Qualcomm Cloud AI 100, AMD EPYC 7003 Series Processor, and Gigabyte server solutions breaks the Peta Operations Per Second barrier for AI Inferencing

The­re is no doubt that AI is the dri­ving for­ce for next-genera­ti­on con­su­mer expe­ri­en­ces. Vir­tual­ly every expe­ri­ence on a mobi­le device somehow invol­ves AI, whe­ther it’s scrol­ling through your favo­ri­te social apps or online shop­ping that offers recom­men­da­ti­ons based on tens of thousands of AI infe­ren­ces on its own. Now what hap­pens when the­se plat­forms are ser­ving mil­li­ons of users on a given day? That real­ly requi­res racks upon racks of power­ful ser­vers that can deli­ver the AI infe­ren­cing per­for­mance requi­red to keep the­se plat­forms hum­ming along.

Today, Qual­comm Tech­no­lo­gies is enab­ling a power­ful ser­ver rack that can meet the­se high-per­for­mance requi­re­ments by pai­ring with the latest AMD EPYC 7003 Seri­es pro­ces­sors and Gigabyte’s latest G292-Z43 ser­ver solu­ti­ons. This amal­ga­ma­ti­on of hard­ware exper­ti­se offers incredi­ble per­for­mance and rai­ses the bar for the modern data cen­ter. Qual­comm Tech­no­lo­gies’ cut­ting-edge Qual­comm Cloud AI 100 fits per­fect­ly into Gigabyte’s ser­ver sys­tem and is capa­ble of dri­ving the incredi­ble AI use cases in the field of high-speed data ana­ly­sis, per­so­na­li­zed recom­men­da­ti­ons, smart cities, 5G com­mu­ni­ca­ti­ons, and more.

The Giga­byte G292-Z43 ser­ver sup­ports two AMD EPYC 7003 Seri­es pro­ces­sors for its pro­ces­sing power along with mul­ti­ple Qual­comm Cloud AI 100 cards for com­pu­ta­tio­nal­ly inten­si­ve app­li­ca­ti­ons sup­por­ting AI infe­ren­cing workloads. The Qual­comm Cloud AI 100 Infe­rence Acce­le­ra­tor boasts up to 400 TOPS with bre­akthrough performance/Watt and that’s just with one sin­gle Qual­comm Cloud AI 100 card. Now ima­gi­ne a Giga­byte ser­ver can host up to 16 Qual­comm Cloud AI 100 infe­ren­cing cards per ser­ver that, cumu­la­tively, can deli­ver up to 6.4 Peta OPS (400 TOPS x 16, one Peta OPS is 1000 TOPS). This marks the first time a Qual­comm Tech­no­lo­gies AI-based solu­ti­on brea­king the Peta­OPs bar­ri­er. And it gets even bet­ter:  A ser­ver rack can host 19 or more of the­se ser­ver units, which easi­ly exceeds 100 Peta OPS. That is a lot of Qual­comm Tech­no­lo­gies AI mus­cle. See the info­gra­phic below on how it is being configured.


To put this into a bit more con­text, a sin­gle 400TOPS HHHL Qual­comm Cloud AI 100 infe­rence card can dri­ve around 19,000 Resnet50 images/sec. That trans­la­tes to more than 6M images per second on one ser­ver rack. This kind of AI per­for­mance can surely enhan­ce, extend, and sca­le AI expe­ri­en­ces to the world. We want to thank AMD and Giga­byte for this ama­zing achievement.

Check out the­se pho­tos of the Qual­comm Cloud AI 100 cards with the AMD EPYC 7003 Seri­es pro­ces­sor-powe­red Giga­byte ser­vers rea­dy to rock and roll.



Qual­comm Cloud AI is a pro­duct of Qual­comm Tech­no­lo­gies, Inc. and/or its sub­si­dia­ries. AMD, the AMD Arrow logo, EPYC, and com­bi­na­ti­ons the­re­of are trade­marks of Advan­ced Micro Devices, Inc.

Durch die weitere Nutzung der Seite stimmst du der Verwendung von Cookies zu. Weitere Informationen

Die Cookie-Einstellungen auf dieser Website sind auf "Cookies zulassen" eingestellt, um das beste Surferlebnis zu ermöglichen. Wenn du diese Website ohne Änderung der Cookie-Einstellungen verwendest oder auf "Akzeptieren" klickst, erklärst du dich damit einverstanden.