Simple and Free.
Just Plug It in.
Max.(0.7W)
16.8TOPS¹
Fast Connect
USB 3.0
per Watt
24.0TOPS
¹ Trillion Operations per Second.
Lightspeeur 2803S
Neural Processor Unit.
Use larger networks via ResNet-101, 152.
GTI's NPU architecture optimises. acclerates data flow
and shows astonishing performance.
Faster inference ability than
Apple A16(15TOPS) or Snapdragon 865(16TOPS).
Lightspeeur 2803S
Edge v1
Compared to Google Edge v1 TPU,
Up to 12x faster compute ability per watt.
Optimised for Computer-vision.
Farewell to
Old-fashioned Device.
Best for Image Recognition, Object Detection, NLP, Visual Analysis¹.
¹ Usage Examples of Gyrfalcon Technology.
Funny How Small is.
Although it has a small size of 50*20mm, can run various frameworks.
Recommended for use with Ubuntu OS.
We will provide
SDK, MDK, and Example Codes.
Related materials will be sent to the email
you registered with VORAMALL or in a separate message.
The photo is an example.
VORA USB 2803
SPEC
VORA 2803
Form Factor
USB 3.0 (W50(63) x H20 x D8, mm)
Processor
-
NPU
Gyrfalcon Inc. Lightspeeur® 2803S,
16.8TOPS @ 0.7W
Memory
-
GPU
-
Storage
-
Image,
Pixel I/O
Image : 448*448, 224*224, 112*112,
56*56, 28*28, 14*14 pixel
Pixel/Layer : up to 4,014,080
16, 32, 64, 128, 256, 512, 1024 ch.
I/O Ports
1x USB 3.0
Environment
Above Python 3.8
Above OpenCV 4.2
GNU
Network
-
Power Supply
USB
Frameworks
TensorFlow, PyTorch(Caffe),
VGG16, ResNet 18/50, MobileNet
Others
-
OS
-
| SPEC | VORA 2803 |
|---|---|
| Form Factor | USB 3.0 (W50(63) x H20 x D8, mm) |
| Processor | - |
| NPU | Gyrfalcon Inc. Lightspeeur® 2803S, 16.8TOPS @ 0.7W |
| Memory | - |
| GPU | - |
| Storage | - |
| Image, Pixel I/O |
Image : 448*448, 224*224, 112*112, 56*56, 28*28, 14*14 pixel Pixel/Layer : up to 4,014,080 16, 32, 64, 128, 256, 512, 1024 ch. |
| I/O Ports | 1x USB 3.0 |
| Environment | Above Python 3.8 Above OpenCV 4.2 GNU |
| Network | - |
| Power Supply | USB |
| Frameworks | TensorFlow, PyTorch(Caffe), VGG16, ResNet 18/50, MobileNet |
| Others | - |
| OS | - |