AMD Enables Server APU Software to Reimagine the Server

Show­ca­ses new gene­ra­ti­on of deve­lo­p­ment tools that will help ser­vers adapt to modern data cen­ter workloads at APU13 Deve­lo­per Summit



SUNNYVALE, Calif. —11/11/2013 

AMD (NYSE: AMD) today announ­ced it is enab­ling its acce­le­ra­ted pro­ces­sing units (APU) for next-gene­ra­ti­on ser­vers through important advance­ments in soft­ware tools deve­lo­ped by AMD and in col­la­bo­ra­ti­on with tech­no­lo­gy part­ners and the open source com­mu­ni­ty. In his key­note address at APU13, AMD Cor­po­ra­te Fel­low Phil Rogers high­ligh­ted the signi­fi­cant pro­gress AMD has made in both deve­lo­ping soft­ware intern­al­ly and empowe­ring others to deve­lop soft­ware to take advan­ta­ge of the capa­bi­li­ties of AMD APU tech­no­lo­gy, which com­bi­nes indus­try lea­ding AMD Rade­on™ gra­phics pro­ces­sing engi­nes with x86 com­pu­ta­tio­nal pro­ces­sing on a sin­gle chip.

Ser­vers must be effi­ci­ent, sca­lable and adap­ta­ble to meet the com­pu­te cha­rac­te­ristics of new and chan­ging workloads. Soft­ware appli­ca­ti­ons that levera­ge ser­ver APUs are desi­gned to dri­ve high­ly effi­ci­ent, low-power, den­se ser­ver solu­ti­ons opti­mi­zed for high­ly par­al­lel and mul­ti­me­dia workloads,” said Suresh Gopa­l­a­krish­n­an, cor­po­ra­te vice pre­si­dent and gene­ral mana­ger of the Ser­ver Busi­ness Unit, AMD. “We have evol­ved our pro­ces­sor road­map to sup­port this oppor­tu­ni­ty, and now we are show­ca­sing how the APU soft­ware eco­sys­tem is gai­ning momen­tum and what deve­lo­pers can do to participate.”

With the rea­liza­ti­on that ser­ver APUs based on Hete­ro­ge­neous Sys­tem Archi­tec­tu­re (HSA) are coming to mar­ket soon, AMD has deve­lo­ped tools for soft­ware deve­lo­pers to take advan­ta­ge of the bene­fits that HSA pro­vi­des. HSA enables the CPU and GPU to work in harm­o­ny on a sin­gle pie­ce of sili­con, seam­less­ly moving the right tasks to the best-sui­ted pro­ces­sing ele­ment with no data trans­fer pen­al­ties and makes more memo­ry available to the GPU so that com­plex pro­ces­sing tasks can fit in a sin­gle node.

Com­bi­ning AMD’s upco­ming HSA com­pli­ant APUs with new east-west fabric archi­tec­tures, like AMD’s Free­dom Fabric and HP’s Moonshot, is a poten­ti­al game chan­ger for many ser­ver-side HPC and big data ana­ly­tics workloads,” said Paul Teich, CTO and seni­or ana­lyst at Moor Insights & Stra­tegy, a lea­ding high-tech ana­lyst firm.
AMD is col­la­bo­ra­ting with its tech­no­lo­gy part­ners and the open source com­mu­ni­ty to pro­vi­de deve­lo­pers with tools that enable them to build ser­ver appli­ca­ti­ons that uti­li­ze both CPU and GPU com­pu­te capa­bi­li­ties available in its revo­lu­tio­na­ry HSA based ser­ver APUs. Tools high­ligh­ted today at APU13 include:

  • Pro­ject Suma­tra  – a joint Ora­cle and AMD pro­ject done in open source that enables deve­lo­pers to code in Java and take advan­ta­ge of GPU compute;
  • GCC/HSA Pro­ject – an AMD and SUSE pro­ject to enable the popu­lar open source Linux com­pi­ler, GCC, to sup­port HSA, tar­ge­ting OpenMP APIs;
  • PGI Acce­le­ra­tor™ Com­pi­ler – a beta ver­si­on is available that enables deve­lo­pers to add Open­ACC direc­ti­ves that sup­port AMD APUs and dis­crete GPUs to Win­dows and Linux For­tran, C and C++ programs;
  • clMathAMD Open­CL math libra­ri­es that were con­tri­bu­ted to open source in August enable deve­lo­pers to acce­le­ra­te com­mon sci­en­ti­fic and engi­nee­ring com­pu­ta­ti­ons on AMD APUs and dis­crete GPUs;
  • Array­Fi­re 2.0 for Open­CL – a fast math libra­ry by Acce­le­rEyes that uti­li­zes clMath for GPU com­pu­ting and offers an easy-to-use API for Win­dows or Linux developers;
  • CodeXL 1.3AMD’s com­pre­hen­si­ve deve­lo­per tool suite for Win­dows and Linux that fea­tures remo­te debug­ging and pro­fil­ing to enable ser­ver appli­ca­ti­on developers.

Seve­ral AMD tech­no­lo­gy part­ners who are enab­ling the ser­ver APU eco­sys­tem are par­ti­ci­pa­ting at APU13, which con­ti­nues through Wed­nes­day, Nov. 13. The APU13 agen­da includes a key­note by Nan­di­ni Rama­ni, Ora­cle, who will over­view the Pro­ject Suma­tra and the bene­fits that GPU com­pu­ting brings to Java. In addi­ti­on, ses­si­ons will be hos­ted by Micro­soft, high­light­ing C++ AMP and Remo­te­FX tech­no­lo­gies; Acce­le­rEyes, show­ca­sing the use of libra­ri­es to enable hete­ro­ge­neous com­pu­ting; and HP, pro­vi­ding an over­view of its HP Moonshot den­se ser­ver envi­ron­ment for hos­ted desk­top. APU13 also fea­tures ses­si­ons that high­light how ser­ver APUs can be used to opti­mi­ze a varie­ty of data cen­ter workloads.

Today, AMD also announ­ced a new uni­fied SDK, tools and acce­le­ra­ted libra­ri­es for hete­ro­ge­neous com­pu­ting deve­lo­pers.

Supporting Resources

About AMD

AMD (NYSE: AMD) designs and inte­gra­tes tech­no­lo­gy that powers mil­li­ons of intel­li­gent devices, inclu­ding per­so­nal com­pu­ters, tablets, game con­so­les and cloud ser­vers that defi­ne the new era of sur­round com­pu­ting. AMD solu­ti­ons enable peo­p­le ever­y­whe­re to rea­li­ze the full poten­ti­al of their favo­ri­te devices and appli­ca­ti­ons to push the boun­da­ries of what is pos­si­ble. For more infor­ma­ti­on, visit www.amd.com.

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