TPU

Creator

Creator

Seonglae Cho

Created

Created

2020 May 17 9:25

Editor

Editor

Seonglae Cho

Edited

Edited

2025 Jul 6 18:29

Refs

Refs

An

ASIC developed by Google, optimized for large-scale matrix multiplication processing and energy efficiency.

Systolic Array + Pipelining to minimize memory access

Ahead-of-Time compilation (XLA) to predetermine memory access patterns, utilizing scratchpads instead of caches

TPU Versions

TPU v6 Trillium

https://jax-ml.github.io/scaling-book/

https://jax-ml.github.io/scaling-book/tpus/

https://jax-ml.github.io/scaling-book/tpus/

https://jax-ml.github.io/scaling-book/tpus/

Compute primitives and memory layouts for different hardware backends. Image by Chen et al., 2018.

https://jax-ml.github.io/scaling-book/tpus/

I've been working with TPUs a lot recently and it's fun to see how they had such different design philosophies compared to GPUs.

https://henryhmko.github.io/posts/tpu/tpu.html

https://github.com/kimbochen/md-blogs/tree/main/tpuv4_v5e

Apple says its AI models were trained on Google's custom chips

Apple is using chips designed by Google in building its advanced AI models, according to a paper published on Monday.

https://www.cnbc.com/2024/07/29/apple-says-its-ai-models-were-trained-on-googles-custom-chips-.html

Apple says its AI models were trained on Google's custom chips

Backlinks

GCP Component jax.profiler AI Compiler Optimization AI Framework

Recommendations

//////