Home GPU Comparison NVIDIA A100 PCIe vs NVIDIA H100 SXM5 64 GB

NVIDIA A100 PCIe vs NVIDIA H100 SXM5 64 GB

AI GPUAI GPU We compared two GPUs: 40GB VRAM A100 PCIe and 64GB VRAM H100 SXM5 64 GB to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.

Main Differences

NVIDIA A100 PCIe 's Advantages
Larger VRAM bandwidth (1555GB/s vs 1008GB/s)
Lower TDP (250W vs 700W)
NVIDIA H100 SXM5 64 GB 's Advantages
Released 2 years and 9 months late
Boost Clock has increased by 40% (1980MHz vs 1410MHz)
More VRAM (64GB vs 40GB)
9984 additional rendering cores

Score

Benchmark

FP32 (float)
A100 PCIe
19.49 TFLOPS
H100 SXM5 64 GB +243%
66.91 TFLOPS
VS

Graphics Card

Jun 2020
Release Date
Mar 2023
Tesla
Generation
Tesla Hopper
AI GPU
Type
AI GPU
PCIe 4.0 x16
Bus Interface
PCIe 5.0 x16

Clock Speeds

765 MHz
Base Clock
1665 MHz
1410 MHz
Boost Clock
1980 MHz
1215 MHz
Memory Clock
1313 MHz

Memory

40GB
Memory Size
64GB
HBM2e
Memory Type
HBM3
5120bit
Memory Bus
3072bit
1555GB/s
Bandwidth
1008GB/s

Render Config

108
SM Count
132
-
Compute Units
-
6912
Shading Units
16896
432
TMUs
528
160
ROPs
24
432
Tensor Cores
528
-
RT Cores
-
192 KB (per SM)
L1 Cache
256 KB (per SM)
40 MB
L2 Cache
50 MB

Theoretical Performance

225.6 GPixel/s
Pixel Rate
47.52 GPixel/s
609.1 GTexel/s
Texture Rate
1045 GTexel/s
77.97 TFLOPS
FP16 (half)
267.6 TFLOPS
19.49 TFLOPS
FP32 (float)
66.91 TFLOPS
9.746 TFLOPS
FP64 (double)
33.45 TFLOPS

Graphics Processor

GA100
GPU Name
GH100
-
GPU Variant
-
Ampere
Architecture
Hopper
TSMC
Foundry
TSMC
7 nm
Process Size
4 nm
54.2 billion
Transistors
80 billion
826 mm²
Die Size
814 mm²

Board Design

250W
TDP
700W
600 W
Suggested PSU
1100 W
No outputs
Outputs
No outputs
8-pin EPS
Power Connectors
8-pin EPS

Graphics Features

N/A
DirectX
N/A
N/A
OpenGL
N/A
3.0
OpenCL
3.0
N/A
Vulkan
N/A
8.0
CUDA
9.0
N/A
Shader Model
N/A

Related GPU Comparisons

© 2024 - TopCPU.net   Contact Us Privacy Policy