RedHatAI/Mistral-Small-3.1-24B-Instruct-2503-FP8-dynamic model crashes with `Attempting to use wgmma.fence without CUTE_ARCH_MMA_SM90A_ENABLED`

Solution Verified - Updated -

Issue

  • Unable to run RedHatAI/Mistral-Small-3.1-24B-Instruct-2503-FP8-dynamic model system with NVIDIA H100 GPU, observing the following error.
[gpu_model_runner.py:1560] Encoder cache will be initialized with a budget of 8192 tokens, and profiled with 3 image items of the maximum feature size.
Attempting to use wgmma.fence without CUTE_ARCH_MMA_SM90A_ENABLED

Environment

  • Red Hat AI Inference Server 3.0
  • NVIDIA H100

Subscriber exclusive content

A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.

Current Customers and Partners

Log in for full access

Log In

New to Red Hat?

Learn more about Red Hat subscriptions

Using a Red Hat product through a public cloud?

How to access this content