AMPD Announces ‘Machine Learning Cloud’ Initiative Built Around AMD Instinct™ MI100 accelerators

- the World’s Fas­test HPC GPU for Sci­en­ti­fic Research*

Van­cou­ver, Bri­tish Colum­bia – March 11, 2021 – AMPD Ven­tures Inc. (“AMPD” or the “Com­pa­ny”, CSE: AMPD; FRA: 2Q0) is plea­sed to announ­ce a ‘Machi­ne Lear­ning Cloud’ initia­ti­ve desi­gned to cater to the requi­re­ments of aca­de­mic insti­tu­ti­ons and com­pa­nies in the arti­fi­cial intel­li­gence (“AI”), machi­ne lear­ning and deep lear­ning sec­tors. The plat­form, fea­turing AMD Instinct™ acce­le­ra­tors along with the AMD ROCm™ open soft­ware plat­form, will initi­al­ly be hos­ted at AMPD’s DC1 data cent­re in Van­cou­ver, Bri­tish Colum­bia, but is expec­ted to expand into other ter­ri­to­ries over the com­ing months.

AMD has been focu­sing on Data Cent­re Deep Lear­ning solu­ti­ons, com­bi­ning its AMD ROCm™ open soft­ware plat­form with its power­ful AMD Instinct™ acce­le­ra­tors. With an open-source tool­set, AMD ROCm™ is desi­gned to bring a rich foun­da­ti­on to advan­ced com­pu­ting with the goal of sol­ving real-world pro­blems. It is the first open-source exas­ca­le-class plat­form for acce­le­ra­ted com­pu­ting that’s also pro­gramming-lan­guage inde­pen­dent. It unlocks an array of dif­fe­rent open com­pu­te lan­guages, com­pi­lers, libra­ries, and tools desi­gned from the ground up to meet the deman­ding needs of AI sci­en­tists and rese­ar­chers — hel­ping to acce­le­ra­te code deve­lo­p­ment and sol­ve the toughest chal­len­ges in the world today.

AMPD expects to be the first com­pa­ny in Cana­da to offer the power of the AMD Instinct MI100 acce­le­ra­tors on a ser­vice basis. The AMD Instinct MI100 was announ­ced in Novem­ber 2020 as the world’s fas­test High-Per­for­mance Com­pu­ting (HPC) GPU, desi­gned with AMD’s all-new AMD CDNA™ archi­tec­tu­re and beco­m­ing the industry’s first data cen­ter GPU to exceed 10 Tera­flops (FP64)*.

AMPD intends to offer sub­si­di­zed access to the­se excep­tio­nal­ly power­ful com­pu­te resour­ces to the aca­de­mic arti­fi­cial intel­li­gence com­mu­ni­ty in Bri­tish Colum­bia, in addi­ti­on to its pro­grams for com­mer­cial cus­to­mers who wish to assess their AI com­pu­te opti­ons in a low-risk envi­ron­ment. AMPD is working with aca­de­mic insti­tu­ti­ons across the pro­vin­ce, as well as the ‘AI in BC’ indus­try asso­cia­ti­on to iden­ti­fy pro­grams that would bene­fit from access to the plat­form, inclu­ding a spe­cial empha­sis on pro­grams desi­gned to incre­a­se access to machi­ne lear­ning tech­no­lo­gy for under­re­pre­sen­ted groups with the aim of hel­ping ensu­re an inclu­si­ve and diver­se arti­fi­cial intel­li­gence future.

It’s clear that AMD and AMPD share the same com­mit­ment to the open-source com­mu­ni­ty and open-source tech­no­lo­gies that are the dri­ving for­ce behind ROCm™ plat­form inno­va­tions,” said Brad McCredie, cor­po­ra­te vice pre­si­dent, Data Cen­ter GPU and Acce­le­ra­ted Pro­ces­sing, AMD. “We are plea­sed to pro­vi­de AMPD with the power of the­se acce­le­ra­tors with the aim of faci­li­ta­ting access for aca­de­mic insti­tu­ti­ons and rese­ar­chers as cost effec­tively as pos­si­ble. Our indus­try-dif­fe­ren­tia­ting approach to acce­le­ra­ted com­pu­te and hete­ro­ge­ne­ous workload deve­lo­p­ment gives users unpre­ce­den­ted fle­xi­bi­li­ty, choice and plat­form auto­no­my, and we are exci­ted that AMPD plans to offer the­se advan­ta­ges on a ser­vice basis.”

Addi­tio­nal­ly, both AMPD and AMD will sup­port the com­mu­ni­ty via the ROCm™ Git­Hub forums, inclu­ding gui­d­ance and sup­port for HIP (Hete­ro­ge­ne­ous-Com­pu­ting Inter­face for Por­ta­bi­li­ty). AMD has deve­lo­ped a suite of tools desi­gned to ease con­ver­si­on of CUDA app­li­ca­ti­ons to por­ta­ble C++ code by auto­ma­ting much of the con­ver­si­on work via a source-to-source trans­for­ma­ti­on from CUDA to HIP.

BC is home to a cri­ti­cal mass of some 200 app­lied AI com­pa­nies as well as lea­ding aca­de­mic and rese­arch pro­grams,” said James Hurst­house, Chief Stra­te­gy Offi­cer at AMPD. “We expect that this cut­ting-edge GPU-based hos­ted envi­ron­ment pro­mo­ting open-source com­pu­ting opti­ons will yield posi­ti­ve out­co­mes for AI com­pa­nies and aca­de­mic rese­ar­chers and we are incredi­b­ly plea­sed that a com­pa­ny with the pedigree of AMD has reco­gni­zed AMPD as a key go-to-mar­ket part­ner. We alrea­dy have initia­ti­ves under way when it comes to the app­li­ca­ti­on of arti­fi­cial intel­li­gence and machi­ne lear­ning in other AMPD core sec­tors, such as the visu­al effects indus­try, as demons­tra­ted by our joint indus­try aca­de­mic pro­ject on arti­fi­cial intel­li­gence in moti­on cap­tu­re in con­junc­tion with AMPD cus­to­mer, Sho­cap Enter­tain­ment, and a team of AI rese­ar­chers at Simon Fra­ser University.”

Hurst­house recent­ly mode­ra­ted two panel dis­cus­sions on the the­me of arti­fi­cial intel­li­gence. On March 4th, 2021, he acted as Panel Chair for a ses­si­on on ‘How to Build an AI Clus­ter’ at the ‘Glo­bal AI: How to Build Across Bor­ders’ vir­tu­al con­fe­rence, hos­ted by CityA­ge, with the sup­port of the AI in BC indus­try asso­cia­ti­on and the Cana­da Korea Busi­ness Asso­cia­ti­on. Panelists inclu­ded repre­sen­ta­ti­ves from Mas­ter­Card, Cis­co AI Lab, and AI in BC. Hurst­house also mode­ra­ted a panel at the SPARK FX 2021 Vir­tu­al Con­fe­rence (“SPARK”) with the tit­le ‘Crea­ti­ve Arti­fi­cial Intel­li­gence and Movie Making: The Artist in the Loop’.

The glo­bal arti­fi­cial intel­li­gence (AI) soft­ware mar­ket is fore­cast to grow rapidly in the com­ing years, reaching around USD $126 bil­li­on by 2025[1]. The over­all AI mar­ket inclu­des a wide array of app­li­ca­ti­ons such as natu­ral lan­guage pro­ces­sing, robo­tic pro­cess auto­ma­ti­on, and machi­ne lear­ning. This growth is expec­ted to repre­sent an oppor­tu­ni­ty for a mul­ti-sec­tor com­pa­ny such as AMPD that sup­plies GPU-based high-per­for­mance com­pu­ting envi­ron­ments to host the­se applications.

AMD, the AMD logo, AMD Instinct, ROCm, AMD CDNA, 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.