Chip comparison
Google TPU v4 vs NVIDIA V100 Tensor Core GPU
Side-by-side specs
Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.
| Field | Google TPU v4 | NVIDIA V100 Tensor Core GPU |
|---|---|---|
| Process node | TSMC N7 | TSMC 12FFN |
| Transistors (B) | — | 21.1 |
| Die size | — | 815 mm² |
| TDP | 192 W | 300 W |
| Memory | 32 GB HBM2 | 32 GB HBM2 |
| Memory bandwidth | 1,200 GB/s | 900 GB/s |
| FP16 (dense) | — | 125 TFLOPS |
| BF16 (dense) | 275 TFLOPS | — |
| INT8 (dense) | 275 TOPS | — |
| Form factor | OAM (per-chip in v4 pod) | SXM2 / PCIe |
| Announced | 2021-05-18 | 2017-05-10 |
| Released | 2022-05-01 | 2017-12-01 |
Robots running Google TPU v4
No robots publicly running it yet.
Robots running NVIDIA V100 Tensor Core GPU
No robots publicly running it yet.
Data centers with Google TPU v4
Data centers with NVIDIA V100 Tensor Core GPU
No data centers publicly running it yet.
Common questions
Google TPU v4 vs NVIDIA V100 Tensor Core GPU: which is faster for training?
Google TPU v4 has 2.20x the dense FP16/BF16 throughput of the other (Google TPU v4: 275 TFLOPS; NVIDIA V100 Tensor Core GPU: 125 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.
Google TPU v4 vs NVIDIA V100 Tensor Core GPU: which has more memory?
NVIDIA V100 Tensor Core GPU carries more HBM (Google TPU v4: 32 GB HBM2; NVIDIA V100 Tensor Core GPU: 32 GB HBM2). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
Google TPU v4 vs NVIDIA V100 Tensor Core GPU: which is more power-efficient?
Dense FP16/BF16 TFLOPS per watt: Google TPU v4 1.43 (275 ÷ 192 W); NVIDIA V100 Tensor Core GPU 0.42 (125 ÷ 300 W). Google TPU v4 wins at the die level; system-level efficiency also depends on cooling and interconnect.
Google TPU v4 vs NVIDIA V100 Tensor Core GPU: which is more widely used?
Google TPU v4: 0 robots and 2 data centers publicly running it. NVIDIA V100 Tensor Core GPU: 0 robots and 0 data centers. Empty here means no operator has said publicly, not zero adoption in the market.
See also: every chip comparison · Google TPU v4 full page · NVIDIA V100 Tensor Core GPU full page.