NVIDIA GeForce RTX 4090 Max-Q vs NVIDIA H200 SXM 141 GB

Comparison NVIDIA GeForce RTX 4090 Max-Q with 16 GB GDDR6 and 9,728 cores vs NVIDIA H200 SXM 141 GB with 141 GB HBM3e and 16,896 cores.

Loading...

Performance Rating

NVIDIA H200 SXM 141 GB outperforms NVIDIA GeForce RTX 4090 Max-Q by 368.89% in the overall GPU ARK performance rating

A100 A100
H200 H200
MI325X MI325X

NVIDIA GeForce RTX 4090 Max-Q

14.4

NVIDIA GeForce RTX 4090 Max-Q

14.4
RX 7900 XTX RX 7900 XTX
MI250 MI250
Instinct MI300X Instinct MI300X

NVIDIA H200 SXM 141 GB

67.4

NVIDIA H200 SXM 141 GB

67.4

Expert Comparison

NVIDIA GeForce RTX 4090 Max-Q лучше для среднего уровня игр и творческих задач благодаря своей более низкой потребляемой мощности (80 Вт) и хорошему соотношению цены и производительности (GPI 14.4). NVIDIA H200 SXM 141 GB же значительно превосходит в мощности для вычислений и больших данных из-за огромной памяти (141 ГБ HBM3e) и высокой пропускной способности (4.89 TB/s), но при этом требует гораздо больше энергии (700 Вт) и доступен только для серверных решений.

Contents:

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

Memory

Memory Size

16 GB
🔥 +781% 141 ГБ

Memory Type

GDDR6 HBM3e

Memory Bandwidth

576.0 GB/s
🔥 4.89 TB/s

Memory Bus Width

256 бит 6,144 бит

ML Performance

FP16 (Half Precision)

28.31 TFLOPS
🔥 +845% 267.6 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

28.31 TFLOPS
🔥 +136% 66.91 TFLOPS

FP64 (Double Precision)

0.4423 TFLOPS
🔥 +7,463% 33.45 TFLOPS

CUDA Cores

9,728
🔥 +74% 16,896

RT Cores

🔥 76
No

Architecture & Compatibility

GPU Architecture

Ada Lovelace Hopper

SM (Streaming Multiprocessor)

76
🔥 +74% 132

PCIe Version

PCIe 4.0 x16 PCIe 5.0 x16

ML Software Support

CUDA Version

8.9
🔥 9.0

CUDA Toolkit (first supported)

v11 v12

CUDA Toolkit status

Supported Supported

Clocks & Performance

Base Clock

930
🔥 +61% 1,500

Boost Clock

1,455
🔥 +36% 1,980

Memory Clock

🔥 +41% 2,250
1,593

Power Consumption

Recommended PSU

No 1100 W

Power Connector

None 8-pin EPS

TDP/TGP

🔥 -89% 80 W
700 W

Rendering

Texture Units (TMU)

304
🔥 +74% 528

ROP

🔥 76
No

L2 Cache

🔥 +28% 64 MB
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

IGP
🔥 SXM Module

Release Date

Jan. 3, 2023 Nov. 18, 2024

Display Outputs

Portable Device Dependent
No outputs

Rent on Intelion Cloud

H200 (141GB)

Available for rent on Intelion Cloud

from 375.0 RUB/hour Rent
Pay-as-you-go, per-second billing
Prices include 22% VAT
EDI & closing documents
Server ready in 3 minutes
Russian-speaking support
Card, SBP or invoice