Skip to content

vulkan: skip zero size tensors in backend copies - #20233

Merged
0cc4m merged 1 commit into
masterfrom
0cc4m/vulkan-fix-cpy-zero-size
Mar 9, 2026
Merged

vulkan: skip zero size tensors in backend copies#20233
0cc4m merged 1 commit into
masterfrom
0cc4m/vulkan-fix-cpy-zero-size

Conversation

@0cc4m

@0cc4m 0cc4m commented Mar 8, 2026

Copy link
Copy Markdown
Contributor

I noticed validation errors in #19955

Validation Error: [ VUID-VkBufferCopy-size-01988 ] | MessageID = 0x636fcc49
vkCmdCopyBuffer(): pRegions[0].size is zero.
The Vulkan spec states: The size must be greater than 0 (https://docs.vulkan.org/spec/latest/chapters/copies.html#VUID-VkBufferCopy-size-01988)
Objects: 1
    [0] VkCommandBuffer 0x57d940f3cda0

I traced it to the backend async tensor copy function being called with zero-size tensors. I added checks to all backend interfaces to skip size 0 tensors.

@0cc4m
0cc4m requested a review from jeffbolznv March 8, 2026 11:12
This was referenced Mar 8, 2026
Closed
Merged
Closed
@github-actions github-actions Bot added Vulkan Issues specific to the Vulkan backend ggml changes relating to the ggml tensor library for machine learning labels Mar 8, 2026
@0cc4m
0cc4m merged commit b2f460b into master Mar 9, 2026
77 of 78 checks passed
@0cc4m
0cc4m deleted the 0cc4m/vulkan-fix-cpy-zero-size branch March 9, 2026 06:23
Seunghhon pushed a commit to Seunghhon/llama.cpp that referenced this pull request Apr 26, 2026
@0cc4m
vulkan: skip zero size tensors in backend copies (ggml-org#20233)
d8afe8f
my-other-github-account pushed a commit to my-other-github-account/llama.cpp that referenced this pull request May 15, 2026
@0cc4m
vulkan: skip zero size tensors in backend copies (ggml-org#20233)
a557d69
fukuro-kun pushed a commit to fukuro-kun/fukuro-llama-cpp-turboquant that referenced this pull request Jul 5, 2026
@0cc4m
vulkan: skip zero size tensors in backend copies (ggml-org#20233)
5bb5a42
MrLordCat referenced this pull request in MrLordCat/llama.cpp-rdna-lab Jul 16, 2026
@0cc4m
vulkan: skip zero size tensors in backend copies (#20233)
ca8a1bc
zommiommy pushed a commit to zommiommy/llama.cpp that referenced this pull request Aug 18, 2026
@0cc4m @zommiommy
vulkan: skip zero size tensors in backend copies (ggml-org#20233)
2178184
frostyautumnleaf pushed a commit to frostyautumnleaf/llama.cpp that referenced this pull request Oct 5, 2026
@0cc4m
vulkan: skip zero size tensors in backend copies (ggml-org#20233)
85e1fd5
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jeffbolznv jeffbolznv jeffbolznv approved these changes

Assignees

No one assigned

Labels

ggml changes relating to the ggml tensor library for machine learning Vulkan Issues specific to the Vulkan backend

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@0cc4m @jeffbolznv