NVIDIA Tesla K20m vs NVIDIA Rubin Ultra GPU
GPU · Specs Compare
Graphics Card
Release Date
Jan 2013
Jan 2027
Generation
Tesla
Server Rubin
Type
Professional
Desktop
Bus Interface
PCIe 2.0 x16
PCIe 6.0 x16
Clock Speeds
Memory Clock
1300 MHz
1900 MHz
Base Clock
—
700 MHz
Boost Clock
—
2267 MHz
Memory
Memory Size
5GB
1024GB
Memory Type
GDDR5
HBM4e
Memory Bus
320bit
32768bit
Bandwidth
208.0GB/s
31.1TB/s
Render Config
Shading Units
2496
57344
TMUs
208
1792
ROPs
40
24
L1 Cache
16 KB (per SMX)
256 KB (per SM)
L2 Cache
1280 KB
128 MB
SM Count
—
448
Tensor Cores
—
1792
Theoretical Performance
Pixel Rate
36.71 GPixel/s
54.41 GPixel/s
Texture Rate
146.8 GTexel/s
4062.5 GTexel/s
FP32 (float)
3.524 TFLOPS
260.0 TFLOPS
FP64 (double)
1175 GFLOPS
65.00 TFLOPS
FP16 (half)
—
520.0 TFLOPS
Graphics Processor
GPU Name
GK110
GR110
Architecture
Kepler
Rubin Ultra
Foundry
TSMC
TSMC
Process Size
28 nm
3 nm
Transistors
7.08 billion
670 billion
Die Size
561 mm²
2912 mm²
Board Design
TDP
225W
3600W
Suggested PSU
550 W
4000 W
Outputs
No outputs
No outputs
Power Connectors
1x 6-pin + 1x 8-pin
—
Graphics Features
DirectX
12 (11_0)
N/A
OpenGL
4.6
N/A
OpenCL
3.0
3.0
Vulkan
1.1
N/A
CUDA
3.5
10.7
Shader Model
5.1
N/A
▲ Specs Compare
Data sourced from official specifications and benchmark tests.