Compatibility
vLLM Optimizer targets Python 3.11–3.12 and native Linux experiment hosts with NVIDIA GPUs. This page records dated evidence rather than promising that every vLLM/CUDA combination works.
The runtime extra targets stable vLLM 0.30.x (>=0.30,<0.31) and GuideLLM
>=0.7.3,<0.8. vLLM 0.30.0 is the latest stable release checked on 2026-10-04;
see its release notes.
Install the current checkout on Linux with pip install -e '.[runtime]'.
Historical evidence and captured requirement files below retain their original
versions.
Validation matrix
The required L40 and H100 native-Linux runs need dedicated hosts and are not available in this workspace. They remain explicitly unvalidated:
| Date | GPU | GPUs | Mode | Driver / CUDA | Python | vLLM / GuideLLM | Model | Outcome |
|---|---|---|---|---|---|---|---|---|
| 2026-09-01 | L40 | 1 | native Linux | — | — | — | — | Not run: host unavailable |
| 2026-09-01 | L40 | 2+ | tensor parallel | — | — | — | — | Not run: host unavailable |
| 2026-09-01 | H100 | 1 | native Linux | — | — | — | — | Not run: host unavailable |
| 2026-09-01 | H100 | 2+ | tensor parallel | — | — | — | — | Not run: host unavailable |
| 2026-09-01 | RTX 3080 | 1 | WSL2 attempt | 596.36 / CUDA 13.2 | 3.12.3 | 0.28.0 / 0.7.3 | facebook/opt-125m | Failed: vLLM reports UVA is not available |
| 2026-09-02 | RTX 3080 | 1 | WSL2 V0 smoke | 596.36 | 3.12.3 | 0.19.0 / 0.7.3 | local OPT-125M | Passed: two repeats/backend, 10/10 requests/backend, clean drains |
| 2026-09-13 | RTX 3080 | 1 | WSL2 showcase and baseline control | 616.64 / CUDA 13.0 | 3.12.3 | 0.28.0 / 0.7.3 installed | facebook/opt-125m | Completed: published vLLM Bench Serve evidence; showcase drift remains inconclusive |
| 2026-10-04 | RTX 3080 | 1 | WSL2 temperature smoke | 616.92 / CUDA 13.0 | 3.12.3 | 0.30.0 / 0.7.4 | local OPT-125M | Passed: default zero, explicit zero, nonzero startup warnings, 6/6 vLLM and 4/4 GuideLLM requests, JSON/CSV/HTML, clean shutdown |
The RTX 3080 rows are environment diagnostics, not supported native-Linux validation. The September 1 attempt produced no accepted benchmark result; the later September 13 runs did. See the published results and provenance. The V0 smoke verified streaming, final counters, request totals, reports, cleanup, and offline reclassification, but used a vLLM version outside the supported range.
The matrix still needs evidence for explicit ports, interruption cleanup, long generation, and at least one multi-GPU run on each supported native-Linux GPU.
Both GuideLLM 0.7.3 and vllm bench serve from vLLM 0.28.0 are the versions
used by the reproducibility procedure. Workload arguments are passed through,
so consult the documentation for the installed engine version when adopting
newer options.
The published WSL2 experiments used these environment workarounds:
The first disables a runner that required unavailable unified virtual addressing; the second avoids a sampler requiring a local CUDA compiler. Native Linux systems may not need either setting.
CUDA compatibility
vLLM Optimizer does not bundle CUDA and does not require one exact CUDA release. The working combination is determined by vLLM, its PyTorch wheel, the NVIDIA driver, the GPU architecture, and any locally compiled kernels. Therefore, vLLM Optimizer cannot promise support for every CUDA version.
Follow the vLLM GPU installation guide for the vLLM release you install. Also check NVIDIA's CUDA compatibility table: newer CUDA toolkits require a sufficiently recent driver, and some PTX or new features require a newer driver even within a compatible major family.
The py3-none-any wheel is one universal pure-Python artifact for Linux,
Windows, and macOS. It is not three OS-specific binaries. Native vLLM execution
is unsupported on Windows; configuration and stored-result inspection work.