Shaoting

I changed the codes in ```cpython-unix/Makefile``` and solved this problem. ``` # Always write out settings files. placeholder_archive_makefiles: $(RUN_BUILD) placeholder_archive makefiles include $(OUTDIR)/Makefile.$(HOST_PLATFORM).$(TARGET_TRIPLE).$(PYBUILD_PYTHON_MAJOR_VERSION) include $(OUTDIR)/versions/VERSION.* # Always write out expanded...

I went into the same problem when building inside a docker container with root permission. While my last comment solved the problem ```multiple target patterns. Stop.```, a new problem occurs...

> I've got the same issue when tried to launch your build scripts from my docker image without root (map host user(not root) uid:guid to image root). The issue was...

@ywang96 Thank you for assigning yourself to the review! After checking the Buildkite performance-benchmark pipeline, I noticed that it is stuck at the `Wait for container to be ready` step,...

> > @ywang96 Thank you for assigning yourself to the review! After checking the Buildkite performance-benchmark pipeline, I noticed that it is stuck at the `Wait for container to be...

> validate how LMCache behaves when used in conjunction with [vLLM](https://github.com/vllm-project/vllm) > * Trigger CI by a `ready` label & after the PR being approved At present, we have two...

> AFIK, this is already setup in the testing environments in CI. [@Shaoting-Feng](https://github.com/Shaoting-Feng) is that correct? I didn't install NIXL myself in the CI server. Right now the CI machine...

We want to add `NIXL` test after https://github.com/LMCache/LMCache/pull/895 is merged.

I think this happens because PyTorch enforces that `torch.cuda.set_device(device_id)` only accepts `device_id` in `[0, num_local_GPUs – 1]`, otherwise it throws “invalid device ordinal”. In multi-node tensor-parallel setups, LMCache (via the...

Hi, I'd like to take on this issue. I'll refactor the current datasets into `benchmark_dataset.py` and let `benchmark_throughput.py` and `benchmark_serving.py` import from it.