AMD Radeon Instinct MI300X vs NVIDIA Switch GPU 20nm

Comparison of AMD Radeon Instinct MI300X with 192 GB HBM3 and 19,456 cores vs NVIDIA Switch GPU 20nm with 4 GB DDR4 and 256 cores.

Loading...

Performance Rating

H200 H200
MI325X MI325X
A100 A100

AMD Radeon Instinct MI300X

AMD Radeon Instinct MI300X

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

NVIDIA Switch GPU 20nm

NVIDIA Switch GPU 20nm

Contents:

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

Memory

Memory Size

192 GB 4 GB

Memory Type

HBM3 DDR4

Memory Bandwidth

10.3 TB/s 25.60 GB/s

Memory Bus Width

8,192 бит 64 бит

ML Performance

FP16 (Half Precision)

653.7 TFLOPS 0.7864 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

81.72 TFLOPS 0.3932 TFLOPS

FP64 (Double Precision)

81.72 TFLOPS 0.0123 TFLOPS

CUDA Cores

19,456 256

RT Cores

No No

Architecture & Compatibility

GPU Architecture

CDNA 3.0 Maxwell 2.0

SM (Streaming Multiprocessor)

No No

PCIe Version

PCIe 5.0 x16 No

ML Software Support

CUDA Version

No 5.3

Clocks & Performance

Base Clock

1,000 384

Boost Clock

2,100 768

Memory Clock

2,525 1,600

Power Consumption

TDP/TGP

750 W
🔥 -98% 15 W

Recommended PSU

1150 W No

Power Connector

None No

Rendering

Texture Units (TMU)

1,216 16

ROP

No No

L2 Cache

16 MB No

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

232.9 tokens/s

MLPerf, mixtral-8x7b (fp8)

5 975 tokens/s

Additional

Slots

OAM Module No

Release Date

Dec. 6, 2023 March 17, 2017

Display Outputs

No outputs
1x USB Type-C

Renting is cheaper than buying