Home NVIDIA L20

NVIDIA L20

This is a GPU manufactured with TSMC 5nm process, based on Nvidia Ada Lovelace architecture and released on Nov 2023. It features 76.3 billion transistors, 11776 CUDA cores and 48GB GDDR6 memory, with 96MB L2 cache, theoretical performance of 59.35TFLOPS, with total power consumption of 275W.

Graphics Card

[Report Issues]
Release Date
Nov 2023
Generation
Tesla Ada
Type
Desktop
Bus Interface
PCIe 4.0 x16

Clock Speeds

[Report Issues]
Base Clock
1440MHz
Boost Clock
2520MHz
Memory Clock
2250MHz

Memory

[Report Issues]
Memory Size
48GB
Memory Type
GDDR6
Memory Bus
384bit
Bandwidth
864.0 GB/s

Render Config

[Report Issues]
SM Count
92
Shading Units
11776
TMUs
368
ROPs
128
Tensor Cores
368
RT Cores
92
L1 Cache
128 KB (per SM)
L2 Cache
96MB

Theoretical Performance

[Report Issues]
Pixel Rate
322.6 GPixel/s
Texture Rate
927.4 GTexel/s
FP16 (half)
59.35 TFLOPS
FP32 (float)
59.35 TFLOPS
FP64 (double)
927.4 GFLOPS

Graphics Processor

[Report Issues]
GPU Name
AD102
GPU Variant
AD102
Architecture
Ada Lovelace
Foundry
TSMC
Process Size
5 nm
Transistors
76,300 million
Die Size
609mm²

Board Design

[Report Issues]
TDP
275W
Suggested PSU
600W
Outputs
4x DisplayPort 1.4a
Power Connectors
1x 16-pin

Graphics Features

[Report Issues]
DirectX
12 Ultimate (12_2)
OpenGL
4.6
OpenCL
3.0
Vulkan
1.3
CUDA
8.9
Shader Model
6.7

Rankings

[Report Issues]
FP32 (float)
61.39 TFLOPS
61.32 TFLOPS
NVIDIA L20 48 GB GDDR6
59.35 TFLOPS
NVIDIA H800 SXM5 80 GB HBM3
59.3 TFLOPS
NVIDIA H100 CNX 80 GB HBM2e
53.84 TFLOPS
52.43 TFLOPS
52.43 TFLOPS

Related Comparisons

© 2024 - TopCPU.net   Contact Us Privacy Policy