Tesla K20s vs GRID A100A
We compared two Professional market GPUs: 5GB VRAM Tesla K20s and 48GB VRAM GRID A100A to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Tesla K20s 's Advantages
Lower TDP (225W vs 400W)
GRID A100A 's Advantages
Released 7 years and 3 months late
Boost Clock1005MHz
More VRAM (48GB vs 5GB)
Larger VRAM bandwidth (1866GB/s vs 208.0GB/s)
4416 additional rendering cores
Score
Benchmark
FP32 (float)
Tesla K20s
2870
GRID A100A
+383%
13890
Graphics Card
Feb 2013
Release Date
May 2020
Tesla
Generation
GRID
Professional
Type
Professional
PCIe 2.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
-
Base Clock
900 MHz
-
Boost Clock
1005 MHz
1300 MHz
Memory Clock
1215 MHz
Memory
5GB
Memory Size
48GB
GDDR5
Memory Type
HBM2e
320bit
Memory Bus
6144bit
208.0GB/s
Bandwidth
1866GB/s
Render Config
-
SM Count
108
2496
Shading Units
6912
208
TMUs
432
40
ROPs
192
-
Tensor Cores
432
16 KB (per SMX)
L1 Cache
192 KB (per SM)
1280 KB
L2 Cache
48 MB
Theoretical Performance
29.90 GPixel/s
Pixel Rate
193.0 GPixel/s
119.6 GTexel/s
Texture Rate
434.2 GTexel/s
-
FP16 (half)
55.57 TFLOPS
2.870 TFLOPS
FP32 (float)
13.89 TFLOPS
956.8 GFLOPS
FP64 (double)
6.947 TFLOPS
Board Design
225W
TDP
400W
550 W
Suggested PSU
800 W
No outputs
Outputs
No outputs
-
Power Connectors
None
Graphics Processor
GK110
GPU Name
GA100
Kepler
Architecture
Ampere
TSMC
Foundry
TSMC
28 nm
Process Size
7 nm
7.08 billion
Transistors
54.2 billion
561 mm²
Die Size
826 mm²
Graphics Features
12 (11_0)
DirectX
N/A
4.6
OpenGL
N/A
3.0
OpenCL
3.0
1.1
Vulkan
N/A
3.5
CUDA
8.0
5.1
Shader Model
N/A
