AMD Radeon Instinct MI8 vs NVIDIA Tesla K20s
We compared two Professional market GPUs: 4GB VRAM Radeon Instinct MI8 and 5GB VRAM Tesla K20s to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
AMD Radeon Instinct MI8 's Advantages
Released 3 years and 10 months late
Larger VRAM bandwidth (512.0GB/s vs 208.0GB/s)
1600 additional rendering cores
Lower TDP (175W vs 225W)
NVIDIA Tesla K20s 's Advantages
More VRAM (5GB vs 4GB)
Score
Benchmark
FP32 (float)
Radeon Instinct MI8
+185%
8.192 TFLOPS
Tesla K20s
2.87 TFLOPS
Graphics Card
Dec 2016
Release Date
Feb 2013
Radeon Instinct
Generation
Tesla
Professional
Type
Professional
PCIe 3.0 x16
Bus Interface
PCIe 2.0 x16
Clock Speeds
-
-
-
-
-
-
500 MHz
Memory Clock
1300 MHz
Memory
4GB
Memory Size
5GB
HBM
Memory Type
GDDR5
4096bit
Memory Bus
320bit
512.0GB/s
Bandwidth
208.0GB/s
Render Config
64
Compute Units
-
-
-
-
4096
Shading Units
2496
256
TMUs
208
64
ROPs
40
-
-
-
-
-
-
16 KB (per CU)
L1 Cache
16 KB (per SMX)
2 MB
L2 Cache
1280 KB
-
-
-
Theoretical Performance
64.00 GPixel/s
Pixel Rate
29.90 GPixel/s
256.0 GTexel/s
Texture Rate
119.6 GTexel/s
8.192 TFLOPS
FP16 (half)
-
8.192 TFLOPS
FP32 (float)
2.870 TFLOPS
512.0 GFLOPS
FP64 (double)
956.8 GFLOPS
Board Design
175W
TDP
225W
450 W
Suggested PSU
550 W
No outputs
Outputs
No outputs
1x 8-pin
Power Connectors
-
Graphics Processor
Fiji
GPU Name
GK110
Fiji XT CA
(215-0862120)
GPU Variant
-
GCN 3.0
Architecture
Kepler
TSMC
Foundry
TSMC
28 nm
Process Size
28 nm
8.9 billion
Transistors
7.08 billion
596 mm²
Die Size
561 mm²
Graphics Features
12 (12_0)
DirectX
12 (11_0)
4.6
OpenGL
4.6
2.1
OpenCL
3.0
1.2.170
Vulkan
1.1
-
CUDA
3.5
6.5
Shader Model
5.1
