ASUS RTX 3070 Max-Q vs NVIDIA H200 SXM 141 GB

Comparison of ASUS RTX 3070 Max-Q with 8 GB GDDR6 and 5,120 cores vs NVIDIA H200 SXM 141 GB with 141 GB HBM3e and 16,896 cores.

Loading...

Performance Rating

H200 H200
MI325X MI325X
A100 A100

ASUS RTX 3070 Max-Q

ASUS RTX 3070 Max-Q

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

NVIDIA H200 SXM 141 GB

67.4

NVIDIA H200 SXM 141 GB

67.4

Contents:

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

Memory

Memory Size

8 GB
🔥 +1,662% 141 ГБ

Memory Type

GDDR6 HBM3e

Memory Bandwidth

384.0 GB/s
🔥 4.89 TB/s

Memory Bus Width

256 бит 6,144 бит

ML Performance

FP16 (Half Precision)

14.44 TFLOPS
🔥 +1,753% 267.6 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

14.44 TFLOPS
🔥 +363% 66.91 TFLOPS

FP64 (Double Precision)

0.2256 TFLOPS
🔥 +14,727% 33.45 TFLOPS

CUDA Cores

5,120
🔥 +230% 16,896

RT Cores

40 No

Architecture & Compatibility

GPU Architecture

Ampere Hopper

SM (Streaming Multiprocessor)

40
🔥 +230% 132

PCIe Version

PCIe 4.0 x16 PCIe 5.0 x16

ML Software Support

CUDA Version

8.6
🔥 9.0

Clocks & Performance

Base Clock

930
🔥 +61% 1,500

Boost Clock

1,410
🔥 +40% 1,980

Memory Clock

1,500
🔥 +6% 1,593

Power Consumption

TDP/TGP

🔥 -87% 90 W
700 W

Recommended PSU

No 1100 W

Power Connector

None 8-pin EPS

Rendering

Texture Units (TMU)

160
🔥 +230% 528

ROP

40 No

L2 Cache

4 MB
🔥 +1,150% 50 MB

Benchmarks

MLPerf, llama2-70b-99.9 (UNSET)

3 534 tokens/s

MLPerf, llama2-70b-99.9 (fp16)

3 553 tokens/s

MLPerf, llama2-70b-99.9 (fp8)

2 444 tokens/s

MLPerf, llama3.1-405b (fp16)

40.8 tokens/s

MLPerf, llama3.1-405b (fp8)

25.3 tokens/s

MLPerf, llama3.1-8b (fp8)

5 161 tokens/s

MLPerf, deepseek-r1 (fp8)

1 113 tokens/s

MLPerf, mixtral-8x7b (fp8)

7 132 tokens/s

Additional

Slots

No
🔥 SXM Module

Release Date

Jan. 12, 2021 Nov. 18, 2024

Display Outputs

Portable Device Dependent
No outputs

Renting is cheaper than buying