NVIDIA A100 PCIe 80 GB vs NVIDIA H100 SXM5 94 GB

Comparison of NVIDIA A100 PCIe 80 GB with 80 GB HBM2e and 6,912 cores vs NVIDIA H100 SXM5 94 GB with 94 GB HBM3 and 16,896 cores.

Loading...

Performance Rating

H200 H200
MI325X MI325X
A100 A100

NVIDIA A100 PCIe 80 GB

NVIDIA A100 PCIe 80 GB

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

NVIDIA H100 SXM5 94 GB

57.8

NVIDIA H100 SXM5 94 GB

57.8

Contents:

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

Memory

Memory Size

80 GB
🔥 +18% 94 ГБ

Memory Type

HBM2e HBM3

Memory Bandwidth

1.94 TB/s
🔥 +73% 3.36 TB/s

Memory Bus Width

5,120 бит 5,120 бит

ML Performance

FP16 (Half Precision)

77.97 TFLOPS
🔥 +243% 267.6 TFLOPS

BF16 (Brain Float)

311.84 TFLOPS No

TF32 (TensorFloat)

155.92 No

Compute Power

FP32 (Single Precision)

19.49 TFLOPS
🔥 +243% 66.91 TFLOPS

FP64 (Double Precision)

9.746 TFLOPS
🔥 +243% 33.45 TFLOPS

CUDA Cores

6,912
🔥 +144% 16,896

RT Cores

No No

Architecture & Compatibility

GPU Architecture

Ampere Hopper

SM (Streaming Multiprocessor)

108
🔥 +22% 132

PCIe Version

PCIe 4.0 x16 PCIe 5.0 x16

ML Software Support

CUDA Version

8.0
🔥 9.0

Clocks & Performance

Base Clock

1,065
🔥 +27% 1,350

Boost Clock

1,410
🔥 +40% 1,980

Memory Clock

1,512
🔥 1,313

Power Consumption

TDP/TGP

🔥 -57% 300 W
700 W

Recommended PSU

🔥 -36% 700 W
1100 W

Power Connector

8-pin EPS 8-pin EPS

Rendering

Texture Units (TMU)

432
🔥 +22% 528

ROP

No No

L2 Cache

80 MB
🔥 50 MB

Benchmarks

MLPerf, llama2-70b-99.9 (fp8)

2 106 tokens/s

MLPerf, mixtral-8x7b (fp8)

4 714 tokens/s

Additional

Slots

Dual-slot
🔥 SXM Module

Release Date

June 28, 2021 March 21, 2023

Display Outputs

No outputs
No outputs

Renting is cheaper than buying