NVIDIA Tesla K20X vs AMD Radeon Instinct MI8
We compared two Professional market GPUs: 6GB VRAM Tesla K20X and 4GB VRAM Radeon Instinct MI8 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA Tesla K20X 's Advantages
More VRAM (6GB vs 4GB)
AMD Radeon Instinct MI8 's Advantages
Released 4 years and 1 months late
Larger VRAM bandwidth (512.0GB/s vs 249.6GB/s)
1408 additional rendering cores
Lower TDP (175W vs 235W)
Score
Benchmark
FP32 (float)
Tesla K20X
3.935 TFLOPS
Radeon Instinct MI8
+108%
8.192 TFLOPS
Graphics Card
Nov 2012
Release Date
Dec 2016
Tesla
Generation
Radeon Instinct
Professional
Type
Professional
PCIe 3.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
-
-
-
-
-
-
1300 MHz
Memory Clock
500 MHz
Memory
6GB
Memory Size
4GB
GDDR5
Memory Type
HBM
384bit
Memory Bus
4096bit
249.6GB/s
Bandwidth
512.0GB/s
Render Config
-
Compute Units
64
-
-
-
2688
Shading Units
4096
224
TMUs
256
48
ROPs
64
-
-
-
-
-
-
16 KB (per SMX)
L1 Cache
16 KB (per CU)
1536 KB
L2 Cache
2 MB
-
-
-
Theoretical Performance
40.99 GPixel/s
Pixel Rate
64.00 GPixel/s
164.0 GTexel/s
Texture Rate
256.0 GTexel/s
-
FP16 (half)
8.192 TFLOPS
3.935 TFLOPS
FP32 (float)
8.192 TFLOPS
1312 GFLOPS
FP64 (double)
512.0 GFLOPS
Board Design
235W
TDP
175W
550 W
Suggested PSU
450 W
No outputs
Outputs
No outputs
1x 6-pin + 1x 8-pin
Power Connectors
1x 8-pin
Graphics Processor
GK110
GPU Name
Fiji
-
GPU Variant
Fiji XT CA
(215-0862120)
Kepler
Architecture
GCN 3.0
TSMC
Foundry
TSMC
28 nm
Process Size
28 nm
7.08 billion
Transistors
8.9 billion
561 mm²
Die Size
596 mm²
Graphics Features
12 (11_0)
DirectX
12 (12_0)
4.6
OpenGL
4.6
3.0
OpenCL
2.1
1.1
Vulkan
1.2.170
3.5
CUDA
-
5.1
Shader Model
6.5
