forward fix PR 14245, restore build on ROCm 6.2

[jeffdaily]

PR #14245 broke compilation for ROCm 6.2.

__hip_fp8_e4m3 only exists starting in ROCm 6.3. PR #14245 added VLLM_DISPATCH_FP8_TYPES that on ROCm instantiates both OCP and FNUZ kernels. We need to replace the new __hip_fp8_e4m3 HW cvt with something reasonable that doesn't rely on the ROCm 6.3 feature. This allows compiling on ROCm 6.2 or newer.

[github-actions]

👋 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 fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

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 ready label to the PR or enable auto-merge.

🚀

[ProExpertProg]

Thanks for the quick fix!

[huydhn]

@jeffdaily Thank you for the fix! For the context, I'm experimenting with benchmarking vLLM v1 on an internal MI300X GPU (dashboard), and the host is on ROCm 6.2 and builds start to fail (we probably want to use OSS AMD runners to run this benchmark eventually).

[ProExpertProg]

I retried the failing CI job so let's hope this passes.

[tlrmchlsmth]

Thanks for the contribution and the explanation!