Radeon RX 580 and H100 PCIe - video card comparison
Radeon RX 580
H100 PCIe
Architecture
Polaris (2016−2019)
Hopper (2022)
GPU code name
Polaris 20
GH100
Market segment
Desktop
Desktop
Release date
April 2017
March 2022
Interface
PCIe 3.0 x16
PCIe 5.0 x16
Supplementary power connectors
1x 8-pin
8-pin EPS
Boost clock speed
1340 MHz
1650 MHz
Core clock speed
1257 MHz
-
Floating-point performance
6,175 gflops
-
Manufacturing process technology
14 nm
4 nm
Number of transistors
5,700 million
80,000 million
Pipelines / CUDA cores
2304
7296
Power consumption (TDP)
185 Watt
350 Watt
Texture fill rate
193.0
752.4
Memory type
GDDR5
HBM2e
Maximum RAM amount
8 GB
80 GB
Memory bus width
256 Bit
5120 Bit
Memory clock speed
8000 MHz
2 GB/s
Memory bandwidth
256.0 GB/s
1,280 GB/s
Display Connectors
1x HDMI, 3x DisplayPort
No outputs
HDMI
+
-
DirectX
12 (12_0)
N/A
CUDA
-
9.0
OpenCL
2.0
3.0
OpenGL
4.6
N/A
Shader Model
6.4
N/A
Vulkan
1.2.131
N/A