Together Compute ships early Rubin GPU kernels, unlocking wider MMA steps and new memory features
togethercompute · x · 2026-09-11
Together Compute published a blog detailing how it optimizes GEMM kernels for NVIDIA's Rubin GPUs: starting from its original B200 GEMMs and progressively adding Rubin-specific features such as wider MMA steps, more tensor and shared memory, B-side collectors, and early A release. The team stresses the kernels are still early, with LUT GEMMs, hardware-native megakernels, and new engine-level optimizations coming next, and credits NVIDIA for early access and support.
Related event: Together AI ports ThunderKittens to Vera Rubin, hits 22 PFLOPS(6 posts)→
More from Infra
- L3Harris says fine-tuned open-source models beat frontier AI in 48 hours at 95% lower cost — eliano · 2026-09-11
- Qwen3-TTS 1.7B hits 1.6x real-time voice cloning on CPU via llama.cpp — alexcovo_eth · 2026-09-11
- NVIDIA ships NVFP4-quantized Qwen3.8-27B, trending on Hugging Face — nvidia · 2026-09-11
- Qwen 3.8 125B-A6B runs 15.3% faster on Mac via speculative decoding on mlx.fast — TheMoonMidas · 2026-09-11
- Olam Labs CEO: only compute and data remain as bottlenecks to AGI — garrytan · 2026-09-11
- Open-sourced inference acceleration for structure-based models ships benchmarked and documented — AllThingsApx · 2026-09-11