Inventec collaborates with AMD to provide Deep Learning Solutions

TAIPEI, Tai­wanMay 2, 2019– Com­pu­ter and ser­ver manu­fac­tu­rer Inven­tec Enter­pri­se Busi­ness Group (Inven­tec EBG) today announ­ced the release of its P47G4 ser­ver solu­ti­on, opti­mi­zed for AMD deep lear­ning tech­no­lo­gies.

The P47G4 ser­ver is one of four opti­mi­zed ser­ver solu­ti­ons and fea­tures a 2U, sin­gle-socket sys­tem equip­ped with AMD EPYC™ pro­ces­sors and up to four AMD Rade­on Instinct™ MI25 GPU acce­le­ra­tors. 

Inven­tec P47G4: Opti­mal, deep lear­ning sys­tem solu­ti­on fea­turing 2U sin­gle-socket AMD EPYC™ pro­ces­sor ser­ver with 4x Rade­on Instinct™ MI25 acce­le­ra­tors

The com­bi­na­ti­on of world-class ser­vers, high-per­for­mance AMD Rade­on Instinct GPU acce­le­ra­tors and the AMD ROCm open soft­ware plat­form, with its MIO­pen deep lear­ning libra­ries, pro­vi­des easy-to-deploy, pre-con­fi­gu­red solu­ti­ons for lea­ding deep lear­ning frame­works, enab­ling rese­ar­chers, sci­en­tists and data ana­lysts to acce­le­ra­te dis­co­very.

The­se tech­no­lo­gies pro­vi­de cus­to­mers with solu­ti­ons for quick deep lear­ning pro­ject deploy­ments. High levels of sys­tem and app­li­ca­ti­on inte­gra­ti­on are brought tog­e­ther to pro­vi­de easy access to the power­ful AMD Rade­on Instinct™ GPU acce­le­ra­tors for deep lear­ning on plat­forms that embrace an open envi­ron­ment.

With the AMD Rade­on Instinct™ MI25 acce­le­ra­tor, the Inven­tec P47G4 ser­ver meets the emer­ging sca­le-out deman­ds of dat­a­cen­ters. In addi­ti­on, with the incre­a­sing uti­liz­a­ti­on deman­ds of AI in high per­for­mance com­pu­ting (HPC) mar­kets, it enab­les orga­niz­a­ti­ons to find the right fit for key datacenter/HPC workloads with a low total cost of owners­hip.

The Inven­tec P47G4 is a per­fect, com­pact solu­ti­on to acce­le­ra­te machi­ne lear­ning dis­co­very on pre-con­fi­gu­red, rea­dy-to-deploy AMD Rade­on Instinct™ GPU-powe­red ser­vers with the ROCm open soft­ware plat­form and lea­ding deep lear­ning frame­works, inclu­ding Ten­sor­Flow,” said Jack Tsai, Gene­ral Mana­ger of Inven­tec Enter­pri­se Busi­ness Group.

Working clo­se­ly with AMD, Inven­tec opti­mi­zed the P47G4 ser­ver for deep lear­ning  workloads. It deli­vers up to 100 Tera­flops of FP16 com­pu­te per­for­mance at low acqui­si­ti­on cos­ts without com­pro­mi­sing the per­for­mance and fea­tures of a tra­di­tio­nal two-socket sys­tem. It pro­vi­des out­stan­ding com­pu­te, memo­ry, I/O and secu­ri­ty capa­bi­li­ties to help acce­le­ra­te mis­si­on cri­ti­cal app­li­ca­ti­ons.

About Inven­tec Data Cen­ter Solu­ti­ons (Inven­tec EBG)

Inven­tec Data Cen­ter Solu­ti­ons (Inven­tec EBG) was estab­lis­hed in 1998 and has been focu­sing on the design and manu­fac­tu­ring of ser­ver sys­tems in Inven­tec Cor­po­ra­ti­on. Over deca­des, Inven­tec EBG has been the key ser­ver sys­tem sup­plier of the glo­bal bran­ding cli­ents. 

For more infor­ma­ti­on, plea­se visit:

Fol­low “Inven­tec Data Cen­ter Solu­ti­ons” on Lin­kedIn and Wechat to recei­ve their latest news and announ­ce­ments.

Inven­tec logos are trade­marks or regis­tered trade­marks of Inven­tec Cor­po­ra­ti­on. 

AMD, the AMD Arrow logo, EPYC, Rade­on and com­bi­na­ti­ons the­re­of, are trade­marks of Advan­ced Micro Devices, Inc. Other names are for infor­ma­tio­nal pur­po­ses only and may be trade­marks of their respec­ti­ve owners.

Inven­tec EBG Media:

Vero­ni­ca Wu 
Office TEL:+886 3 390‑0000 #23948

Inven­tec EBG Con­ta­ct:

Ste­ven Lin    
Office TEL: +886–3‑3900000 ext. 23950

SOURCE Inven­tec

Related Links