[Kernel] support non-zero cuda devices in punica kernels

[jeejeelee]

FILL IN THE PR DESCRIPTION HERE

• avoid punica kernel executeed on non-zero GPU resulted in RuntimeError: CUDA error: an illegal memory access was encountered.
• Improve test_punica.py

FIX #xxxx (link existing issues this PR will resolve)

BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE

---

PR Checklist (Click to Expand)

Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.

PR Title and Classification

Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:

• [Bugfix] for bug fixes.
• [CI/Build] for build or continuous integration improvements.
• [Doc] for documentation fixes and improvements.
• [Model] for adding a new model or improving an existing model. Model name should appear in the title.
• [Frontend] For changes on the vLLM frontend (e.g., OpenAI API server, LLM class, etc.)
• [Kernel] for changes affecting CUDA kernels or other compute kernels.
• [Core] for changes in the core vLLM logic (e.g., LLMEngine, AsyncLLMEngine, Scheduler, etc.)
• [Hardware][Vendor] for hardware-specific changes. Vendor name should appear in the prefix (e.g., [Hardware][AMD]).
• [Misc] for PRs that do not fit the above categories. Please use this sparingly.

Note: If the PR spans more than one category, please include all relevant prefixes.

Code Quality

The PR need to meet the following code quality standards:

• We adhere to Google Python style guide and Google C++ style guide.
• Pass all linter checks. Please use format.sh to format your code.
• The code need to be well-documented to ensure future contributors can easily understand the code.
• Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.
• Please add documentation to docs/source/ if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.

Notes for Large Changes

Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with rfc-required and might not go through the PR.

What to Expect for the Reviews

The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:

• After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.
• After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.
• After the review, the reviewer will put an action-required label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.
• Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion.

Thank You

Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!

[jeejeelee]

after changing the aforementioned code, test_punica's result:

6956 passed, 3996 skipped in 5907.20s (1:38:27)...

[Yard1]

LGTM, thanks!

[simon-mo]

Thanks for the fix! Please fix formatting.

[yyccli]

Hi, i'm not sure whether this problem came back in some cases after introducing full tp in lora(see #3524 )
when i run func test_linear_parallel in test_layers.py and got the following errors:

could i trouble you take some time to look at this?
cc @Yard1 @FurtherAI

[yyccli]

more info after appending -s flag in pytest

FAILED[E ProcessGroupNCCL.cpp:1182] [Rank 0] NCCL watchdog thread terminated with exception: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Exception raised from c10_cuda_check_implementation at ../c10/cuda/CUDAException.cpp:44 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f2ae3ac3d87 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10.so)
frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::string const&) + 0x64 (0x7f2ae3a7475f in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10.so)
frame #2: c10::cuda::c10_cuda_check_implementation(int, char const*, char const*, int, bool) + 0x118 (0x7f2ae3b948a8 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10_cuda.so)
frame #3: c10d::ProcessGroupNCCL::WorkNCCL::finishedGPUExecutionInternal() const + 0x6c (0x7f2a995d23ac in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #4: c10d::ProcessGroupNCCL::WorkNCCL::isCompleted() + 0x58 (0x7f2a995d64c8 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #5: c10d::ProcessGroupNCCL::workCleanupLoop() + 0x15a (0x7f2a995d9bfa in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #6: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x119 (0x7f2a995da839 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #7: <unknown function> + 0xdc253 (0x7f2ae32b0253 in /usr/lib/x86_64-linux-gnu/libstdc++.so.6)
frame #8: <unknown function> + 0x94ac3 (0x7f2ae53fdac3 in /usr/lib/x86_64-linux-gnu/libc.so.6)
frame #9: clone + 0x44 (0x7f2ae548ebf4 in /usr/lib/x86_64-linux-gnu/libc.so.6)

terminate called after throwing an instance of 'c10::DistBackendError'
  what():  [Rank 0] NCCL watchdog thread terminated with exception: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Exception raised from c10_cuda_check_implementation at ../c10/cuda/CUDAException.cpp:44 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f2ae3ac3d87 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10.so)
frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::string const&) + 0x64 (0x7f2ae3a7475f in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10.so)
frame #2: c10::cuda::c10_cuda_check_implementation(int, char const*, char const*, int, bool) + 0x118 (0x7f2ae3b948a8 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10_cuda.so)
frame #3: c10d::ProcessGroupNCCL::WorkNCCL::finishedGPUExecutionInternal() const + 0x6c (0x7f2a995d23ac in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #4: c10d::ProcessGroupNCCL::WorkNCCL::isCompleted() + 0x58 (0x7f2a995d64c8 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #5: c10d::ProcessGroupNCCL::workCleanupLoop() + 0x15a (0x7f2a995d9bfa in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #6: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x119 (0x7f2a995da839 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #7: <unknown function> + 0xdc253 (0x7f2ae32b0253 in /usr/lib/x86_64-linux-gnu/libstdc++.so.6)
frame #8: <unknown function> + 0x94ac3 (0x7f2ae53fdac3 in /usr/lib/x86_64-linux-gnu/libc.so.6)
frame #9: clone + 0x44 (0x7f2ae548ebf4 in /usr/lib/x86_64-linux-gnu/libc.so.6)

Exception raised from ncclCommWatchdog at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1186 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f2ae3ac3d87 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libc10.so)
frame #1: <unknown function> + 0xdf6b11 (0x7f2a99330b11 in /workspace/vllm/.root_vllm/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
frame #2: <unknown function> + 0xdc253 (0x7f2ae32b0253 in /usr/lib/x86_64-linux-gnu/libstdc++.so.6)
frame #3: <unknown function> + 0x94ac3 (0x7f2ae53fdac3 in /usr/lib/x86_64-linux-gnu/libc.so.6)
frame #4: clone + 0x44 (0x7f2ae548ebf4 in /usr/lib/x86_64-linux-gnu/libc.so.6)

Fatal Python error: Aborted

[jeejeelee]

@yyccli Did you test using only one GPU?

To my understanding, test_layers.py::test_linear_parallel only passes when using a single gpu, has nothing to do with #3524

[yyccli]

single gpu is ok, but ... if i specify two gpus, the test will just do tests on both gpu individually? 😰

[jeejeelee]

single gpu is ok, but ... if i specify two gpus, the test will just do tests on both gpu individually? 😰

yes