AMD Radeon Instinct MI250 vs NVIDIA A40 PCIe

Comparison of AMD Radeon Instinct MI250 with 128 GB HBM2e and 13,312 cores vs NVIDIA A40 PCIe with 48 GB GDDR6 and 10,752 cores.

Loading...

Performance Rating

AMD Radeon Instinct MI250 outperforms NVIDIA A40 PCIe by 138.14% in the overall GPU ARK performance rating

A100 A100
H200 H200
MI325X MI325X

AMD Radeon Instinct MI250

50.9

AMD Radeon Instinct MI250

50.9
RX 7900 XTX RX 7900 XTX
MI250 MI250
Instinct MI300X Instinct MI300X

NVIDIA A40 PCIe

21.4

NVIDIA A40 PCIe

21.4

Contents:

Memory ML Performance Compute Power Architecture & Compatibility ML Software Support Clocks & Performance Power Consumption Rendering Benchmarks Additional

Memory

Memory Size

🔥 +167% 128 ГБ
48 ГБ

Memory Type

HBM2e GDDR6

Memory Bandwidth

🔥 3.28 TB/s
695.8 GB/s

Memory Bus Width

8,192 бит 384 бит

ML Performance

FP16 (Half Precision)

🔥 +868% 362.1 TFLOPS
37.42 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

🔥 +21% 45.26 TFLOPS
37.42 TFLOPS

FP64 (Double Precision)

🔥 +7,642% 45.26 TFLOPS
0.5846 TFLOPS

CUDA Cores

🔥 +24% 13,312
10,752

RT Cores

No
🔥 84

Architecture & Compatibility

GPU Architecture

CDNA 2.0 Ampere

SM (Streaming Multiprocessor)

No
🔥 84

PCIe Version

PCIe 4.0 x16 PCIe 4.0 x16

ML Software Support

CUDA Version

No 8.6

Clocks & Performance

Base Clock

1,000
🔥 +30% 1,305

Boost Clock

1,700
🔥 +2% 1,740

Memory Clock

1,600
🔥 +13% 1,812

Power Consumption

TDP/TGP

500 W
🔥 -40% 300 W

Recommended PSU

900 W
🔥 -22% 700 W

Power Connector

2x 8-pin 8-pin EPS

Rendering

Texture Units (TMU)

🔥 +148% 832
336

ROP

No
🔥 84

L2 Cache

16 MB
🔥 6 MB

Benchmarks

llama.cpp, llama-2-7b-Q4_0

63.9 tokens/s

Additional

Slots

Dual-slot Dual-slot

Release Date

Nov. 8, 2021 Oct. 5, 2020

Display Outputs

No outputs
3x DisplayPort 1.4a

Renting is cheaper than buying