AMD MI250 vs Nvidia V100

Two data center GPUs, one CDNA2 and one Volta: 128 GB HBM2e against 32 GB HBM2 and 3.6x the memory bandwidth on the MI250.

At a glance

Updated 7 minutes ago

AMD MI250

Nvidia V100

Cheapest listed price per GPU per hour by billing type, and the median across providers.

Billing	MI250	V100	Difference
On-demand	Not sold on demand	$0.14
Spot	No spot listings	$0.09
Reserved	$1.30	$0.13	10x more expensive on MI250
Median	$1.50	$1.57	Within 5%

Median: Median across providers, one vote each, so a large catalog does not outweigh a small one.

Weekly median on-demand price per GPU per hour

Last 12 months

Aggregating historical prices...

Spec	MI250	V100	Difference
Memory
Capacity	128 GB HBM2e	32 GB HBM2	4x more on MI250
Bandwidth	3,200 GB/s	900 GB/s	3.6x higher on MI250
Bus width	8,192-bit	4,096-bit	2x wider on MI250
Compute · dense
FP16 / BF16	362.1 TFLOPS	125 TFLOPS	2.9x higher on MI250
INT8	362.1 TOPS
FP32	90.5 TFLOPS	15.7 TFLOPS	5.8x higher on MI250
FP64	45.3 TFLOPS	7.8 TFLOPS	5.8x higher on MI250
Platform
Interconnect	Infinity Fabric 400 GB/s	NVLink 300 GB/s	1.3x higher on MI250
Board power	500 W	300 W	1.7x higher on MI250
Released	Q4 2021	Q2 2017	4 years newer on MI250
Precision
Supported formats	FP16BF16FP32FP64INT4INT8	FP16FP32FP64INT8

Figures are for each model's headline configuration. Where a form factor carries a different number, it is named under the figure.

Compute: Vendor peak figures, dense and per GPU. Not measured throughput.

Providers listing these GPUs, those with both first, so you can switch without moving clouds.