NEW ORLEANS, La., Nov. 17 — The Open­ACC stan­dards group today announ­ced that Advan­ced Micro Devices (AMD) and PathSca­le, Inc., have joi­n­ed the orga­niza­ti­on, alig­ning with 18 other mem­bers and sup­port­ers world­wi­de to dri­ve the deve­lo­p­ment of appli­ca­ti­ons for acce­le­ra­tor-enab­led super­com­pu­ters, clus­ters and other high per­for­mance com­pu­ting (HPC) systems.

Open­ACC pro­vi­des an effi­ci­ent and per­for­mance-por­ta­ble path for deve­lo­ping mas­si­ve­ly par­al­lel pro­grams across a wide ran­ge of acce­le­ra­tors, inclu­ding GPUs, many core copro­ces­sors and mul­ti-core CPUs. By joi­ning the Open­ACC stan­dards group, AMD and PathSca­le bring con­sidera­ble exper­ti­se in high per­for­mance com­pu­ting tech­no­lo­gy and pro­gramming capa­bi­li­ties, pro­vi­ding deve­lo­pers with more diver­si­ty and choice in com­pu­ting archi­tec­tures and soft­ware tools, and pro­mi­sing to extend the reach of Open­ACC-sup­port­ed solu­ti­ons to new and exis­ting users worldwide.

Expert Open­ACC mem­bers will be on hand at the Super­com­pu­ting Con­fe­rence (SC14) in New Orleans, Loui­sia­na USA this week, to ans­wer Open­ACC pro­gramming ques­ti­ons, dis­cuss the latest deve­lo­p­ments, and pro­vi­de tech­no­lo­gy demos. For more infor­ma­ti­on visit the Open­ACC booth (#657) or the Open­ACC web­site: http://www.openacc.org/SC14.

Open­ACC Sim­pli­fies Appli­ca­ti­ons Acce­le­ra­ti­on on AMD APUs

Advan­ced Micro Devices is a lea­ding mul­ti­na­tio­nal semi­con­duc­tor com­pa­ny with vast expe­ri­ence in ener­gy-effi­ci­ent high-per­for­mance com­pu­ting. The Open­ACC stan­dard helps pro­gramm­ers take full advan­ta­ge of AMD’s high-per­for­mance Acce­le­ra­ted Pro­ces­sing Units (APUs).

“The HPC com­mu­ni­ty is dri­ving the need for rich hete­ro­ge­neous com­pu­ting solu­ti­ons to bet­ter meet cus­to­mer per­for­mance and power needs. With this tran­si­ti­on the­re is a strong need for a rich, sim­pli­fied pro­g­ra­ming model to sup­port new hete­ro­ge­neous appli­ca­ti­on deve­lo­p­ment and pull in lega­cy C/C++/Fortran appli­ca­ti­ons,“  said Greg Stoner,  seni­or direc­tor of Deve­lo­per Tech­no­lo­gy, Pro­fes­sio­nal Gra­phics, AMD, and new­ly appoin­ted mem­ber of the Open­ACC board of direc­tors. “Open­ACC 2.0 via its sin­gle source direc­ti­ves-based pro­gramming model puts in place a solid foun­da­ti­on to dri­ve key bene­fits of a hete­ro­ge­neous hard­ware plat­form. AMD is loo­king for­ward to working with the Open­ACC orga­niza­ti­on to help it to con­ti­nue to evol­ve this important pro­gramming standard.

PathSca­le Joins OpenACC

PathSca­le is the deve­lo­per of inno­va­ti­ve ENZO 2014 com­pi­lers that enhan­ce the per­for­mance and ener­gy effi­ci­en­cy of acce­le­ra­ted Linux HPC clus­ters and sys­tems. By sup­port­ing the Open­ACC stan­dard, PathSca­le pro­vi­des users with more opti­ons and fle­xi­bi­li­ty to choo­se the ide­al com­pi­ler solu­ti­on that fits their indi­vi­du­al needs.

“It is important in any HPC deve­lo­p­ment envi­ron­ment to have access to a varie­ty of com­pi­lers,” noted Oscar Her­nan­dez, prin­ci­ple inves­ti­ga­tor at Oak Ridge Natio­nal Labo­ra­to­ry. “Some­ti­mes one com­pi­ler will exhi­bit bet­ter per­for­mance on a par­ti­cu­lar code than ano­ther, or behave in a more desi­ra­ble way. We ful­ly intend to make PathSca­le available to deve­lo­pers crea­ting appli­ca­ti­ons for ORNL’s Titan Cray XK7 supercomputer.”

Open­ACC Events at SC14

Open­ACC mem­bers are par­ti­ci­pa­ting in num­ber of pre­sen­ta­ti­ons, talks and dis­cus­sions at SC14 on the latest deve­lo­p­ments. They will also pro­vi­de tech­no­lo­gy demons­tra­ti­ons in a varie­ty of mem­ber booths, inclu­ding demons­tra­ti­ons of the latest Open­ACC fea­tures as deve­lo­ped for the next major release of the popu­lar open-source GCC compiler.

For more infor­ma­ti­on about Open­ACC demons­tra­ti­ons and events visit http://www.openacc.org/SC14

Key Open­ACC events at SC14 include:

• Work­shop — “Work­shop on Acce­le­ra­tor Pro­gramming Using Direc­ti­ves” — Mon­day, Nov. 17, 2014, 8am to 5pm in Rooms 276 and 277.
• Tuto­ri­al — “Open­ACC: Pro­duct, Por­ta­ble Per­for­mance on Hybrid Sys­tems using High Level Com­pi­lers and Tools” — Mon­day, Nov. 17, 8am to 5pm in Room 396.
• Open­ACC Birds of a Fea­ther (BoF) ses­si­on— “Open­ACC API User Expe­ri­ence, Ven­dor Reac­tion, Rele­van­ce and Road­map” — Wed­nes­day, Nov. 19, 2014 at 5:30pm in Rooms 275/276/277. BoF high­lights will include reports on the recent hacka­thon at ORNL, the first Open­ACC Seis­mic work­shop at the Uni­ver­si­ty of Hous­ton, and recent pro­gress in imple­men­ting and enhan­cing the Open­ACC standard.

The Open­ACC stan­dard sim­pli­fies the pro­gramming of acce­le­ra­ted-com­pu­ting sys­tems through the use of direc­ti­ves, enab­ling appli­ca­ti­on deve­lo­pers world­wi­de to easi­ly take advan­ta­ge of the trans­for­ma­ti­ve power of the­se sys­tems to dri­ve advan­ces in sci­ence, engi­nee­ring and indus­try.  It is now available in com­pi­ler pro­ducts from many lea­ding com­pi­ler ven­dors and is sup­port­ed in a wide ran­ge of debug­gers, pro­fi­lers and other pro­gramming tools.

About the Open­ACC Stan­dards Group

Open­ACC is a non-pro­fit cor­po­ra­ti­on that crea­ted and fos­ters a cross-plat­form API which allows any sci­en­tist or pro­gramm­er to more easi­ly acce­le­ra­te appli­ca­ti­ons on modern many-core and mul­ti-core pro­ces­sors. Open­ACC direc­ti­ves iden­ti­fy com­pu­te-inten­si­ve code to a com­pi­ler for acce­le­ra­ti­on or off­load, while pre­ser­ving a sin­gle code base and cross-plat­form por­ta­bi­li­ty. Open­ACC is com­ple­men­ta­ry to and inter­ope­ra­ble with exis­ting pro­gramming models, inclu­ding OpenMP, MPI, CUDA and Open­CL. Orga­niza­ti­ons actively invol­ved in high-per­for­mance hete­ro­ge­neous com­pu­ting are encou­ra­ged to par­ti­ci­pa­te. Plea­se visit http://www.openacc.org for more information.

About AMD

AMD is a semi­con­duc­tor design inno­va­tor lea­ding the next era of vivid digi­tal expe­ri­en­ces with its ground­brea­king AMD Acce­le­ra­ted Pro­ces­sing Units (APUs) that power a wide ran­ge of com­pu­ting devices. AMD’s ser­ver com­pu­ting pro­ducts are focu­sed on dri­ving indus­try-lea­ding cloud com­pu­ting and vir­tua­liza­ti­on envi­ron­ments. AMD’s supe­ri­or gra­phics tech­no­lo­gies are found in a varie­ty of solu­ti­ons ran­ging from game con­so­les, PCs to super­com­pu­ters. For more infor­ma­ti­on, visit www.amd.com.

About PathSca­le

PathSca­le Inc. has deve­lo­ped indus­try lea­ding high per­for­mance For­tran, C and C++ com­pi­ler pro­ducts for AMD, Intel and NVIDIA pro­ces­sors. The PathSca­le com­pi­ler tech­no­lo­gy has the world’s most advan­ced opti­miza­ti­on infra­struc­tu­re and can ful­ly exploit the poten­ti­als of many-core and hete­ro­ge­neous archi­tec­tures. The company’s goal is to deli­ver robust and high per­for­mance com­pi­lers tail­o­red to clus­te­red, GPGPU and mul­ti-core com­pu­ting envi­ron­ments. More infor­ma­ti­on about PathSca­le is available on the web at http://www.pathscale.com

Source: Open­ACC Stan­dards Group