DEVICE/API: Add tests #814
Open
+1,783
−1,079
Conversation
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
|
👋 Hi michal-shalev! Thank you for contributing to ai-dynamo/nixl. Your PR reviewers will review your contribution then trigger the CI to test your changes. 🚀 |
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
5dd69d7 to
a06b812
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
a06b812 to
6cd8c5e
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
6cd8c5e to
5e07f27
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
5d203f0 to
9124a60
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
9124a60 to
c9d565b
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
c9d565b to
cc0dc83
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
cc0dc83 to
cccf367
Compare
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
cccf367 to
e162030
Compare
Signed-off-by: Michal Shalev <[email protected]>
michal-shalev
force-pushed
the
add-device-api-tests
branch
from
e162030 to
55df3ae
Compare
ofirfarjun7
reviewed
Dec 1, 2025
| } | ||
|
|
||
| void | ||
| copyToHost(T *host_data, size_t count) const { |
Contributor
There was a problem hiding this comment.
Maybe use common copy function and pass cudaMemcpyHostToDevice/cudaMemcpyDeviceToHost to it
| if constexpr (level == nixl_gpu_level_t::THREAD) { | ||
| return 1; | ||
| } else if constexpr (level == nixl_gpu_level_t::WARP) { | ||
| return 32; |
| req_ptr); | ||
| break; | ||
|
|
||
| case NixlDeviceOperation::FULL_WRITE: |
| req_ptr); | ||
| break; | ||
|
|
||
| case NixlDeviceOperation::SIGNAL_READ: { |
Contributor
There was a problem hiding this comment.
maybe SIGNAL_POLL/SIGANL_WAIT?
| } while (value != params.signalRead.expectedValue); | ||
|
|
||
| if (params.signalRead.resultPtr != nullptr) { | ||
| *params.signalRead.resultPtr = value; |
Contributor
There was a problem hiding this comment.
Why we need resultPtr if we know this function exists only if value == params.signalRead.expectedValue?
| } | ||
| nixlGpuWriteSignal<level>(params.signalWrite.signalAddr, | ||
| params.signalWrite.value); | ||
| return NIXL_SUCCESS; |
Contributor
There was a problem hiding this comment.
Maybe just set status and break, use common code in end of func to return status.
Do the same for previous cases that returns status.
|
|
||
| namespace { | ||
|
|
||
| constexpr size_t maxThreadsPerBlock = 256; |
| return NIXL_ERR_BACKEND; | ||
| } | ||
|
|
||
| return NIXL_SUCCESS; |
Contributor
There was a problem hiding this comment.
Suggested change
| return NIXL_SUCCESS; | |
| const cudaError_t launch_error = cudaGetLastError(); | |
| if (launch_error != cudaSuccess) { | |
| std::cerr << "CUDA kernel launch error: " << cudaGetErrorString(launch_error) << "\n"; | |
| return NIXL_ERR_BACKEND; | |
| } | |
| const cudaError_t sync_error = cudaDeviceSynchronize(); | |
| if (sync_error != cudaSuccess) { | |
| std::cerr << "CUDA synchronization error: " << cudaGetErrorString(sync_error) << "\n"; | |
| return NIXL_ERR_BACKEND; | |
| } |
|
|
||
| createXferRequest(data.srcBuffers, data.dstBuffers, mem_type, | ||
| data.xferReq, data.gpuReqHandle); | ||
| } |
Contributor
There was a problem hiding this comment.
Looks like common code... other tests uses same code for creating gpuReqHandle...
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.
What?
Adds a comprehensive suite of tests for the Device API, covering Single Write, Partial Write, Full Write, and Signal operations.
Why?
To validate the functionality and correctness of the Device API implementation.
How?
test/gtest/device_api/common/for shared kernels and utilities (device_kernels,mem_buffer, etc.).test/gtest/device_api/tests/for each operation type.