NVIDIA H200 SXM 141 GB vs NVIDIA L4

Comparison of NVIDIA H200 SXM 141 GB with 141 GB HBM3e and 16,896 cores vs NVIDIA L4 with 24 GB GDDR6 and 7,424 cores.

Loading...

Performance Rating

NVIDIA H200 SXM 141 GB outperforms NVIDIA L4 by 398.37% in the overall GPU ARK performance rating

A100 A100
H200 H200
MI325X MI325X

NVIDIA H200 SXM 141 GB

67.4

NVIDIA H200 SXM 141 GB

67.4
RX 7900 XTX RX 7900 XTX
MI250 MI250
Instinct MI300X Instinct MI300X

NVIDIA L4

13.5

NVIDIA L4

13.5

Contents:

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

Memory

Memory Size

🔥 +488% 141 ГБ
24 ГБ

Memory Type

HBM3e GDDR6

Memory Bandwidth

🔥 4.89 TB/s
300.1 GB/s

Memory Bus Width

6,144 бит 192 бит

ML Performance

FP16 (Half Precision)

🔥 +783% 267.6 TFLOPS
30.29 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

🔥 +121% 66.91 TFLOPS
30.29 TFLOPS

FP64 (Double Precision)

🔥 +6,967% 33.45 TFLOPS
0.4733 TFLOPS

CUDA Cores

🔥 +128% 16,896
7,424

RT Cores

No
🔥 60

Architecture & Compatibility

GPU Architecture

Hopper Ada Lovelace

SM (Streaming Multiprocessor)

🔥 +120% 132
60

PCIe Version

PCIe 5.0 x16 PCIe 4.0 x16

ML Software Support

CUDA Version

🔥 9.0
8.9

Clocks & Performance

Base Clock

🔥 +89% 1,500
795

Boost Clock

1,980
🔥 +3% 2,040

Memory Clock

🔥 +2% 1,593
1,563

Power Consumption

TDP/TGP

700 W
🔥 -90% 72 W

Recommended PSU

1100 W
🔥 -77% 250 W

Power Connector

8-pin EPS None

Rendering

Texture Units (TMU)

🔥 +120% 528
240

ROP

No
🔥 60

L2 Cache

🔥 +4% 50 MB
48 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

🔥 SXM Module
Single-slot

Release Date

Nov. 18, 2024 March 21, 2023

Display Outputs

No outputs
No outputs

Renting is cheaper than buying