Connect with us

GMS Velocity 2U “4×4” server configuration includes four 28-core Intel Scalable Xeon CPUs and four Nvidia A100 GPGPUs

Published

on

The American company General Micro Systems (GMS), specializing in the production of secure servers, has presented the Velocity 2U “4×4” server for artificial intelligence applications. According to the manufacturer, the server is intended for use “on the battlefield”: in command posts, tactical operations centers and protected platforms such as autonomous vehicles, as well as in on-board surveillance, data mining and in other cases when a combination of high performance, small size and air cooling. The server is designed as a 2U rackmount unit.

Its configuration includes four 28-core Intel Scalable Xeon CPUs for a total of 112 cores, and four Nvidia A100 GPGPUs that are included in PCIe Gen 3 x16 slots. The DDR4 ECC RAM is 4TB. The server includes a storage subsystem that can contain up to 22 SSDs with a total capacity of up to 264 TB with NVMe and RAID support. There are 16 ports of 100 GbE for network connection. To connect custom I / O modules, 144 PCIe Gen 3 lanes are available, which are brought out into 8 slots. The server uses hot-plug power supply redundancy. The server is designed to operate at temperatures from 0 ° C to +40 ° C (optionally from -20 ° C to +75 ° C).

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

News

TSMC presented the N4P process technology

Published

on

TSMC introduced the N4P process, a performance-optimized enhancement to the 5nm technology platform. New process technology added to N5, N4, N3. According to the manufacturer, the choice between them gives customers the opportunity to choose the most suitable combination of power consumption, performance, die area and product cost.

The new process technology is the third major improvement on TSMC’s 5nm platform. The company estimates that N4P delivers an 11% increase in productivity over the original N5 technology and 6% over N4. Compared to N5, N4P will also provide 22% more energy efficiency and 6% more transistor density. In addition, N4P reduces process complexity and shorter insert cycle times by reducing the number of masks.

N4P projects are said to be well supported by TSMC’s comprehensive design ecosystem. The first products based on N4P technology are expected to go into production by the second half of 2022.

Continue Reading

News

Xturismo Flying Motorcycle Orders Started in Japan

Published

on

The Japanese company ALI Technologies has presented the “flying motorcycle” Xturismo, capable of speeds up to 100 km / h and stay in the air for up to 40 minutes. During the demonstration, the device rose several meters above the ground and flew for about one and a half minutes. The weight of the device, designed to carry one person, is 300 kg.

ALI Technologies, backed by Kyocera and Mitsui Sumitomo Insurance Venture Capital, has already started accepting orders and plans to begin shipping the first batch of 200 pieces in the first half of next year. The Xturismo is priced at approximately $ 682,000.

Xturismo Flying Motorcycle Orders Started in Japan

Interestingly, in the Xturismo, the main propulsion system does not use electric motors, but an internal combustion engine that drives the two main propellers. Auxiliary propellers help maintain balance and perform maneuvers. To do this, the “motorcycle” uses technology tested by ALI in drones.

One of the applications of the apparatus is called rescue operations on the water. True, it is not clear how exactly the assistance is supposed to be provided – neither the carrying capacity, nor the cruising range, nor the lack of the possibility of landing on water contribute to such an application, not to mention the fact that for aircraft operated over water, it is mandatory to reserve engines with the ability to continue flight if one of them fails.

Continue Reading

News

NASA: You Can Help Teach Rovers To Explore Mars Better

Published

on

The US Aeronautics and Space Administration NASA has published a note with the intriguing headline “You Can Help Teach NASA Mars Rovers to Explore Mars Better.”

As the department explains, artificial intelligence has tremendous potential to change the way spacecraft study the universe. But since all machine learning algorithms require human training, NASA is asking members of the public to help identify elements of scientific interest in images taken by the Perseverance rover.

The project, called AI4Mars, is a follow-up to a project launched last year based on imagery from the Curiosity rover.

NASA: You Can Help Teach Rovers To Explore Mars Better

Early contributors to this project mapped out nearly half a million images, using a tool to highlight objects such as sand and rock that NASA’s Jet Propulsion Laboratory (JPL) Mars rover drivers typically look at when planning routes on the Red Planet. The end result was an algorithm called SPOC (Classification of Soil and Object Properties) that could correctly identify these features in almost 98% of cases.

The SPOC algorithm is still in development, and NASA hopes that one day it can be sent to Mars aboard a spacecraft of the future that can do more autonomous driving than Perseverance’s AutoNav technology allows.

AI4Mars now provides additional markers to indicate more precise details, allowing people to choose options such as moving rocks or nodules (globular mineral formations).

The goal is to hone an algorithm that could help a future rover pick “needles out of a haystack” of data sent from Mars.

Equipped with 19 cameras, Perseverance sends tens to hundreds of images to Earth every day for scientists and engineers to view specific geologic features. But time is running out: after these images have traveled millions of kilometers from Mars to Earth, NASA team members have hours to develop the next set of instructions, based on what they see in these images, to send to Perseverance.

Continue Reading

Most Popular