Tag
This article discusses how next-generation AI chip startups are attacking the data movement aspect from different angles in an attempt to challenge Nvidia's dominance, citing a list from July 2026.
Proposes RED-PIM, an algorithm-architecture co-design that reduces inter-bank data movement from O(N^2) to O(N) and shrinks attention matrices, achieving significant inference time reductions (16% to 99.99%) for transformer models.
A detailed analysis of how NVIDIA GPU programming evolved from Volta to Blackwell, highlighting the shift from synchronous thread models to asynchronous dataflow and the challenges of feeding Tensor Cores. The article discusses new hardware features like TMA, TMEM, and tcgen05 MMA, and shows how modern kernels like FlashAttention-3 and FlashMLA exploit these changes for higher utilization.
The author discusses building a small VLM for desktop GUI automation to move data between apps without APIs, expressing interest in non-coding autonomous use cases for local models.