ENBIK SMALL AI CARD NVIDIA RTX4070-8GB/65W
ENBIK Small Edge AI computing card NVIDIA RTX4070, only 16.8 cm long and 4 cm thick for dual slot installation space, only 65W low-power, provide single-precision 15.71 TFLOPS and integer 321 TOPS computing capability.
- AI Chip : RTX4070 Ada Lovelace
- Memory : 8G GDDR6 128bit
- Computing: 15.71 TFLOPS/321 TOPS/4608 CUDA/144 Tensor/36 RT Core
- Dimension : 16.8 x 7.84 x 4 cm
- Dual-slot installation thickness
- 65W power consumption, no external PSU required: draw power directly from the PCIe slot
- The best choice for installation of mini-pc and small expansion box pc
- Category: GPU For Edge AI, PCIe GPU AI Card
Low power consumption, high efficiency, achieve the best deep learning and inference at the edge
ENBIK EGD4708P65 is an AI computing card specially designed for real edge applications. It is only 16.8cm long and 7.84 cm high, and the most amazing thing is that it is only 4 cm thick. Under such an amazing size, it can output a single-precision computing power of up to 15.71 TFLOPS, the computing power converted into integer operations is also as high as 321 TOPS, and it has 4608 CUDA and 144 Tensor Core. In such a high computing power output, only 65W of power consumption can achieve excellent performance, seamless AI multitasking and amazing visual experience, all of which take place in an amazing compact size.
EGD4708P65 can be easily installed and integrated into a small AI computing pc and effectively solve the edge AI deployment with limited power supply. It can perform API transmission learning functions through the pre-training model, and then directly realize the fine-tuning model and low-latency reasoning at the edge.
What’s more shocking is that the EGD4708P65 actually adopts the RTX4070 AI chip of NVIDIA’s latest ADA Lovelace architecture, which undoubtedly makes software developers more excited. The rich NVIDIA software eco system makes the edge AI deployment more fast and easy to implement, effectively solve the installation and low-latency reasoning of various AI models at the edge, and quickly make various decisions and judgments at the edge. Therefore, EGD4708P65 is definitely the best choice for demanding AI tasks and limited space and limited power supply deployment among various edge equipment, whether it is industrial equipment, commercial game consoles or edge AI devices.


Low power consumption, high efficiency, achieve the best deep learning and inference at the edge
ENBIK EGD4708P65 is an AI computing card specially designed for real edge applications. It is only 16.8cm long and 7.84 cm high, and the most amazing thing is that it is only 4 cm thick. Under such an amazing size, it can output a single-precision computing power of up to 15.71 TFLOPS, the computing power converted into integer operations is also as high as 321 TOPS, and it has 4608 CUDA and 144 Tensor Core. In such a high computing power output, only 65W of power consumption can achieve excellent performance, seamless AI multitasking and amazing visual experience, all of which take place in an amazing compact size.
EGD4708P65 can be easily installed and integrated into a small AI computing pc and effectively solve the edge AI deployment with limited power supply. It can perform API transmission learning functions through the pre-training model, and then directly realize the fine-tuning model and low-latency reasoning at the edge.
What’s more shocking is that the EGD4708P65 actually adopts the RTX4070 AI chip of NVIDIA’s latest ADA Lovelace architecture, which undoubtedly makes software developers more excited. The rich NVIDIA software eco system makes the edge AI deployment more fast and easy to implement, effectively solve the installation and low-latency reasoning of various AI models at the edge, and quickly make various decisions and judgments at the edge. Therefore, EGD4708P65 is definitely the best choice for demanding AI tasks and limited space and limited power supply deployment among various edge equipment, whether it is industrial equipment, commercial game consoles or edge AI devices.

