AMD Radeon Instinct MI300X vs Intel Data Center GPU Max Subsystem

Comparison AMD Radeon Instinct MI300X with 192 GB HBM3 and 19,456 cores vs Intel Data Center GPU Max Subsystem with 128 GB HBM2e and 16,384 cores.

Loading...

Performance Rating

AMD Radeon Instinct MI300X outperforms Intel Data Center GPU Max Subsystem by 71.28% in the overall GPU ARK performance rating

A100 A100
H200 H200
MI325X MI325X

AMD Radeon Instinct MI300X

94.1

AMD Radeon Instinct MI300X

94.1
RX 7900 XTX RX 7900 XTX
MI250 MI250
Instinct MI300X Instinct MI300X

Intel Data Center GPU Max Subsystem

54.9

Intel Data Center GPU Max Subsystem

54.9

Expert Comparison

AMD Radeon Instinct MI300X лучше по пропускной способности памяти (10.3 TB/s против 3.21 TB/s) и FP32 вычислениям (81.72 TFLOPS против 52.43 TFLOPS), что делает его более подходящим для высокопроизводительных вычислений. Однако Intel Data Center GPU Max Subsystem имеет значительно большую мощность потребления (2400 Вт против 750 Вт), что может быть важным фактором при выборе оборудования с ограниченными мощностями.

Contents:

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

Memory

Memory Size

🔥 +50% 192 ГБ
128 GB

Memory Type

HBM3 HBM2e

Memory Bandwidth

🔥 +221% 10.3 TB/s
3.21 TB/s

Memory Bus Width

8,192 бит 8,192 бит

ML Performance

FP16 (Half Precision)

🔥 +1,147% 653.7 TFLOPS
52.43 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

🔥 +56% 81.72 TFLOPS
52.43 TFLOPS

FP64 (Double Precision)

🔥 +56% 81.72 TFLOPS
52.43 TFLOPS

CUDA Cores

🔥 +19% 19,456
16,384

RT Cores

No
🔥 128

Architecture & Compatibility

GPU Architecture

CDNA 3.0 Generation 12.5

SM (Streaming Multiprocessor)

No No

PCIe Version

PCIe 5.0 x16 PCIe 5.0 x16

ML Software Support

CUDA Version

No No

CUDA Toolkit status

Supported Supported

Clocks & Performance

Base Clock

🔥 +11% 1,000
900

Boost Clock

🔥 +31% 2,100
1,600

Memory Clock

🔥 +61% 2,525
1,565

Power Consumption

Recommended PSU

🔥 -59% 1150 W
2800 W

Power Connector

None 1x 16-pin

TDP/TGP

🔥 -69% 750 W
2400 W

Rendering

Texture Units (TMU)

🔥 +19% 1,216
1,024

ROP

No
🔥 128

L2 Cache

16 MB
🔥 +2,450% 408 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

232.9 tokens/s

MLPerf, mixtral-8x7b (fp8)

5 975 tokens/s

Additional

Slots

OAM Module Dual-slot

Release Date

Dec. 6, 2023 Jan. 10, 2023

Display Outputs

No outputs
No outputs

Renting is cheaper than buying