At SC19, the pre­mier annu­al event for super­com­pu­ting, AMD (NASDAQ: AMD) is exten­ding its per­for­mance lead in high-per­for­mance com­pu­ting (HPC) with a ran­ge of new cus­to­mer wins in top rese­arch sys­tems world­wi­de, new plat­forms sup­port­ing AMD EPYC™ pro­ces­sors and Rade­on Instinct™ acce­le­ra­tors and the new­ly announ­ced ROCm 3.0, brin­ging sup­port for new com­pi­lers and HPC applications.

“AMD enters SC19 as the pro­ces­sor pro­vi­der for the upco­ming Fron­tier super­com­pu­ter, expec­ted to be the hig­hest per­for­mance super­com­pu­ter in the world when it is deli­ver­ed in 2021,” said Mark Paper­mas­ter, exe­cu­ti­ve vice pre­si­dent and chief tech­no­lo­gy offi­cer, AMD. “Show atten­de­es can spend time with the same foun­da­tio­nal AMD tech­no­lo­gies that will go into that exas­ca­le-class sys­tem this week. From high per­for­mance AMD EPYC CPUs and Rade­on Instinct GPUs working tog­e­ther across high-speed inter­con­nects, to our open soft­ware eco­sys­tem, all deli­ver­ed by the big­gest names in super­com­pu­ting, it’s all here at SC19.”

At SC19, Paper­mas­ter will be on a panel along with CTOs from Cray, CERN and Xilinx to dis­cuss the uni­que inno­va­tions in exas­ca­le sys­tems and bey­ond, inclu­ding new tech­ni­cal breakth­roughs across hard­ware, soft­ware and pro­gramming tools.

New Amazon EC2 Compute-Optimized Instances to Use 2^nd Gen AMD EPYC

Ama­zon Web Ser­vices (AWS) is expan­ding its use of AMD EPYC pro­ces­sors, announ­cing the upco­ming launch of two new Ama­zon Ela­s­tic Com­pu­te Cloud (Ama­zon EC2) com­pu­te-opti­mi­zed ins­tances. The­se ins­tances, C5a and C5ad, will be powered by cus­tom 2^nd Gen AMD EPYC pro­ces­sors, run­ning at fre­quen­ci­es up to 3.3Ghz. C5a and C5ad will come in 8 vir­tua­li­zed sizes with up to 96 vCPUs which will pro­vi­de addi­tio­nal choices to help cus­to­mers opti­mi­ze both cost and per­for­mance for a varie­ty of com­pu­te inten­si­ve workloads, inclu­ding batch pro­ces­sing, dis­tri­bu­ted ana­ly­tics, and web appli­ca­ti­ons. Both ins­tances will be available in bare metal vari­ants which will enable cus­to­mers to run appli­ca­ti­ons that bene­fit from direct access to the pro­ces­sor and memo­ry resour­ces of the under­ly­ing ser­ver. The­se ins­tances will have 192 logi­cal pro­ces­sors on 96 phy­si­cal cores, which will be twice the lar­gest ins­tance size offe­red in the EC2 com­pu­te-opti­mi­zed ins­tance fami­ly. The C5a and C5ad bare metal ins­tance will be able to uti­li­ze 100 Gbps net­work band­width and will be com­pa­ti­ble with Ela­s­tic Fabric Adap­ter, enab­ling cus­to­mers to sca­le up High Per­for­mance Com­pu­ting and other lar­ge com­pu­te inten­si­ve workloads. The new ins­tances will be available soon across mul­ti­ple AWS regions.

AMD Powers Supercomputers to New Levels and EPYC Joins TOP500

HPC orga­niza­ti­ons are con­ti­nuing to adopt the 2^nd Gen AMD EPYC pro­ces­sor and Rade­on Instinct acce­le­ra­tors for more powerful and effi­ci­ent super­com­pu­ting sys­tems. The 2^nd Gen EPYC pro­ces­sors pro­vi­de twice the manu­fac­tu­ring appli­ca­ti­on per­for­mance¹ and up to 60% fas­ter Life Sci­en­ces simu­la­ti­ons than com­pe­ting solu­ti­ons,^2,3 while the Rade­on Instinct GPU acce­le­ra­tor pro­vi­des up to 6.6 peak theo­re­ti­cal TFLOPS Dou­ble Pre­cis­i­on per­for­mance for HPC workloads. Both pro­ducts sup­port PCIe^® 4.0 enab­ling high-band­width inter­con­nect for fas­ter com­pu­te across hete­ro­ge­nous systems.

The latest cus­to­mers deploy­ing AMD pro­ces­sors and acce­le­ra­tors include:

• Atos, a glo­bal lea­der in digi­tal trans­for­ma­ti­on, is sup­p­ly­ing two Bull­Se­qua­na XH2000 super­com­pu­ters based on 2^nd Gen EPYC to Météo-France for ope­ra­tio­nal wea­ther fore­cas­ting and rese­arch in atmo­sphe­ric, oce­an and cli­ma­te science.
• Atos and the French natio­nal high-per­for­mance com­pu­ting orga­niza­ti­on, GENCI, announ­ced that the latest exten­si­on of its Joli­ot-Curie super­com­pu­ter, loca­ted and admi­nis­te­red by the CEA teams at its TGCC (Very Lar­ge Com­pu­ting Cent­re), and based on Atos’ Bull­Se­qua­na XH2000 solu­ti­on and 2^nd Gen EPYC is now operational.
• The exten­si­on of the Joli­ot-Curie super­com­pu­ter joi­n­ed the 54th edi­ti­on of the TOP500, making it the first AMD EPYC 7H12 64-core 280W⁴equip­ped super­com­pu­ter in the glo­bal ranking.
• Cray, a Hew­lett Packard Enter­pri­se com­pa­ny, recent­ly announ­ced two new super­com­pu­ters equip­ped with the 2nd Gen AMD EPYC pro­ces­sors and based on the Shas­ta™ super­com­pu­ter archi­tec­tu­re, ARCHER2 and Vul­can.
• Dell Tech­no­lo­gies is sup­p­ly­ing the San Die­go Super­com­pu­ter Cen­ter and its Expan­se super­com­pu­ter with Dell EMC PowerEdge ser­vers and 2^nd Gen EPYC Processors.
• ETH Zurich is using AMD EPYC 7742 pro­ces­sors in its Euler VI system.
• NEC is sup­p­ly­ing the Ger­man wea­ther fore­cas­ting ser­vice, Deut­scher Wet­ter­dienst, a sys­tem using 2^nd Gen EPYC pro­ces­sors in con­junc­tion with NEC SX-Auro­ra TSUBASA vec­tor engines.

“We cho­se the AMD EPYC 7742 becau­se the pro­ces­sor not only pro­vi­ded broad sup­port for most com­mon soft­ware appli­ca­ti­ons our rese­ar­chers use, but it pro­vi­des the raw per­for­mance, impres­si­ve memo­ry and I/O band­width and most important­ly the price/performance that is cri­ti­cal to mee­ting the demands of the­se rese­ar­chers,” said Chris­ti­an Bol­li­ger, sci­en­ti­fic IT ser­vices, ETH Zurich. “With the Euler VI sys­tem using 2^nd Gen AMD EPYC pro­ces­sors, our users now have access to a sys­tem that will pro­vi­de them the capa­bi­li­ties nee­ded to advan­ce their research.”

Bringing Supercomputing to the Cloud with AMD EPYC

The HPC indus­try is evol­ving to sup­port new workloads, grea­ter demands for per­for­mance and most important­ly, easier access for long term or tem­po­ra­ry use. This is being done through the Cloud, which for the first time can pro­vi­de users with the same levels of per­for­mance, at lower use cos­ts, com­pared to what they can deploy on-premise.

Micro­soft Azu­re announ­ced pre­vious­ly unob­tainable levels of per­for­mance for com­pu­ta­tio­nal flu­id dyna­mics (CFD) using an Azu­re HB cloud ins­tance run­ning on an 1st Gen AMD EPYC-pro­ces­sor based system.

Now, Azu­re is pushing the boun­da­ries of HPC in the cloud even fur­ther with the pre­view of Azu­re HBv2 vir­tu­al machi­nes for high-per­for­mance com­pu­ting. Based on the AMD EPYC 7742 pro­ces­sor, the­se vir­tu­al machi­nes pro­vi­de cus­to­mers with access to super­com­pu­ter per­for­mance, sup­port­ing 200Gbps HDR Infi­ni­Band, and up to 80,000 cores for a sin­gle job, in the ease and sim­pli­ci­ty through the Cloud.

AMD Intro­du­ces ROCm 3.0

Com­mu­ni­ty sup­port for the pre-exas­ca­le soft­ware eco­sys­tem con­ti­nues to grow. This eco­sys­tem is built on ROCm, the foun­da­tio­nal open source com­pon­ents for GPU com­pu­te pro­vi­ded by AMD. The ROCm deve­lo­p­ment cycle fea­tures month­ly releases offe­ring deve­lo­pers a regu­lar cadence of con­ti­nuous impro­ve­ments and updates to com­pi­lers, libra­ri­es, pro­fi­lers, debug­gers and sys­tem manage­ment tools. Major deve­lo­p­ment mile­sto­nes fea­tured at SC19 include:

• Intro­duc­tion of ROCm 3.0 with new inno­va­tions to sup­port HIP-clang – a com­pi­ler built upon LLVM, impro­ved CUDA con­ver­si­on capa­bi­li­ty with hipi­fy-clang, libra­ry opti­miza­ti­ons for both HPC and ML.
• ROCm upstream inte­gra­ti­on into lea­ding Ten­sor­Flow and PyTorch machi­ne lear­ning frame­works for appli­ca­ti­ons like rein­force­ment lear­ning, auto­no­mous dri­ving, and image and video detection.
• Expan­ded acce­le­ra­ti­on sup­port for HPC pro­g­ra­ming models and appli­ca­ti­ons like OpenMP pro­g­ra­ming, LAMMPS, and NAMD. 
• New sup­port for sys­tem and workload deploy­ment tools like Kuber­netes, Sin­gu­la­ri­ty, SLURM, TAU and others.

A Growing Hardware Ecosystem

As HPC sys­tems are asked to sup­port more deman­ding workloads, the need for hete­ro­ge­nous com­pu­te powered by CPUs and acce­le­ra­tors is cri­ti­cal for modern HPC sys­tems. AMD part­ners are crea­ting plat­forms that sup­port this demand, as well as tra­di­tio­nal CPU only computing:

• GIGABYTE announ­ced four new G‑Series GPU ser­vers that sup­port 2^nd Gen AMD EPYC pro­ces­sors, the G292-Z22, G292-Z42, G482-Z50 and G482-Z51. The G482-Z51 can sup­port up to eight PCIe 4.0 GPU cards, giving cus­to­mers gre­at “AMD + AMD” opti­ons for a ran­ge of acce­le­ra­ted com­pu­ting workloads.
• The world-record shat­te­ring HPE Pro­Li­ant DL325 Gen10 and DL385 Gen10 ser­vers are joi­n­ed by new Gen10 Plus models signi­fi­cant­ly enhan­cing per­for­mance and effi­ci­en­cy for vital workloads like vir­tua­liza­ti­on, HPC and Big Data with up to 64 pro­ces­sor cores, 3200 MT/s memo­ry for 9 per­cent fas­ter memo­ry per­for­mance, 2X grea­ter I/O band­width with PCIe 4.0 sup­port and 2.4X grea­ter sto­rage capa­ci­ty ver­sus pre­vious generations.
• With the new Pen­gu­in Alt­us^® XE4218GT, sup­port­ing up to 8 GPUs, Pen­gu­in cus­to­mers have an “AMD + AMD” solu­ti­on that uses the PCIe 4.0 sup­port in the 2^nd Gen EPYC and Rade­on Instinct MI50 to power machi­ne lear­ning, big data ana­ly­tics, and simi­lar workloads.
• Tyan also announ­ced new plat­forms in its HPC-focu­sed Trans­port HX pro­duct line and data­ba­se-focu­sed Trans­port SX pro­duct line powered by 2^nd Gen AMD EPYC processors.

With the intro­duc­tion of PCIe 4.0 sup­port in the 2^nd Gen AMD EPYC pro­ces­sors and Rade­on Instinct GPU acce­le­ra­tors, AMD has led the ena­blem­ent of the PCIe 4.0 eco­sys­tem by working clo­se­ly with key indus­try part­ners. The 2^nd Gen EPYC sup­ports PCIe 4.0 enab­led devices such as high speed ether­net and Infi­ni­Band^® inter­con­nects, NICs and swit­ches, com­pu­te acce­le­ra­tors (GPUs and FPGAs) and sto­rage (NVME SSD) devices. Some of the part­ners offe­ring PCIe 4.0 pro­ducts for 2^nd Gen EPYC include:

• Broad­com Thor NIC for 200 GB ethernet.
• Mel­lan­ox ConnectX‑6 NIC show­ing ~400 GB/s Infi­ni­Band performance.
• Sam­sung Gen4 PM1733 NVME SSD – Show­ca­sing 2x of IOPS over the Sam­sung Gen3 SSD.
• Xilinx Alveo U50, U280 FPGAs.

Additional Resources

• AMD EPYC on AMD.com
• Learn more about AMD Rade­on Instinct™ MI50 accelerators
• Learn more about the ROCm 3.0 open soft­ware plat­form here
• Learn about how AMD EPYC per­forms with HPC soft­ware here
• Fol­low AMD dat­a­cen­ter deve­lo­p­ments on Twit­ter @AMDServer 
• Fol­low AMD Rade­on Instinct™ on Twit­ter @RadeonInstinct

About AMD

For 50 years, AMD has dri­ven inno­va­ti­on in high-per­for­mance com­pu­ting, gra­phics and visua­liza­ti­on tech­no­lo­gies ― the buil­ding blocks for gam­ing, immersi­ve plat­forms, and the data cen­ter. Hundreds of mil­li­ons of con­su­mers, lea­ding For­tu­ne 500 busi­nesses and cut­ting-edge sci­en­ti­fic rese­arch faci­li­ties around the world rely on AMD tech­no­lo­gy dai­ly to impro­ve how they live, work and play. AMD employees around the world are focu­sed on buil­ding gre­at pro­ducts that push the boun­da­ries of what is pos­si­ble. For more infor­ma­ti­on about how AMD is enab­ling today and inspi­ring tomor­row, visit the AMD (NASDAQ: AMD) web­site, blog, and Face­book and Twit­ter pages.

Cautionary Statement

This press release con­ta­ins for­ward-loo­king state­ments con­cer­ning Advan­ced Micro Devices, Inc. (AMD) inclu­ding the fea­tures, func­tion­a­li­ty, avai­la­bi­li­ty, timing, deploy­ment and expec­ta­ti­ons of AMD future pro­ducts and tech­no­lo­gies as well as future col­la­bo­ra­ti­ons and the expec­ted bene­fits of tho­se col­la­bo­ra­ti­ons, which are made pur­su­ant to the Safe Har­bor pro­vi­si­ons of the Pri­va­te Secu­ri­ties Liti­ga­ti­on Reform Act of 1995. For­ward-loo­king state­ments are com­mon­ly iden­ti­fied by words such as “would,” “intends,” “belie­ves,” “expects,” “may,” “will,” “should,” “seeks,” “intends,” “plans,” “pro for­ma,” “esti­ma­tes,” “anti­ci­pa­tes,” or the nega­ti­ve of the­se words and phra­ses, other varia­ti­ons of the­se words and phra­ses or com­pa­ra­ble ter­mi­no­lo­gy. Inves­tors are cau­tio­ned that the for­ward-loo­king state­ments in this docu­ment are based on cur­rent beliefs, assump­ti­ons and expec­ta­ti­ons, speak only as of the date of this docu­ment and invol­ve risks and uncer­tain­ties that could cau­se actu­al results to dif­fer mate­ri­al­ly from cur­rent expec­ta­ti­ons. Such state­ments are sub­ject to cer­tain known and unknown risks and uncer­tain­ties, many of which are dif­fi­cult to pre­dict and gene­ral­ly bey­ond AMD’s con­trol, that could cau­se actu­al results and other future events to dif­fer mate­ri­al­ly from tho­se expres­sed in, or impli­ed or pro­jec­ted by, the for­ward-loo­king infor­ma­ti­on and state­ments. Mate­ri­al fac­tors that could cau­se actu­al results to dif­fer mate­ri­al­ly from cur­rent expec­ta­ti­ons include, wit­hout limi­ta­ti­on, the fol­lo­wing: Intel Corporation’s domi­nan­ce of the micro­pro­ces­sor mar­ket and its aggres­si­ve busi­ness prac­ti­ces may limit AMD’s abili­ty to com­pe­te effec­tively; AMD reli­es on third par­ties to manu­fac­tu­re its pro­ducts, and if they are unable to do so on a time­ly basis in suf­fi­ci­ent quan­ti­ties and using com­pe­ti­ti­ve tech­no­lo­gies, AMD’s busi­ness could be mate­ri­al­ly adver­se­ly affec­ted; fail­ure to achie­ve expec­ted manu­fac­tu­ring yields for AMD’s pro­ducts could nega­tively impact its finan­cial results; AMD has a wafer sup­p­ly agree­ment with GF with obli­ga­ti­ons to purcha­se all of its micro­pro­ces­sor and APU pro­duct requi­re­ments, and a cer­tain por­ti­on of its GPU pro­duct requi­re­ments, from GLOBALFOUNDRIES Inc. (GF) with limi­t­ed excep­ti­ons. If GF is not able to satis­fy AMD’s manu­fac­tu­ring requi­re­ments, its busi­ness could be adver­se­ly impac­ted; the suc­cess of AMD’s busi­ness is depen­dent upon its abili­ty to intro­du­ce pro­ducts on a time­ly basis with fea­tures and per­for­mance levels that pro­vi­de value to its cus­to­mers while sup­port­ing and coin­ci­ding with signi­fi­cant indus­try tran­si­ti­ons; if AMD can­not gene­ra­te suf­fi­ci­ent reve­nue and ope­ra­ting cash flow or obtain exter­nal finan­cing, it may face a cash short­fall and be unable to make all of its plan­ned invest­ments in rese­arch and deve­lo­p­ment or other stra­te­gic invest­ments; the loss of a signi­fi­cant cus­to­mer may have a mate­ri­al adver­se effect on AMD; AMD’s receipt of reve­nue from its semi-cus­tom SoC pro­ducts is depen­dent upon its tech­no­lo­gy being desi­gned into third-par­ty pro­ducts and the suc­cess of tho­se pro­ducts; glo­bal eco­no­mic uncer­tain­ty may adver­se­ly impact AMD’s busi­ness and ope­ra­ting results; AMD’s ope­ra­ti­ons are sub­ject to poli­ti­cal, legal and eco­no­mic risks and natu­ral dis­as­ters which could have a mate­ri­al adver­se effect on AMD; govern­ment actions and regu­la­ti­ons such as export admi­nis­tra­ti­on regu­la­ti­ons, tariffs and trade pro­tec­tion mea­su­res, may limit AMD’s abili­ty to export its pro­ducts to cer­tain cus­to­mers; AMD pro­ducts may be sub­ject to secu­ri­ty vul­nerabi­li­ties that could have a mate­ri­al adver­se effect on AMD; IT outa­ges, data loss, data brea­ches and cyber-attacks could com­pro­mi­se AMD’s intellec­tu­al pro­per­ty or other sen­si­ti­ve infor­ma­ti­on, be cos­t­ly to reme­dia­te and cau­se signi­fi­cant dama­ge to its busi­ness and repu­ta­ti­on; AMD’s ope­ra­ting results are sub­ject to quar­ter­ly and sea­so­nal sales pat­terns; AMD may not be able to gene­ra­te suf­fi­ci­ent cash to ser­vice its debt obli­ga­ti­ons or meet its working capi­tal requi­re­ments; AMD has a lar­ge amount of indeb­ted­ness which could adver­se­ly affect its finan­cial posi­ti­on and pre­vent it from imple­men­ting its stra­tegy or ful­fil­ling its con­trac­tu­al obli­ga­ti­ons; the agree­ments gover­ning AMD’s notes and the Secu­red Revol­ving Line of Cre­dit impo­se rest­ric­tions on AMD that may adver­se­ly affect its abili­ty to ope­ra­te its busi­ness; the mar­kets in which AMD’s pro­ducts are sold are high­ly com­pe­ti­ti­ve; the con­ver­si­on of the 2.125% Con­ver­ti­ble Seni­or Notes due 2026 may dilute the owner­ship inte­rest of its exis­ting stock­hol­ders, or may other­wi­se depress the pri­ce of its com­mon stock; uncer­tain­ties invol­ving the orde­ring and ship­ment of AMD’s pro­ducts could mate­ri­al­ly adver­se­ly affect it; the demand for AMD’s pro­ducts depends in part on the mar­ket con­di­ti­ons in the indus­tries into which they are sold. Fluc­tua­tions in demand for AMD’s pro­ducts or a mar­ket decli­ne in any of the­se indus­tries could have a mate­ri­al adver­se effect on its results of ope­ra­ti­ons; AMD’s abili­ty to design and intro­du­ce new pro­ducts in a time­ly man­ner is depen­dent upon third-par­ty intellec­tu­al pro­per­ty; AMD depends on third-par­ty com­pa­nies for the design, manu­fac­tu­re and sup­p­ly of mother­boards, soft­ware and other com­pu­ter plat­form com­pon­ents to sup­port its busi­ness; if AMD loses Micro­soft Corporation’s sup­port for its pro­ducts or other soft­ware ven­dors do not design and deve­lop soft­ware to run on AMD’s pro­ducts, its abili­ty to sell its pro­ducts could be mate­ri­al­ly adver­se­ly affec­ted; and AMD’s reli­ance on third-par­ty dis­tri­bu­tors and AIB part­ners sub­jects it to cer­tain risks.  Inves­tors are urged to review in detail the risks and uncer­tain­ties in AMD’s Secu­ri­ties and Exch­an­ge Com­mis­si­on filings, inclu­ding but not limi­t­ed to AMD’s Quar­ter­ly Report on Form 10‑Q for the quar­ter ended Sep­tem­ber 28, 2019.

Links to third par­ty sites are pro­vi­ded for con­ve­ni­ence and unless expli­cit­ly sta­ted, AMD is not respon­si­ble for the con­tents of such lin­ked sites and no endor­se­ment is implied.

AMD, the AMD Arrow logo, EPYC, Rade­on, Rade­on Instinct, 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.

1. Based on AMD inter­nal test­ing of ANSYS FLUENT 19.1, lm6000_16m bench­mark, as of July 17, 2019 of a 2P EPYC 7742 powered refe­rence ser­ver ver­sus a 2P Intel Xeon Pla­ti­num 8280 powered ser­ver. Results may vary. ROM-42
2. AMD Inter­nal test­ing as of 30July2019 of a 2P AMD EPYC 7742 powered refe­rence plat­form ver­sus a 2P Intel Pla­ti­num 8280 powered pro­duc­tion ser­ver, on GROMACS ver­si­on 2019.3 bench­mark. Results may vary. ROM-113
3. “Best-in-class based on indus­try-stan­dard pin-based (LGA) X86 pro­ces­sors. EPYC-08”
4. EPYC 7H12 pro­ces­sor boost fre­quen­ci­es may be achie­ved only with a coo­ling solu­ti­on that meets group ‘Z’ requi­re­ments.  Achie­va­ble boost fre­quen­ci­es may vary depen­ding on the effec­ti­ve­ness of the actu­al coo­ling solu­ti­on. ROM-282