Analyzing Deepseek’s “undefined” NVIDIA PTX optimizations (with benchmarks!) Blog 08/06/2026 · 0 Comment Analyzing Deepseek's "undefined" NVIDIA PTX optimizations (with benchmarks!)Nvidia CUDA in 100 SecondsNVIDIA CUDA’s moat is collapsing? The significance of DeepSeek engineers’ CUDA bypass approach[Video Special] DeepSeek-V4 Architecture and KV Cache OptimizationInsights from DeepSeek & Nvidia GPU Innovation via PTX (Is Nvidia Stock Doomed?)Breaking the CUDA Moat: How DeepSeek TileLang is Ending NVIDIA Lock-InDeepSeek's GPU optimization tricks | Lex Fridman PodcastLLM throughput benchmark on 13 GPUs and 10 modelsDeepSeek V4 Code Review: Why This AI is Nvidia's Worst NightmareNVIDIA Just Solved Quantum Computing's Biggest Problem12
![[Video Special] DeepSeek-V4 Architecture and KV Cache Optimization](https://i.ytimg.com/vi/rvLL8tQl4hg/hq720.jpg?sqp=-oaymwFFCNAFEJQDSFryq4qpAzcIARUAAIhCGAHYAQHiAQoIGBACGAY4AUAB8AEB-AH-CYAC0AWKAgwIABABGGQgZChkMA-4AvMY&rs=AOn4CLDD9BgcdpCGOCCvYzDCLWZEUjXJeQ)
