LLM Reddit Apr 24, 2026 2 min read
LocalLLaMA upvoted this because it felt like real plumbing, not another benchmark screenshot. The excitement was about DeepSeek open-sourcing faster expert-parallel communication and reusable GPU kernels.
#gpu-kernels
RSS FeedAI X/Twitter Mar 20, 2026 2 min read
NVIDIA announced SOL-ExecBench on March 20, 2026, a benchmark for real-world GPU kernels that scores optimized CUDA and PyTorch code against Speed-of-Light hardware bounds on NVIDIA B200 systems. The release packages 235 kernel optimization problems drawn from 124 AI models across BF16, FP8, and NVFP4 workloads.