NVIDIA H100 PCIe 80 GB vs AMD Radeon HD 6850 1440SP Edition
GPU · Specs Compare
Graphics Card
Release Date
Oct 2022
Aug 2012
Generation
Server Hopper
Northern Islands
Type
AI GPU
Desktop
Bus Interface
PCIe 5.0 x16
PCIe 2.0 x16
Clock Speeds
Base Clock
1095 MHz
—
Boost Clock
1755 MHz
—
Memory Clock
1593 MHz
1000 MHz
Memory
Memory Size
80GB
1024MB
Memory Type
HBM2e
GDDR5
Memory Bus
5120bit
256bit
Bandwidth
2.04TB/s
128.0GB/s
Render Config
SM Count
114
—
Shading Units
14592
1440
TMUs
456
72
ROPs
24
32
Tensor Cores
456
—
L1 Cache
256 KB (per SM)
8 KB (per CU)
L2 Cache
50 MB
512 KB
Compute Units
—
18
Theoretical Performance
Pixel Rate
42.12 GPixel/s
23.20 GPixel/s
Texture Rate
800.3 GTexel/s
52.20 GTexel/s
FP16 (half)
204.9 TFLOPS
—
FP32 (float)
51.22 TFLOPS
2.088 TFLOPS
FP64 (double)
25.61 TFLOPS
417.6 GFLOPS
Graphics Processor
GPU Name
GH100
Cypress
Architecture
Hopper
TeraScale 2
Foundry
TSMC
TSMC
Process Size
5 nm
40 nm
Transistors
80 billion
2.154 billion
Die Size
814 mm²
334 mm²
GPU Variant
—
Cypress PRO(215-0735043)
Board Design
TDP
350W
151W
Suggested PSU
750 W
450 W
Outputs
No outputs
2x DVI1x HDMI 1.3a1x DisplayPort 1.1
Power Connectors
1x 16-pin
2x 6-pin
Graphics Features
DirectX
N/A
11.2 (11_0)
OpenGL
N/A
4.4
OpenCL
3.0
1.2
Vulkan
N/A
N/A
CUDA
9.0
—
Shader Model
N/A
5.0
▲ Specs Compare
Data sourced from official specifications and benchmark tests.