AMD Radeon Instinct MI250 vs AMD Radeon Instinct MI300X

Comparison of AMD Radeon Instinct MI250 with 128 GB HBM2e and 13,312 cores vs AMD Radeon Instinct MI300X with 192 GB HBM3 and 19,456 cores.

Loading...

Performance Rating

AMD Radeon Instinct MI300X outperforms AMD Radeon Instinct MI250 by 84.91% 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

AMD Radeon Instinct MI300X

94.1

AMD Radeon Instinct MI300X

94.1

Contents:

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

Memory

Memory Size

128 ГБ
🔥 +50% 192 ГБ

Memory Type

HBM2e HBM3

Memory Bandwidth

3.28 TB/s
🔥 +214% 10.3 TB/s

Memory Bus Width

8,192 бит 8,192 бит

ML Performance

FP16 (Half Precision)

362.1 TFLOPS
🔥 +81% 653.7 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

45.26 TFLOPS
🔥 +81% 81.72 TFLOPS

FP64 (Double Precision)

45.26 TFLOPS
🔥 +81% 81.72 TFLOPS

CUDA Cores

13,312
🔥 +46% 19,456

RT Cores

No No

Architecture & Compatibility

GPU Architecture

CDNA 2.0 CDNA 3.0

SM (Streaming Multiprocessor)

No No

PCIe Version

PCIe 4.0 x16 PCIe 5.0 x16

ML Software Support

CUDA Version

No No

Clocks & Performance

Base Clock

1,000 1,000

Boost Clock

1,700
🔥 +24% 2,100

Memory Clock

1,600
🔥 +58% 2,525

Power Consumption

TDP/TGP

🔥 -33% 500 W
750 W

Recommended PSU

🔥 -22% 900 W
1150 W

Power Connector

2x 8-pin None

Rendering

Texture Units (TMU)

832
🔥 +46% 1,216

ROP

No No

L2 Cache

16 MB 16 MB

Benchmarks

MLPerf, llama2-70b-99.9 (UNSET)

1 983 tokens/s

MLPerf, llama2-70b-99.9 (fp16)

1 740 tokens/s

MLPerf, llama2-70b-99.9 (fp8)

1 057 tokens/s

MLPerf, llama3.1-405b (UNSET)

30.4 tokens/s

MLPerf, llama3.1-405b (fp16)

34.8 tokens/s

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

63.9 tokens/s
🔥 +264% 232.9 tokens/s

MLPerf, mixtral-8x7b (fp8)

5 975 tokens/s

Additional

Slots

Dual-slot OAM Module

Release Date

Nov. 8, 2021 Dec. 6, 2023

Display Outputs

No outputs
No outputs

Renting is cheaper than buying