Tesla K80 and Tesla K20X - video card comparison
Tesla K80
Tesla K20X
Architecture
Kepler 2.0 (2013−2015)
Kepler (2012−2018)
GPU code name
GK210
GK110
Market segment
Workstation
Workstation
Release date
November 2014
November 2012
Interface
PCIe 3.0 x16
PCIe 3.0 x16
Supplementary power connectors
1x 8-pin
1x 6-pin + 1x 8-pin
Boost clock speed
824 MHz
-
Core clock speed
562 MHz
732 MHz
Floating-point performance
-
3,935 gflops
Manufacturing process technology
28 nm
28 nm
Number of transistors
7,100 million
7,080 million
Pipelines / CUDA cores
2496
2688
Power consumption (TDP)
300 Watt
235 Watt
Texture fill rate
171.4
164.0
Memory type
GDDR5
GDDR5
Maximum RAM amount
12 GB
6 GB
Memory bus width
384 Bit
384 Bit
Memory clock speed
5012 MHz
5200 MHz
Memory bandwidth
240.6 GB/s
249.6 GB/s
Display Connectors
No outputs
No outputs
DirectX
12 (11_1)
12 (11_0)
CUDA
3.7
3.5
OpenCL
1.2
1.2
OpenGL
4.6
4.6
Shader Model
5.1
5.1
Vulkan
1.1.126
1.1.126