ADLINK has introduced the pocket-sized and portable NVIDIA RTX A500 GPU that replaces the iGPU in your system for AI acceleration.
ADLINK Pocket AI Targets Putting Graphics Computing in Your Hands with the Power of NVIDIA RTX GPUs
Imagine carrying an AI performance GPU in your pocket. As silly as this may sound, ADLINK has done it with its Pocket AI mobile GPU featuring NVIDIA GPU with Tensor-Core technology.
The concept behind this technology is interesting. It aims to bring AI and graphics capabilities to any device in a “plug and play” manner. ADLINK Pocket AI is specially designed for professionals with AI tasks that require a dedicated GPU. To highlight the potential of this innovative technology, Storage review He uploaded a comprehensive review covering every aspect in detail.
Starting with the integrated GPU, Pocket AI features the NVIDIA RTX A500 GPU, which features 4GB of GDDR6 memory along with 2048 CUDA cores and 64 Tensor cores running in 64-bit interface. It comes with a computing power of 6.54 TFLOPS Peak FP32 with a TDP of 25W. On paper, the GPU has lower specs than the RTX 3050, but if you’re running an older machine with an iGPU it could make for a nice upgrade, especially if you want to focus on AI-intensive tasks.
In terms of design details, Pocket AI is equipped with a rubber sleeve that helps hold the GPU and protects users from potential burns since the NVIDIA RTX A500 is running very hot. The unit is a nice black and green color with two ports on board. The first is Power Delivery, and the second is Thunderbolt 3, which connects the GPU to your device. The Pocket AI’s cooling mechanism is simple, featuring an intake fan at the bottom with wide cutouts to keep airflow circulating.
It all comes down to how Pocket AI performs in real life applications. StorageReview has benchmarked the GPU performance of the Luxmark & UL Labs Procyon AI, and its performance numbers show more than twice that of an iGPU. Here is a look at the numbers obtained below:
|standard||Iris XE 96 core||Adlink Pocket AI|
|Procyon AI inference||59||264||651|
Benchmarks reveal that while Pocket AI doesn’t have massive raw power, it is suitable for AI industry workloads. The product is ideal for situations where it is preferable to continue using an existing device rather than upgrading to a new device. If you want to add enhanced graphics computing capabilities to your device, Pocket AI is the way to go. However, the ability to pack the power of Pocket AI with portability comes at a cost.
The Pocket AI is currently priced at $429, which is steep, judging by the components on board. To give you an idea, the NVIDIA RTX 4060 Ti is currently selling for a similar price, and so given that, I think Pocket AI isn’t priced here. But this is the case with almost all external GPU options and modules. The GPD G1 that comes with an AMD RX 7600M XT GPU costs around $800 which is the same price as the RX 7900 XT enthusiast card.
Furthermore, while with Pocket AI you might gain portability and computing power, in the grand scheme of things most consumers won’t have the problem of carrying around a GPU every day; Instead, they will prefer a more permanent solution which means getting a better device. On top of that, concerns about Pocket AI temperatures, which obviously occur in such devices, have also been reported.
Overall, Pocket AI seems to be an innovation that comes with a lot of problems. Although the trend is here, ADLINK’s Pocket AI cannot be a hit unless it comes with a better price along with improvements within the heat dissipation department.
source: Storage review