DEPLOY

Chip leaderboard

Highest INT8 dense throughput

AMD Instinct MI350 series leads at 4,600 TOPS. 29 chips ranked.

Raw compute at INT8: the standard precision for serving inference. Dense throughput, not sparse.

#ChipValueInputs
1AMD Instinct MI350 series
AMD
4,600 TOPS
2NVIDIA B200 (Blackwell)
NVIDIA
4,500 TOPS
3Tenstorrent Blackhole
Tenstorrent
2,980 TOPS
4AMD Instinct MI300X
AMD
2,614 TOPS
5AMD Instinct MI325X
AMD
2,614 TOPS
6NVIDIA Jetson AGX Thor
NVIDIA
2,070 TOPS
7Biren BR100
Biren Technology
2,048 TOPS
8NVIDIA DRIVE Thor
NVIDIA
2,000 TOPS
9NVIDIA H100 Tensor Core GPU
NVIDIA
1,979 TOPS
10NVIDIA H200 Tensor Core GPU
NVIDIA
1,979 TOPS
11Google TPU v6 (Trillium)
Google
1,836 TOPS
12Huawei Ascend 910C
Huawei
1,600 TOPS
13AWS Trainium 2
Amazon
1,300 TOPS
14Google TPU v5p
Google
918 TOPS
15Huawei Ascend 910B
Huawei
800 TOPS
16Groq LPU (Language Processing Unit)
Groq
750 TOPS
17NVIDIA L40S
NVIDIA
733 TOPS
18Meta MTIA v2
Meta
708 TOPS
19NVIDIA A100 Tensor Core GPU
NVIDIA
624 TOPS
20Tenstorrent Wormhole
Tenstorrent
590 TOPS
21Google TPU v5e
Google
393 TOPS
22AMD Instinct MI250X
AMD
383 TOPS
23Tesla HW4 (FSD Computer 2)
Tesla
300 TOPS
24Google TPU v4
Google
275 TOPS
25NVIDIA Jetson AGX Orin
NVIDIA
275 TOPS
26Cambricon MLU370-X8
Cambricon Technologies
256 TOPS
27NVIDIA A10 Tensor Core GPU
NVIDIA
250 TOPS
28NVIDIA T4 Tensor Core GPU
NVIDIA
130 TOPS
29Mobileye EyeQ6
Mobileye
34 TOPS

29 chips ranked. 31 more are in our catalog but don't publish the numbers needed for this ranking.

Other rankings: Best perf-per-watt (TFLOPS/W) · Most HBM memory per chip · Highest memory bandwidth · Highest FP8 dense throughput · Highest FP16 dense throughput · Cheapest per TFLOP at launch (list price) · Most transistors per die

See also: every chip on record · head-to-head comparisons.