NVIDIA L40S vs NVIDIA A10G
We compared two Professional market GPUs: 48GB VRAM L40S and 12GB VRAM A10G to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA L40S 's Advantages
Released 1 years and 6 months late
Boost Clock has increased by 47% (2520MHz vs 1710MHz)
More VRAM (48GB vs 12GB)
Larger VRAM bandwidth (864.0GB/s vs 600.2GB/s)
8960 additional rendering cores
NVIDIA A10G 's Advantages
Lower TDP (150W vs 300W)
Score
Benchmark
FP32 (float)
L40S
+190%
91.61 TFLOPS
A10G
31.52 TFLOPS
Blender
L40S
+157%
9239
A10G
3589
Graphics Card
Oct 2022
Release Date
Apr 2021
Tesla Ada
Generation
Tesla
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
1110 MHz
Base Clock
1320 MHz
2520 MHz
Boost Clock
1710 MHz
2250 MHz
Memory Clock
1563 MHz
Memory
48GB
Memory Size
12GB
GDDR6
Memory Type
GDDR6
384bit
Memory Bus
384bit
864.0GB/s
Bandwidth
600.2GB/s
Render Config
-
-
-
142
SM Count
72
18176
Shading Units
9216
568
TMUs
288
192
ROPs
96
568
Tensor Cores
288
142
RT Cores
72
128 KB (per SM)
L1 Cache
128 KB (per SM)
48 MB
L2 Cache
6 MB
-
-
-
Theoretical Performance
483.8 GPixel/s
Pixel Rate
164.2 GPixel/s
1431 GTexel/s
Texture Rate
492.5 GTexel/s
91.61 TFLOPS
FP16 (half)
31.52 TFLOPS
91.61 TFLOPS
FP32 (float)
31.52 TFLOPS
1431 GFLOPS
FP64 (double)
985.0 GFLOPS
Board Design
300W
TDP
150W
700 W
Suggested PSU
450 W
1x HDMI 2.1
3x DisplayPort 1.4a
Outputs
No outputs
1x 16-pin
Power Connectors
8-pin EPS
Graphics Processor
AD102
GPU Name
GA102
AD102-???-A1
GPU Variant
GA102-890-A1
Ada Lovelace
Architecture
Ampere
TSMC
Foundry
Samsung
5 nm
Process Size
8 nm
76.3 billion
Transistors
28.3 billion
609 mm²
Die Size
628 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.3
Vulkan
1.3
8.9
CUDA
8.6
6.7
Shader Model
6.6
