Radeon Instinct MI250 vs Tesla K40s
We compared two Professional market GPUs: 128GB VRAM Radeon Instinct MI250 and 12GB VRAM Tesla K40s to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Radeon Instinct MI250 Advantages
Released 8 years late
Boost Clock has increased by 94% (1700MHz vs 876MHz)
More VRAM (128GB vs 12GB)
Larger VRAM bandwidth (3277GB/s vs 288.4GB/s)
10432 additional rendering cores
Tesla K40s Advantages
Lower TDP (245W vs 500W)
Score
Benchmark
FP32 (float)
Radeon Instinct MI250
+796%
45260
Tesla K40s
5046
Graphics Card
Nov 2021
Release Date
Nov 2013
Radeon Instinct
Generation
Tesla
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
1000 MHz
Base Clock
745 MHz
1700 MHz
Boost Clock
876 MHz
1600 MHz
Memory Clock
1502 MHz
Memory
128GB
Memory Size
12GB
HBM2e
Memory Type
GDDR5
8192bit
Memory Bus
384bit
3277GB/s
Bandwidth
288.4GB/s
Render Config
208
Compute Units
-
13312
Shading Units
2880
832
TMUs
240
0
ROPs
48
16 KB (per CU)
L1 Cache
16 KB (per SMX)
16 MB
L2 Cache
1536 KB
Theoretical Performance
0 MPixel/s
Pixel Rate
52.56 GPixel/s
1414 GTexel/s
Texture Rate
210.2 GTexel/s
362.1 TFLOPS
FP16 (half)
-
45.26 TFLOPS
FP32 (float)
5.046 TFLOPS
45.26 TFLOPS
FP64 (double)
1.682 TFLOPS
Board Design
500W
TDP
245W
900 W
Suggested PSU
550 W
No outputs
Outputs
No outputs
2x 8-pin
Power Connectors
-
Graphics Processor
Aldebaran
GPU Name
GK110B
Aldebaran
GPU Variant
-
CDNA 2.0
Architecture
Kepler
TSMC
Foundry
TSMC
6 nm
Process Size
28 nm
58.2 billion
Transistors
7.08 billion
Unknown
Die Size
561 mm²
Graphics Features
N/A
DirectX
12 (11_1)
N/A
OpenGL
4.6
3.0
OpenCL
3.0
N/A
Vulkan
1.1
-
CUDA
3.5
N/A
Shader Model
5.1
