Home
Compare GPUs
MI250 vs A100

AMD MI250 vs Nvidia A100

Two data center GPUs, one CDNA2 and one Ampere: 128 GB HBM2e against 80 GB HBM2e and 65% more memory bandwidth on the MI250.

At a glance

Memory

MI250: 128 GB
A100: 80 GB

1.6x more on MI250

Memory bandwidth

MI250: 3.20 TB/s
A100: 1.94 TB/s

1.7x higher on MI250

Cheapest on-demand

MI250: --
A100: $0.89 /GPU/hr

MI250 is not sold on demand

Price comparison

Median price per GPU per hour by billing type, with the cheapest listing under it.

Billing	MI250	A100
On-demand	No on-demand listings	$1.81 /GPU/hr from $0.89
Reserved	-- /GPU/hr from $1.30 (12 mo)	$1.60 /GPU/hr from $0.66 (24 mo)
Spot	No spot listings	$1.20 /GPU/hr from $0.67

Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Medians are across providers, one vote each, so a large catalog does not outweigh a small one. No median for reserved on the MI250: we only show one when at least 3 providers list the GPU on that billing type. Verify before provisioning. More on how we price.

Price history

MI250 vs A100 price history

Weekly median on-demand price per GPU per hour

Last 12 months

Aggregating historical prices...

Specs comparison

Spec	MI250	A100	Difference
Memory
Capacity	128 GB HBM2e	80 GB HBM2e 40 GB on SXM, PCIe	1.6x more on MI250
Bandwidth	3,200 GB/s	1,935 GB/s 2,039 GB/s on SXM 1,555 GB/s on SXM, PCIe	1.7x higher on MI250
Bus width	8,192-bit	5,120-bit	1.6x wider on MI250
Compute · dense
FP16 / BF16	362.1 TFLOPS	312 TFLOPS	1.2x higher on MI250
INT8	362.1 TOPS	624 TOPS	1.7x higher on A100
FP32	90.5 TFLOPS	19.5 TFLOPS	4.6x higher on MI250
FP64	45.3 TFLOPS	9.7 TFLOPS	4.7x higher on MI250
Platform
Interconnect	Infinity Fabric 400 GB/s	NVLink 600 GB/s	1.5x higher on A100
Board power	500 W	300 W	1.7x higher on MI250
Released	Q4 2021	Q2 2020	2 years newer on MI250
Precision
Supported formats	FP16BF16FP32FP64INT4INT8	FP16BF16TF32FP32FP64INT4INT8

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.

Source: MI250 datasheet and A100 datasheet.

Where to rent them

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

Provider	MI250 /GPU/hr	A100 /GPU/hr
Cirrascale	Reserved (12mo) $1.30	--
GPU.ai	--	On-Demand $0.89
Thunder Compute	--	On-Demand $1.09
Massed Compute	--	On-Demand $1.35
Packet·ai	--	On-Demand $1.43
Spheron	--	On-Demand $1.43
Oblivus	--	On-Demand $1.47
Sesterce	--	On-Demand $1.49
Geodd	--	On-Demand $1.49