AMD Radeon Instinct MI250X vs NVIDIA A10 PCIe
We compared two Professional market GPUs: 128GB VRAM Radeon Instinct MI250X and 24GB VRAM A10 PCIe to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
AMD Radeon Instinct MI250X 's Advantages
Released 7 months late
Boost Clock has increased by 0% (1700MHz vs 1695MHz)
More VRAM (128GB vs 24GB)
Larger VRAM bandwidth (3277GB/s vs 600.2GB/s)
4864 additional rendering cores
NVIDIA A10 PCIe 's Advantages
Lower TDP (150W vs 500W)
Score
Benchmark
FP32 (float)
Radeon Instinct MI250X
+53%
47.87 TFLOPS
A10 PCIe
31.24 TFLOPS
Graphics Card
Nov 2021
Release Date
Apr 2021
Radeon Instinct
Generation
Tesla
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
1000 MHz
Base Clock
885 MHz
1700 MHz
Boost Clock
1695 MHz
1600 MHz
Memory Clock
1563 MHz
Memory
128GB
Memory Size
24GB
HBM2e
Memory Type
GDDR6
8192bit
Memory Bus
384bit
3277GB/s
Bandwidth
600.2GB/s
Render Config
220
Compute Units
-
-
SM Count
72
14080
Shading Units
9216
880
TMUs
288
0
ROPs
96
-
Tensor Cores
288
-
RT Cores
72
16 KB (per CU)
L1 Cache
128 KB (per SM)
16 MB
L2 Cache
6 MB
-
-
-
Theoretical Performance
0 MPixel/s
Pixel Rate
162.7 GPixel/s
1496 GTexel/s
Texture Rate
488.2 GTexel/s
383.0 TFLOPS
FP16 (half)
31.24 TFLOPS
47.87 TFLOPS
FP32 (float)
31.24 TFLOPS
47.87 TFLOPS
FP64 (double)
976.3 GFLOPS
Board Design
500W
TDP
150W
900 W
Suggested PSU
450 W
No outputs
Outputs
No outputs
2x 8-pin
Power Connectors
1x 8-pin
Graphics Processor
Aldebaran
GPU Name
GA102
Aldebaran XT
GPU Variant
GA102-890-A1
CDNA 2.0
Architecture
Ampere
TSMC
Foundry
Samsung
6 nm
Process Size
8 nm
58.2 billion
Transistors
28.3 billion
Unknown
Die Size
628 mm²
Graphics Features
N/A
DirectX
12 Ultimate (12_2)
N/A
OpenGL
4.6
3.0
OpenCL
3.0
N/A
Vulkan
1.3
-
CUDA
8.6
N/A
Shader Model
6.6
