NVIDIA H100 PCIe vs AMD Radeon HD 8970 OEM
GPU · Specs Compare
Graphics Card
Release Date
Mar 2022
Jan 2013
Generation
Tesla Hopper
Sea Islands
Type
AI GPU
Desktop
Bus Interface
PCIe 5.0 x16
PCIe 3.0 x16
Clock Speeds
Base Clock
1095 MHz
—
Boost Clock
1755 MHz
—
Memory Clock
1593 MHz
1375 MHz
Memory
Memory Size
80GB
3GB
Memory Type
HBM2e
GDDR5
Memory Bus
5120bit
384bit
Bandwidth
2039GB/s
264.0GB/s
Render Config
SM Count
114
—
Shading Units
14592
2048
TMUs
456
128
ROPs
24
32
Tensor Cores
456
—
L1 Cache
256 KB (per SM)
16 KB (per CU)
L2 Cache
50 MB
768 KB
Compute Units
—
32
Theoretical Performance
Pixel Rate
42.12 GPixel/s
29.60 GPixel/s
Texture Rate
800.3 GTexel/s
118.4 GTexel/s
FP16 (half)
204.9 TFLOPS
—
FP32 (float)
51.22 TFLOPS
3.789 TFLOPS
FP64 (double)
25.61 TFLOPS
947.2 GFLOPS
Graphics Processor
GPU Name
GH100
Tahiti
Architecture
Hopper
GCN 1.0
Foundry
TSMC
TSMC
Process Size
4 nm
28 nm
Transistors
80 billion
4.313 billion
Die Size
814 mm²
352 mm²
GPU Variant
—
Tahiti XT(215-0821060)
Board Design
TDP
350W
250W
Suggested PSU
750 W
600 W
Outputs
No outputs
1x DVI1x HDMI 1.4a2x mini-DisplayPort 1.2
Power Connectors
1x 16-pin
1x 6-pin + 1x 8-pin
Graphics Features
DirectX
N/A
12 (11_1)
OpenGL
N/A
4.6
OpenCL
3.0
1.2
Vulkan
N/A
1.2
CUDA
9.0
—
Shader Model
N/A
5.1
▲ Specs Compare
Data sourced from official specifications and benchmark tests.