[easy] Print number of needed GPUs in skip message #17594
Conversation
tests/compile/test_basic_correctness requires a specific number of GPUs to run. I changed the skip message to print it. Separately I'm looking into if there's another way to run this test, because I have an 8 GPU machine and I would like to run these without setting CUDA_VISIBLE_DEVICES all the time. Signed-off-by: rzou <[email protected]>
|
👋 Hi! Thank you for contributing to the vLLM project. 💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels. Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging. To run CI, PR reviewers can either: Add 🚀 |
Signed-off-by: rzou <[email protected]> Signed-off-by: Mu Huai <[email protected]>
Signed-off-by: rzou <[email protected]>
Signed-off-by: rzou <[email protected]> Signed-off-by: Yuqi Zhang <[email protected]>
2 participants
tests/compile/test_basic_correctness requires a specific number of GPUs to run. I changed the skip message to print it.
Separately I'm looking into if there's another way to run this test, because I have an 8 GPU machine and I would like to run these without setting CUDA_VISIBLE_DEVICES all the time.