NVIDIA GRID K2 vs NVIDIA L40S
We compared two Professional market GPUs: 4GB VRAM GRID K2 and 48GB VRAM L40S to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA GRID K2 's Advantages
Lower TDP (225W vs 300W)
NVIDIA L40S 's Advantages
Released 9 years and 5 months late
Boost Clock2520MHz
More VRAM (48GB vs 4GB)
Larger VRAM bandwidth (864.0GB/s vs 160.0GB/s)
16640 additional rendering cores
Score
Benchmark
FP32 (float)
GRID K2
2.289 TFLOPS
L40S
+3902%
91.61 TFLOPS
Graphics Card
May 2013
Release Date
Oct 2022
GRID
Generation
Tesla Ada
Professional
Type
Professional
PCIe 3.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
-
Base Clock
1110 MHz
-
Boost Clock
2520 MHz
1250 MHz
Memory Clock
2250 MHz
Memory
4GB
Memory Size
48GB
GDDR5
Memory Type
GDDR6
256bit
Memory Bus
384bit
160.0GB/s
Bandwidth
864.0GB/s
Render Config
-
-
-
-
SM Count
142
1536
Shading Units
18176
128
TMUs
568
32
ROPs
192
-
Tensor Cores
568
-
RT Cores
142
16 KB (per SMX)
L1 Cache
128 KB (per SM)
512 KB
L2 Cache
48 MB
-
-
-
Theoretical Performance
23.84 GPixel/s
Pixel Rate
483.8 GPixel/s
95.36 GTexel/s
Texture Rate
1431 GTexel/s
-
FP16 (half)
91.61 TFLOPS
2.289 TFLOPS
FP32 (float)
91.61 TFLOPS
95.36 GFLOPS
FP64 (double)
1431 GFLOPS
Board Design
225W
TDP
300W
550 W
Suggested PSU
700 W
No outputs
Outputs
1x HDMI 2.1
3x DisplayPort 1.4a
1x 6-pin + 1x 8-pin
Power Connectors
1x 16-pin
Graphics Processor
GK104
GPU Name
AD102
GK104-895-A2
GPU Variant
AD102-???-A1
Kepler
Architecture
Ada Lovelace
TSMC
Foundry
TSMC
28 nm
Process Size
5 nm
3.54 billion
Transistors
76.3 billion
294 mm²
Die Size
609 mm²
Graphics Features
12 (11_0)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.2.175
Vulkan
1.3
3.0
CUDA
8.9
6.5 (5.1)
Shader Model
6.7
