions. If this header is pulled in, the subsequently included headers will have these features disabled.
Signed-off-by: Michał Zientkiewicz #### Why we need this PR? *Pick one, remove the rest* - It adds new feature needed for sample masking/selection/conditional execution #### What happened in this PR?...
Signed-off-by: Michał Zientkiewicz #### Why we need this PR? *Pick one, remove the rest* - It fixes a bug: per-sample external source don't adjust to dynamic batch size #### What...
Workaround for issue #6255 This change makes it possible to set a thread-local CUDA context and stream by explosing the following functions: `halide_cuda_set_current_context` and `halide_cuda_set_current_stream`. Executing kernels and buffer copies...
This is a proposal for the interface rework that would satisfy the requirements outlined in #660 ## The changes are: - add notion of `memory_kind` and `allocation_order` - add a...
Signed-off-by: Michal Zientkiewicz Co-authored-by: Joaquin Anton ## Category: **New feature** (*non-breaking change which adds functionality*) **Refactoring** (*Redesign of existing code that doesn't affect functionality*) ## Description: It adds a high-level...
Signed-off-by: Michal Zientkiewicz ## Category: ## Description: ## Additional information: ### Affected modules and functionalities: ### Key points relevant for the review: ### Tests: - [ ] Existing tests apply...
Signed-off-by: Michal Zientkiewicz ## Category: **Other** Documentation / run-time message ## Description: Ths PR makes the video loader print a warning about imminent change of a default value only when...
## Category: **Other** Performance tuning ## Description: The logical blocks seem to have been too large for resize and resulted in GPU underutilization. This PR adds two environment variables that...