Hedra speeds up Krea 3 diffusion transformer 6.9x on AMD Instinct MI355X
henloitsjoyce · x · 2026-11-04
Hedra Labs used its autonomous research system to optimize Krea AI's Krea 3 diffusion transformer on AMD Instinct MI355X, cutting generation time from 22.9s to 3.3s (6.9x faster).
Key stats:
- 2.8x faster MXFP6 GEMM vs BF16
- GPU transfer wait time cut from 18% to 2%
- Quality retained at 98.3% of BF16 (SSIM)
More from Infra
- Big tech has contracted over 10 GW of nuclear power, but contracts aren't generation — ingliguori · 2026-11-04
- Fully Local AI Short Film Rendered on a Mac, Workflow Open-Sourced as Phosphene — Opening-Ad5541 · 2026-11-04
- Anthropic's 15-person chip team uses Claude to design its own AI chips — kimmonismus · 2026-11-04
- Anthropic's 15-person chip team uses Claude to design its own AI chips ahead of $1.5T IPO — kimmonismus · 2026-11-04
- AAOI signs $194M InP wafer supply deal with Yunnan Germanium at ~$1,077 per wafer — zephyr_z9 · 2026-11-04
- Google Cloud taps Tetrate to extend Gemini platform AI routing to multi-cloud and on-prem — IanAndrewsDC · 2026-11-04