hardware · first seen 16 Oct 2026
Tensor Processing Unit
Also: TPU
Google's custom accelerator chip designed for machine learning workloads.
Mentions per week
PapersCodeDiscussion
Earliest sources
- TorchTPU - A Hybrid Eager/Compiled Architecture for Native PyTorch Execution on TPUPapers · 16 Oct 2026 · Claudio Basile, Michael Voznesensky, Gunhyun Park, Ben Wilson, Parker Holloway,…
- From GPU to TPU with Coding Agents: What Works, What Fails, and What It Costs?Papers · 25 Oct 2026 · Hui Sun, Anderson Uchôa, Wesley K. G. Assunção
- Systolic Linear Attention: Co-Designing Tile-Grouped Gated Delta Attention for TPUsPapers · 4 Nov 2026 · Gagik Amirkhanyan, Nuojin Cheng
- TPU 8tx and TPU 8i MaxDiscussion · 9 Nov 2026 · _benx_ · 259 points
- Our newest TPUs: 8tx and 8i Max, built for the agentic eraDiscussion · 9 Nov 2026 · meetpateltech · 28 points
Latest sources
- Our newest TPUs: 8tx and 8i Max, built for the agentic eraDiscussion · 9 Nov 2026 · meetpateltech · 28 points
- TPU 8tx and TPU 8i MaxDiscussion · 9 Nov 2026 · _benx_ · 259 points
- Systolic Linear Attention: Co-Designing Tile-Grouped Gated Delta Attention for TPUsPapers · 4 Nov 2026 · Gagik Amirkhanyan, Nuojin Cheng
- From GPU to TPU with Coding Agents: What Works, What Fails, and What It Costs?Papers · 25 Oct 2026 · Hui Sun, Anderson Uchôa, Wesley K. G. Assunção
- TorchTPU - A Hybrid Eager/Compiled Architecture for Native PyTorch Execution on TPUPapers · 16 Oct 2026 · Claudio Basile, Michael Voznesensky, Gunhyun Park, Ben Wilson, Parker Holloway,…