NVIDIA H100 PCIe 80 GB vs NVIDIA A100 SXM4 40 GB

Comparative analysis of NVIDIA H100 PCIe 80 GB and NVIDIA A100 SXM4 40 GB videocards for all known characteristics in the following categories: Essentials, Technical info, Video outputs and ports, Compatibility, dimensions and requirements, API support, Memory. Benchmark videocards performance analysis: Geekbench - OpenCL, GFXBench 4.0 - Car Chase Offscreen (Frames), GFXBench 4.0 - Car Chase Offscreen (Fps), GFXBench 4.0 - Manhattan (Frames), GFXBench 4.0 - Manhattan (Fps), GFXBench 4.0 - T-Rex (Frames), GFXBench 4.0 - T-Rex (Fps).

NVIDIA H100 PCIe 80 GB

NVIDIA A100 SXM4 40 GB

Differences

Reasons to consider the NVIDIA H100 PCIe 80 GB

Videocard is newer: launch date 1 year(s) 10 month(s) later
Around 24% higher boost clock speed: 1755 MHz vs 1410 MHz
Around 31% higher texture fill rate: 800.3 GTexel/s vs 609.1 GTexel/s
2.1x more pipelines: 14592 vs 6912
A newer manufacturing process allows for a more powerful, yet cooler running videocard: 4 nm vs 7 nm
Around 14% lower typical power consumption: 350 Watt vs 400 Watt
2x more maximum memory size: 80 GB vs 40 GB
Around 31% higher memory clock speed: 1593 MHz, 3.2 Gbps effective vs 1215 MHz (2.4 Gbps effective)
Around 40% better performance in Geekbench - OpenCL: 281702 vs 200574

Specifications (specs)
Launch date	22 Mar 2022 vs 14 May 2020
Boost clock speed	1755 MHz vs 1410 MHz
Texture fill rate	800.3 GTexel/s vs 609.1 GTexel/s
Pipelines	14592 vs 6912
Manufacturing process technology	4 nm vs 7 nm
Thermal Design Power (TDP)	350 Watt vs 400 Watt
Maximum memory size	80 GB vs 40 GB
Memory clock speed	1593 MHz, 3.2 Gbps effective vs 1215 MHz (2.4 Gbps effective)
Benchmarks
Geekbench - OpenCL	281702 vs 200574

Compare benchmarks

GPU 1: NVIDIA H100 PCIe 80 GB
GPU 2: NVIDIA A100 SXM4 40 GB

Geekbench - OpenCL

GPU 1

GPU 2

281702

200574

Name	NVIDIA H100 PCIe 80 GB	NVIDIA A100 SXM4 40 GB
Geekbench - OpenCL	281702	200574
GFXBench 4.0 - Car Chase Offscreen (Frames)		21006
GFXBench 4.0 - Car Chase Offscreen (Fps)		21006
GFXBench 4.0 - Manhattan (Frames)		27823
GFXBench 4.0 - Manhattan (Fps)		27823
GFXBench 4.0 - T-Rex (Frames)		51880
GFXBench 4.0 - T-Rex (Fps)		51880

Compare specifications (specs)

	NVIDIA H100 PCIe 80 GB	NVIDIA A100 SXM4 40 GB
Essentials
Architecture	Hopper	Ampere
Code name	GH100	GA100
Launch date	22 Mar 2022	14 May 2020
Place in performance rating	3	6
Technical info
Boost clock speed	1755 MHz	1410 MHz
Core clock speed	1095 MHz	1095 MHz
Manufacturing process technology	4 nm	7 nm
Pipelines	14592	6912
Pixel fill rate	42.12 GPixel/s	225.6 GPixel/s
Texture fill rate	800.3 GTexel/s	609.1 GTexel/s
Thermal Design Power (TDP)	350 Watt	400 Watt
Transistor count	80000 million	54200 million
Peak Double Precision (FP64) Performance		9.746 TFLOPS (1:2)
Peak Half Precision (FP16) Performance		77.97 TFLOPS (4:1)
Peak Single Precision (FP32) Performance		19.49 TFLOPS
Video outputs and ports
Display Connectors	No outputs	No outputs
Compatibility, dimensions and requirements
Form factor	Dual-slot	IGP
Interface	PCIe 5.0 x16	PCIe 4.0 x16
Length	268 mm, 10.6 inches
Recommended system power (PSU)	750 Watt	800 Watt
Supplementary power connectors	1x 16-pin	None
Width	111 mm, 4.4 inches
API support
OpenCL	3.0	3.0
Memory
Maximum RAM amount	80 GB	40 GB
Memory bandwidth	2,039 GB/s	1555 GB/s
Memory bus width	5120 bit	5120 bit
Memory clock speed	1593 MHz, 3.2 Gbps effective	1215 MHz (2.4 Gbps effective)
Memory type	HBM2e	HBM2e
High bandwidth memory (HBM)