AMD Radeon Instinct MI250 vs NVIDIA Tesla M40

Comparison of AMD Radeon Instinct MI250 with 128 GB HBM2e and 13,312 cores vs NVIDIA Tesla M40 with 12 GB GDDR5 and 3,072 cores.

Loading...

Performance Rating

AMD Radeon Instinct MI250 outperforms NVIDIA Tesla M40 by 791.24% 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 Tesla M40

5.7

NVIDIA Tesla M40

5.7

Contents:

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

Memory

Memory Size

🔥 +967% 128 ГБ
12 ГБ

Memory Type

HBM2e GDDR5

Memory Bandwidth

🔥 3.28 TB/s
288.4 GB/s

Memory Bus Width

8,192 бит 384 бит

ML Performance

FP16 (Half Precision)

🔥 362.1 TFLOPS
No

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

🔥 +562% 45.26 TFLOPS
6.832 TFLOPS

FP64 (Double Precision)

🔥 +21,099% 45.26 TFLOPS
0.2135 TFLOPS

CUDA Cores

🔥 +333% 13,312
3,072

RT Cores

No No

Architecture & Compatibility

GPU Architecture

CDNA 2.0 Maxwell 2.0

SM (Streaming Multiprocessor)

No No

PCIe Version

PCIe 4.0 x16 PCIe 3.0 x16

ML Software Support

CUDA Version

No 5.2

Clocks & Performance

Base Clock

🔥 +5% 1,000
948

Boost Clock

🔥 +53% 1,700
1,112

Memory Clock

🔥 +7% 1,600
1,502

Power Consumption

TDP/TGP

500 W
🔥 -50% 250 W

Recommended PSU

900 W
🔥 -33% 600 W

Power Connector

2x 8-pin 8-pin EPS

Rendering

Texture Units (TMU)

🔥 +333% 832
192

ROP

No No

L2 Cache

🔥 +433% 16 MB
3 MB

Benchmarks

llama.cpp, gpt-oss 20B Q4_K - Medium

47.0 tokens/s

llama.cpp, llama 7B Q4_0

36.7 tokens/s

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

🔥 +53% 63.9 tokens/s
41.7 tokens/s

llama.cpp, qwen3 32B Q4_K - Medium

7.19 tokens/s

llama.cpp, qwen3moe 30B.A3B Q4_K - Medium

35.1 tokens/s

Additional

Slots

Dual-slot Dual-slot

Release Date

Nov. 8, 2021 Nov. 10, 2015

Display Outputs

No outputs
No outputs

Renting is cheaper than buying