https://www.servethehome.com/amd-instinct-mi300x-architecture-at-hot-chips-2024/ Server and Workstation Reviews Tue, 27 Aug 2024 18:25:14 +0000 hourly 1 https://wordpress.org/?v=6.7.1 https://www.servethehome.com/amd-instinct-mi300x-architecture-at-hot-chips-2024/#comment-584682 Tue, 27 Aug 2024 18:25:14 +0000 https://www.servethehome.com/?p=79985#comment-584682 OK, now put it on a PCIe card so us mere mortals can make use of it!

]]> https://www.servethehome.com/amd-instinct-mi300x-architecture-at-hot-chips-2024/#comment-584647 Tue, 27 Aug 2024 02:04:15 +0000 https://www.servethehome.com/?p=79985#comment-584647 Is it known what batch size AMD is using for their MI300X performance claims? Last time AMD claimed these numbers, Nvidia posted this blog post basically showing how AMD cherrypicked useless numbers (batch size 1) that are irrelevant to real world application
https://developer.nvidia.com/blog/achieving-top-inference-performance-with-the-nvidia-h100-tensor-core-gpu-and-nvidia-tensorrt-llm/

]]>