NVIDIA A10 PCIe vs AMD Radeon Instinct MI6
We compared two Professional market GPUs: 24GB VRAM A10 PCIe and 8GB VRAM Radeon Instinct MI6 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA A10 PCIe 's Advantages
Released 4 years and 4 months late
Boost Clock has increased by 37% (1695MHz vs 1233MHz)
More VRAM (24GB vs 8GB)
Larger VRAM bandwidth (600.2GB/s vs 224.0GB/s)
6912 additional rendering cores
Score
Benchmark
FP32 (float)
A10 PCIe
+449%
31.24 TFLOPS
Radeon Instinct MI6
5.682 TFLOPS
Graphics Card
Apr 2021
Release Date
Dec 2016
Tesla
Generation
Radeon Instinct
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
885 MHz
Base Clock
1120 MHz
1695 MHz
Boost Clock
1233 MHz
1563 MHz
Memory Clock
1750 MHz
Memory
24GB
Memory Size
8GB
GDDR6
Memory Type
GDDR5
384bit
Memory Bus
256bit
600.2GB/s
Bandwidth
224.0GB/s
Render Config
-
Compute Units
36
72
SM Count
-
9216
Shading Units
2304
288
TMUs
144
96
ROPs
32
288
Tensor Cores
-
72
RT Cores
-
128 KB (per SM)
L1 Cache
16 KB (per CU)
6 MB
L2 Cache
2 MB
-
-
-
Theoretical Performance
162.7 GPixel/s
Pixel Rate
39.46 GPixel/s
488.2 GTexel/s
Texture Rate
177.6 GTexel/s
31.24 TFLOPS
FP16 (half)
5.682 TFLOPS
31.24 TFLOPS
FP32 (float)
5.682 TFLOPS
976.3 GFLOPS
FP64 (double)
355.1 GFLOPS
Board Design
150W
TDP
150W
450 W
Suggested PSU
450 W
No outputs
Outputs
No outputs
1x 8-pin
Power Connectors
1x 6-pin
Graphics Processor
GA102
GPU Name
Ellesmere
GA102-890-A1
GPU Variant
Ellesmere XT
Ampere
Architecture
GCN 4.0
Samsung
Foundry
GlobalFoundries
8 nm
Process Size
14 nm
28.3 billion
Transistors
5.7 billion
628 mm²
Die Size
232 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 (12_0)
4.6
OpenGL
4.6
3.0
OpenCL
2.1
1.3
Vulkan
1.2
8.6
CUDA
-
6.6
Shader Model
6.4
