NVIDIA L40 CNX vs NVIDIA GRID K2
We compared two Professional market GPUs: 24GB VRAM L40 CNX and 4GB VRAM GRID K2 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA L40 CNX 's Advantages
Released 9 years and 5 months late
Boost Clock2475MHz
More VRAM (24GB vs 4GB)
Larger VRAM bandwidth (864.0GB/s vs 160.0GB/s)
16640 additional rendering cores
NVIDIA GRID K2 's Advantages
Lower TDP (225W vs 300W)
Score
Benchmark
FP32 (float)
L40 CNX
+3830%
89.97 TFLOPS
GRID K2
2.289 TFLOPS
Graphics Card
Oct 2022
Release Date
May 2013
Tesla Ada
Generation
GRID
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
1005 MHz
Base Clock
-
2475 MHz
Boost Clock
-
2250 MHz
Memory Clock
1250 MHz
Memory
24GB
Memory Size
4GB
GDDR6
Memory Type
GDDR5
384bit
Memory Bus
256bit
864.0GB/s
Bandwidth
160.0GB/s
Render Config
-
-
-
142
SM Count
-
18176
Shading Units
1536
568
TMUs
128
192
ROPs
32
568
Tensor Cores
-
142
RT Cores
-
128 KB (per SM)
L1 Cache
16 KB (per SMX)
48 MB
L2 Cache
512 KB
-
-
-
Theoretical Performance
475.2 GPixel/s
Pixel Rate
23.84 GPixel/s
1406 GTexel/s
Texture Rate
95.36 GTexel/s
89.97 TFLOPS
FP16 (half)
-
89.97 TFLOPS
FP32 (float)
2.289 TFLOPS
1406 GFLOPS
FP64 (double)
95.36 GFLOPS
Board Design
300W
TDP
225W
700 W
Suggested PSU
550 W
1x HDMI 2.1
3x DisplayPort 1.4a
Outputs
No outputs
1x 16-pin
Power Connectors
1x 6-pin + 1x 8-pin
Graphics Processor
AD102
GPU Name
GK104
AD102-???-A1
GPU Variant
GK104-895-A2
Ada Lovelace
Architecture
Kepler
TSMC
Foundry
TSMC
4 nm
Process Size
28 nm
76.3 billion
Transistors
3.54 billion
608 mm²
Die Size
294 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 (11_0)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.3
Vulkan
1.2.175
8.9
CUDA
3.0
6.6
Shader Model
6.5 (5.1)
