Fleet: A hierarchical task model for improving cache locality on multi-die GPUs
Modern GPUs adopt chiplet-based designs with multiple private cache hierarchies, but current programming models (CUDA/HIP) expose a flat execution hierarchy that cannot express chiplet-level locality…
Read the full articleYou might also wanna read
Instruction Pipeline Hazards in the Nvidia B200 GPU: An Empirical Analysis
hiraditya.github.io·4d ago
Reverse-engineering NVIDIA's cuda-checkpoint: Achieving 4x faster CUDA process restoration
blog.doubleword.ai·8d ago
A deep technical dive into CUDA kernel execution: from source code to GPU warps
fergusfinn.com·18d agoA deep technical dive into CUDA kernel execution: from source code to GPU warps
fergusfinn.com·18d ago
Roofline Model for Estimating Speculative Decoding Speedup in LLM Inference
modal.com·26d ago黙って通していた確認ツールを、ソフト拒否から allow ルールに起こす
antigravitylab.net·just now
Comments
Sign in to join the conversation.
No comments yet. Be the first.