I benchmarked LLM performance on AMD Radeon AI PRO R9700 with Ollama, comparing ROCm 7.1 vs 6.4 across 8 models (Mistral, Llama, Qwen, GPT-OSS, DeepSeek, and more). Result: visible gains in prompt throughput (+87% avg) and faster responses (+11% avg). Full tables, setup, and notes in the post.

https://meefik.dev/2026/05/31/llms-performance-rocm7/

#AMD #ROCm #Ollama #LLM #AI

AMD ROCm 7.2.4 Released With Performance & Stability Fixes

AMD ROCm 7.2.4 Released With Performance & Stability Fixes #rocm #amd #videocard #ai

https://kbin.melroy.org/m/technology@lemmy.zip/t/1719334

AMD ROCm 7.2.4 Released With Performance & Stability Fixes

AMD ROCm 7.2.4 Released With Performance & Stability Fixes #rocm #amd #videocard #ai

https://kbin.melroy.org/m/[email protected]p/t/1719329

Log in - MelroyBin

Creator of Mbin: content aggregator and micro-blogging platform for the fediverse

Anyone had success running Distrobox (Debian/Ubuntu vm) under Alpine Linux to get Blender with Cycles/HIP support working on an Alpine host? Something I gotta try...

#Linux #Alpine #AlpineLinux #Blender #ROCm #AMD #HIP #Distrobox #Podman #Docker

We’re excited to welcome another student contributor to the FreeBSD community.

Please join us in welcoming Sourojeet Adhikari, who recently shared that he’ll be joining the FreeBSD Foundation as an intern.

As part of his work, Sourojeet is currently exploring “Porting ROCm from Linux to FreeBSD”, an ambitious and highly technical project focused on GPU computing support.

https://racha.ca/myPosts/ROCmFreeBSD_pt1.html

#FreeBSD #OpenSource #GSoC #ROCm

第912回 26.04 LTSとNVIDIA/AMDのGPGPU事情 | gihyo.jp

今回は執筆時点でのUbuntu 26.04 LTSのNVIDIA/AMDのGPGPU対応状況を報告します。

gihyo.jp

RT @TeksEdge: 🚀 Neue MTP-Unterstützung für Strix Halo veröffentlicht!

mehr auf Arint.info

#AI #AMD #MTP #Qwen #ROCm #StrixHalo #arint_info

https://x.com/TeksEdge/status/2058728175388262761#m

Arint - SEO+KI (@[email protected])

<p>RT @TeksEdge: 🚀 Neue MTP-Unterstützung für Strix Halo veröffentlicht!</p> <p><a href="https://arint.info/@Arint/116634638485149145">mehr</a> auf <a href="https://arint.info/">Arint.info</a></p> <p>#AI #AMD #MTP #Qwen #ROCm #StrixHalo #arint_info</p> <p><a href="https://x.com/TeksEdge/status/2058728175388262761#m">https://x.com/TeksEdge/status/2058728175388262761#m</a></p>

Mastodon Glitch Edition

Pretraining large language models with MXFP4 on Native FP4 Hardware

#ROCm #LLM #MXFP4 #FP4

https://hgpu.org/?p=30811

Pretraining large language models with MXFP4 on Native FP4 Hardware

Why does full-pipeline FP4 training of large language models often diverge, even when forward activations and activation gradients remain stable? We address this question through a controlled study…

hgpu.org

40% Cost Gap Developer Cloud Vs On-Prem Instinct

A 40% cost advantage and instant ROCm performance show why AMD's developer cloud beats on‑prem Instinct setups. See the data and learn how to spin up a VM in minutes.

https://cloudplatform.digital/40-percent-cost-gap-developer-cloud-vs-on-prem-ins/

#developercloud #developercloudamd #instinct #rocm #clouddevelopertools

40% Cost Gap Developer Cloud Vs On-Prem Instinct

Discover how the AMD developer cloud cuts costs by 40% compared to on‑prem Instinct deployments, with faster provisioning, ROCm performance gains, and flexible

Cloud Platform

https://www.youtube.com/watch?v=32cjdHVoSRo&t=1s

builds a fully open source stack with fedora after some difficulty #cluster #rocm #rdma #rhel gateway #operator error #userland

Three months wrong about why my 4-node AMD cluster was slow

YouTube