En vivo
AI hardware

Google TPU v6e Trillium

FP16 throughput
9.2×10¹⁴ FLOP/s
Memory
29.8 GB
TDP
380 W

Google TPU v6e Trillium is an AI accelerator designed by Google (TPU), released in May 2024.

It delivers 9.2×10¹⁴ FLOP/s of tensor FP16/BF16 throughput, 9.2×10¹⁴ FLOP/s at FP8. It carries 29.8 GB of memory at 1.5 TB/s of bandwidth. Its thermal design power is 380 W.

Full record
Manufacturer
Google
Type
TPU
Release date
May 14, 2024
Tensor FP16/BF16
9.2×10¹⁴ FLOP/s
FP8
9.2×10¹⁴ FLOP/s
INT8
1.8×10¹⁵ OP/s
Memory
29.8 GB
Memory bandwidth
1.5 TB
TDP
380 W
More from Google
SourceEpoch AI, 'Machine Learning Hardware'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/machine-learning-hardware. Licensed under CC BY 4.0.
← All ai hardware