Demystifying the Nvidia Ampere Architecture through Microbenchmarking and Instruction-level Analysis · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2208.11174/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTDemystifying the Nvidia Ampere Architecture through Microbenchmarking and Instruction-level Analysis — mentioned by dragontamerhttps://news.ycombinator.com/item?id=34808704urn:hackerpapers:mention:2208.11174:34808704Wed, 15 Feb 2023 18:51:01 GMT<p>Those who want &quot;excitement&quot; in instruction sets should study GPUs, POWER10, ARMv8, AVX512, and ARM&#x27;s SVE these days. GPU instructions, such as bpermute, permute, warp-vote, and other such concepts are the biggest &quot;conceptual&quot; changes to compute I&#x27;ve ever seen. DirectX12, Vulkan, CUDA, HIP, it doesn&#x27;t really matter what you study per se, the GPU community knows brute force performance and is advancing it. From this perspective, your study guides are the NVidia PTX guide (CUDA&#x27;s assembly language), and AMD&#x27;s RDNA2 ISA. Also keep in mind the NVidia PTX wgmma instruction, which is key to 16-bit…</p><p><a href="https://hackerpaper.kaipfister.de/p/2208.11174/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.AR