← sources

FURIOSA IR snapshot

source: ir (ir:FURIOSA)published 2026-05-23 08:05ingested 2026-05-23 08:10

extract

Renegade 2026 Keynote EFFICIENT AI INFERENCE IS HERE Tensor contraction, not matmul The fundamental computation of modern day deep learning is tensor contraction, a higher dimensional generalization of matrix multiplication. However, most commercial deep learning accelerators today incorporate fixed-sized matmul instructions as primitives. RNGD breaks away from that, unlocking powerful performance and efficiency. Tensor mapping for max utilization This fundamental design choice streamlines programming, maximizing parallelism and data reuse, while providing flexibility and reconfigurability of compute and maximizes memory resources based on tensor shapes. Furiosa Compiler leverages this flexibility and reconfigurability of hardware to select the most optimized tactics, delivering powerful and efficient deep learning acceleration for all scales of deployment. Tensor Contraction Processor INFERENCE WITHOUT CONSTRAINTS Performance Efficiency Programmability SOFTWARE FOR LLM DEPLOYMENT BE IN THE KNOW Blog Experience RENEGADE Summit 2026 [Experience RENEGADE Summit 2026](/blog/experience-renegade-summit-2026) Furiosa SDK 2026.2: Boosting RNGD throughput and accelerating deployments [Furiosa SDK 2026.2: Boosting RNGD throughput and accelerating deployments](/blog/furiosa-sdk-2026-2) FuriosaAI establishes European flagship office in Portugal [FuriosaAI establishes European flagship office in Portugal](/blog/furiosaai-establishes-european-flagship-office-in-portugal)

retained text

No document body was retained for this record — only its metadata was stored.