Sunday, July 5th 2020

Eurocom launches MXM3 to PCIe Adapter for GPU Developers

Embedded, GPU-accelerated computing is the use of a graphics processing unit (GPU) combined with a central processing unit (CPU) to facilitate advanced computing processes such as Artificial Intelligence of Things (AIoT), Deep Learning, Prototyping/Pre-concept design, CUDA development and medical Hi-Res imaging, to name a few.

Embedded, GPU-accelerated computing is gaining popularity as its potential becomes realized in many high-tech labs and R&D facilities around the world. GPU-accelerated computing plays a huge role in accelerating applications in platforms within these bleeding-edge industries. Whether it is designing self-driving cars, rendering ultra-HD scenes with millions of polygons (4K and above), or working with advanced CAD models for GIS mapping, a multi-GPU system will significantly increase the performance and capabilities advanced developers have available.
Through embedded GPU-accelerated technology, processing and computing times can be cut into a fraction of the time of what could be accomplished compared to traditional, less-sophisticated computing infrastructures. The combination of high-performance computing along with lower power consumption can result in a more significant and more extensive deep learning in neural networks.
Who can benefit from GPU development?
Many fields/industries that can greatly benefit from the utilization of GPU development in their pre-conceptual phases are: research and AI analytics through Deep Learning, Artificial Intelligence of Things (such as development of self-driving cars), VR/3D Development, Military and many other cutting-edge industries that want to speed up the pre-conceptual process and save time to advance to the prototyping process as quickly as possible.

Deep Learning and Artificial Intelligence of Things (AIoT)
Advanced modeling and analysis are always essential for the development of safe and reliable systems. The codes and algorithms can be parallelized using multiple GPUs through multi-threading and multi-streaming in order to reduce this process' high computational cost. In the course of the optimization of the algorithm, the computational performance is improved drastically, thus boosting and accelerating large-scale simulations.

MXM3 to PCIe Adapter from Eurocom
The EUROCOM MXM3 to PCIe Adapter is specifically designed for use with MXM3 AMD or NVIDIA graphics cards via PCI Express 3.0 (x16) bus interface. This allows developers to use the PCI Express system and boost compute performance of the MXM module in the early conceptual stages of their projects.

Using the thousands of computing cores inside the most powerful GPUs today, Eurocom's embedded MXM development kits are able to process massive amounts of video/image data in real time provides a high level of energy efficiency/TDP that is crucial for high density computing and analysis.
  • The EUROCOM MXM3 to PCIe Adapter allows for more flexibility and higher energy efficiency as MXM3 Modules can be used in a variety of computing platforms such as desktops, industrial PC's, military computers and other cutting-edge computer infrastructures.
  • The EUROCOM MXM3 to PCIe Adapter features an 8-pin power connector with 4 HDMI ports and a 2-pin fan connector. The advantage of low TDP without sacrificing compute power/performance makes these MXM3 to PCIe kits ideal for use in small form factor PC's.
For more information, visit the product page.
Add your own comment

4 Comments on Eurocom launches MXM3 to PCIe Adapter for GPU Developers

#1
FreedomEclipse
~Technological Technocrat~
I can see these being something great on the Chinese market.... Sellers on Sites like AliExpress repurposing/recycling MXM cards off gaming laptops and selling them to people who wants to buy them for cheap.

Since it's a mobile part. It probably won't cost a lot of money to machine an aluminium heatsink for it as the TDP is lower.
Posted on Reply
#2
Haile Selassie
MXM gpus are usualy twice as expensive compared to their desktop counterpart, regardless of generation. New or old.
Posted on Reply
#3
ThisMayBeYou
699$ for a pcb with some connectors? No thanks, I will wait till Chinese engineers start to reproduce those.
Posted on Reply
#4
Jism
Would be cool for a OC contest.
Posted on Reply