GPU engineer: stop saying CUDA Core/Tensor Core, they're marketing terms obscuring SM microarchitecture
yacineMTB · x · 2026-09-07
vikhyatk argues that the first step to understanding SM microarchitecture is dropping terms like CUDA Core and Tensor Core—they're marketing labels that convey only orders of magnitude of performance, not real architectural detail. Shared with an explainer video.
More from Infra
- This Week's AI Must-Reads: OpenAI's Research Acceleration Report and Broadcom's $16.7B AI Chip Quarter — VibeMarketer_ · 2026-09-07
- On-device Android agent with Gemma 4 E2B hits 2.6 tok/s live vs 11 tok/s on replay — HowDevelop · 2026-09-07
- Dev wrote Marlin-style FP4/FP8 kernels for RTX 3090 — NVIDIA declined to upstream them — QuixiAI · 2026-09-07
- Prediction: OpenAI's agent compute spend may exceed total employee wages by early next year — haider1 · 2026-09-07
- Huawei's Mate XT 2 debuts Tau Scaling Law Kirin chip and ships with HarmonyOS 7 — pstAsiatech · 2026-09-07
- Huawei unveils Kirin 9050 Pro based on its Tau Scaling Law, claims 42% performance jump — pstAsiatech · 2026-09-07