Intel Data Center GPU Max Subsystem vs NVIDIA L40S
We compared two Professional market GPUs: 128GB VRAM Data Center GPU Max Subsystem and 48GB VRAM L40S to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Intel Data Center GPU Max Subsystem 's Advantages
More VRAM (128GB vs 48GB)
Larger VRAM bandwidth (3205GB/s vs 864.0GB/s)
NVIDIA L40S 's Advantages
Boost Clock has increased by 58% (2520MHz vs 1600MHz)
1792 additional rendering cores
Lower TDP (300W vs 2400W)
Score
Benchmark
FP32 (float)
Data Center GPU Max Subsystem
52.43 TFLOPS
L40S
+74%
91.61 TFLOPS
Graphics Card
Jan 2023
Release Date
Oct 2022
Data Center GPU
Generation
Tesla Ada
Professional
Type
Professional
PCIe 5.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
900 MHz
Base Clock
1110 MHz
1600 MHz
Boost Clock
2520 MHz
1565 MHz
Memory Clock
2250 MHz
Memory
128GB
Memory Size
48GB
HBM2e
Memory Type
GDDR6
8192bit
Memory Bus
384bit
3205GB/s
Bandwidth
864.0GB/s
Render Config
-
-
-
-
SM Count
142
16384
Shading Units
18176
1024
TMUs
568
0
ROPs
192
1024
Tensor Cores
568
128
RT Cores
142
64 KB (per EU)
L1 Cache
128 KB (per SM)
408 MB
L2 Cache
48 MB
-
-
-
Theoretical Performance
0 MPixel/s
Pixel Rate
483.8 GPixel/s
1638 GTexel/s
Texture Rate
1431 GTexel/s
52.43 TFLOPS
FP16 (half)
91.61 TFLOPS
52.43 TFLOPS
FP32 (float)
91.61 TFLOPS
52.43 TFLOPS
FP64 (double)
1431 GFLOPS
Board Design
2400W
TDP
300W
2800 W
Suggested PSU
700 W
No outputs
Outputs
1x HDMI 2.1
3x DisplayPort 1.4a
1x 16-pin
Power Connectors
1x 16-pin
Graphics Processor
Ponte Vecchio
GPU Name
AD102
-
GPU Variant
AD102-???-A1
Generation 12.5
Architecture
Ada Lovelace
Intel
Foundry
TSMC
10 nm
Process Size
5 nm
100 billion
Transistors
76.3 billion
1280 mm²
Die Size
609 mm²
Graphics Features
12 (12_1)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
N/A
Vulkan
1.3
-
CUDA
8.9
6.6
Shader Model
6.7
