Local [Superintelligence Supercomputing] Signed by @LisaSu 💻 🚀🚀🚀 Ryzen AI Max PRO 395 (Strix Halo) with ROCm * runs GPT-OSS locally * runs Battlefied 6 like a desktop * 16x Zen5 cores for builds @sama @gdb Sarah As promised please tag me if you run into any issues 🤙

🚀 New blog: Win on TCO: How AMD Instinct™ MI355X Achieves Cost-Competitive Distributed Inference Through SGLang with MoRI AMD Instinct™ MI355X beats B200 on TCO for DeepSeek-R1 disaggregated inference, with 5% lower cost than B200 TRT-LLM, and 1.25× higher throughput/GPU than B200 SGLang. Together with @AMD, we achieved competitive TCO through full-stack optimizations: 1. MoRI quantized all-to-all (FP4 dispatch FP8 combine): 2.56× bandwidth reduction 2. MoRI-IO KV cache backend: ~10% higher throughput than Mooncake 3. Two-Batch Overlap with SDMA: zero-compute-overhead async transfers 4. AITER GEMM FlyDSL FusedMoE: tuned kernels for TP & DP EP on MI355X 5. Specv2 MTP on ROCm: delivers 4% total token throughput and -3.6% TPOT 6. CPU streaming: 20% output throughput, -16% TPOT at 2,048 concurrency Results live on @SemiAnalysis_ InferenceX dashboard.

“Open is more than a license to us. It’s a philosophy.” That approach is showing up in the real world, with ROCm powering 8 of the top 10 AI labs. @AnushElangovan at AMD #AIDevDay.

Introducing AMD AI Playbooks, your shortcut to building AI on AMD. Step-by-step guides to help you run real AI workloads, customize them for your environment, and get up and running faster. New playbooks drop every month. Start building: amd.com/en/developer/resourc…

Build commands shared here reddit.com/r/ROCm/comments/1… much simpler than before.

CANN toolkit (ala cuda-toolkit, rocm) clocks in at 3.6GB of disk install. That's a lot for a single arch toolkit. Makes AMD's bloated ROCm looks like a lightweight.

@AnushElangovan Has this challenge concluded? If it's still ongoing, how many prizes remain?

SemiAnalysis: Engineering samples and low-volume production of $AMD's first rack-scale MI455X system will begin in H2 2026. Due to manufacturing delays, mass production ramp, and the first production tokens generated on the MI455X will not occur until Q2 2027

✅ Day 0 support of MiniMax-2.5 on AMD GPU 2 MI300X GPUs are all you need, instead of 4 Hopper GPUs, to run it in full context. uv pip install vllm --extra-index-url lnkd.in/gJdnn3kJ VLLM_ROCM_USE_AITER=1 vllm serve MiniMaxAI/MiniMax-M2.5 \ --tensor-parallel-size 2 \ --tool-call-parser minimax_m2 \ --reasoning-parser minimax_m2_append_think \ --enable-auto-tool-choice \ --trust-remote-code