NVIDIA L20 vs NVIDIA CMP 50HX
We compared two Desktop platform GPUs: 48GB VRAM L20 and 10GB VRAM CMP 50HX to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA L20 's Advantages
Released 2 years and 5 months late
Boost Clock has increased by 63% (2520MHz vs 1545MHz)
More VRAM (48GB vs 10GB)
Larger VRAM bandwidth (864.0GB/s vs 560.0GB/s)
8192 additional rendering cores
NVIDIA CMP 50HX 's Advantages
Lower TDP (250W vs 275W)
Score
Benchmark
FP32 (float)
L20
+436%
59.35 TFLOPS
CMP 50HX
11.07 TFLOPS
Graphics Card
Nov 2023
Release Date
Jun 2021
Tesla Ada
Generation
Mining GPUs
Desktop
Type
Desktop
PCIe 4.0 x16
Bus Interface
PCIe 1.0 x4
Clock Speeds
1440 MHz
Base Clock
1350 MHz
2520 MHz
Boost Clock
1545 MHz
2250 MHz
Memory Clock
1750 MHz
Memory
48GB
Memory Size
10GB
GDDR6
Memory Type
GDDR6
384bit
Memory Bus
320bit
864.0GB/s
Bandwidth
560.0GB/s
Render Config
-
-
-
92
SM Count
56
11776
Shading Units
3584
368
TMUs
192
128
ROPs
80
368
Tensor Cores
448
92
RT Cores
56
128 KB (per SM)
L1 Cache
64 KB (per SM)
96 MB
L2 Cache
5 MB
-
-
-
Theoretical Performance
322.6 GPixel/s
Pixel Rate
123.6 GPixel/s
927.4 GTexel/s
Texture Rate
296.6 GTexel/s
59.35 TFLOPS
FP16 (half)
22.15 TFLOPS
59.35 TFLOPS
FP32 (float)
11.07 TFLOPS
927.4 GFLOPS
FP64 (double)
346.1 GFLOPS
Board Design
275W
TDP
250W
600 W
Suggested PSU
600 W
4x DisplayPort 1.4a
Outputs
No outputs
1x 16-pin
Power Connectors
2x 8-pin
Graphics Processor
AD102
GPU Name
TU102
AD102
GPU Variant
TU102-100-A1
Ada Lovelace
Architecture
Turing
TSMC
Foundry
TSMC
5 nm
Process Size
12 nm
76.3 billion
Transistors
18.6 billion
609 mm²
Die Size
754 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
7.5
6.7
Shader Model
6.8
