fix: add missing CUDA error check + correct Vulkan enum type mismatch - #414
Open
Scottcjn wants to merge 1 commit into
Open
fix: add missing CUDA error check + correct Vulkan enum type mismatch#414Scottcjn wants to merge 1 commit into
Scottcjn wants to merge 1 commit into
Conversation
1. simple.c: Wrap cuDeviceGetName() with checkCudaErrors() to catch failures instead of silently using uninitialized device name. Fixes NVIDIA#413 2. simpleVulkan/main.cpp: Replace VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_* with VK_EXTERNAL_MEMORY_HANDLE_TYPE_* when checking memory handle type. The parameter is VkExternalMemoryHandleTypeFlagBits, not semaphore flags. Works by coincidence (same bit values) but is incorrect per the Vulkan spec. Fixes NVIDIA#409
Scottcjn
force-pushed
the
fix/cuda-error-check-and-vulkan-enum
branch
from
May 15, 2026 16:51
9732801 to
d474bc4
Compare
Author
|
Rebased against current upstream master. Ready for re-review. |
Author
|
Checking in on this. The added CUDA error check and Vulkan external-memory fix are small and self-contained. Happy to adjust if a maintainer has a preferred style. |
Collaborator
|
Hi @Scottcjn we are in the process of reworking the CUDA Samples repo, so we'll take a look at this in due time, thanks. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Two small fixes.
1. Missing error check on
cuDeviceGetName(#413)File:
cpp/7_libNVVM/simple/simple.c:72cuDeviceGetName()is called withoutcheckCudaErrors(), so on failure the sample prints an uninitialisednamebuffer. Every other driver API call incudaDeviceInit()is wrapped, so this just brings it in line.2. Wrong Vulkan enum family in external memory import (#409)
File:
cpp/5_Domain_Specific/simpleVulkan/main.cpp:331-337importCudaExternalMemory()takes aVkExternalMemoryHandleTypeFlagBits handleType, but tests it againstVK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_*constants, which belong toVkExternalSemaphoreHandleTypeFlagBits. Replaced with the matchingVK_EXTERNAL_MEMORY_HANDLE_TYPE_*constants.This is a type correctness fix with no behavioural change, because the two enums use the same bit values:
OPAQUE_FD_BITOPAQUE_WIN32_BITOPAQUE_WIN32_KMT_BIT(An earlier version of this description said
VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_WIN32_BITwas 0x00000001. That was wrong, it is 0x00000002. The values above are fromvulkan_core.hin KhronosGroup/Vulkan-Headers.)The other two
VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_*sites in this file, at lines 369 and 379, are insideimportCudaExternalSemaphore(), whose parameter really isVkExternalSemaphoreHandleTypeFlagBits. Those are correct and are deliberately left alone.Fixes #413, Fixes #409