AMD EPYC™ Processors and New AMD Instinct™ MI100 Accelerator Redefine Performance for HPC and Scientific Research

Sup­port for next genera­ti­on AMD EPYC pro­ces­sors, code­n­a­med ‘Milan,’ in new HBv3 Vir­tu­al Machi­nes announ­ced by Microsoft—

AMD Instinct™ MI100 acce­le­ra­tor is first acce­le­ra­tor to use new AMD CDNA archi­tec­tu­re dedi­ca­ted to high-per­for­mance com­pu­ting (HPC) workloads—

SANTA CLARA, Calif. — Novem­ber 16, 2020 — During this year’s SC20 vir­tu­al tra­de­show, AMD (NASDAQ: AMD) is show­ca­sing its lea­ders­hip in the high per­for­mance com­pu­ting (HPC) indus­try. It laun­ched the new AMD Instinct™ MI100 acce­le­ra­tor with ROCm™ 4.0 open eco­sys­tem sup­port, show­ca­sed a gro­wing list of AMD EPYCCPU and AMD Instinct acce­le­ra­tor based deploy­ments, and high­ligh­ted the growth AMD EPYC CPUs and AMD Instinct™ GPUs on the TOP500 list. AMD also remains on track to begin volu­me ship­ments of the 3rd Gen EPYC pro­ces­sors with “Zen 3” core to select HPC and cloud cus­to­mers this quar­ter in advan­ce of the expec­ted public launch in Q1 2021, ali­gned with OEM availability.

The new AMD Instinct™ MI100 acce­le­ra­tor, is the world’s fas­test HPC GPU acce­le­ra­tor for sci­en­ti­fic workloads and the first to sur­pass the 10 tera­flops (FP64) per­for­mance bar­ri­er i. Built on the new AMD CDNA archi­tec­tu­re, the AMD Instinct MI100 GPU enab­les a new class of acce­le­ra­ted sys­tems for HPC and AI when pai­red with 2nd Gen AMD EPYC pro­ces­sors. Sup­por­ted by new acce­le­ra­ted com­pu­te plat­forms from Dell, HPE, Giga­byte and Super­mi­cro, the MI100, com­bi­ned with AMD EPYC CPUs and ROCm 4.0 soft­ware, is desi­gned to pro­pel new dis­co­ve­ries ahead of the exas­ca­le era.

No two cus­to­mers are the same in HPC, and AMD is pro­vi­ding a path to today’s most advan­ced tech­no­lo­gies and capa­bi­li­ties that are cri­ti­cal to sup­port their HPC work, from small clus­ters on pre­mi­se, to vir­tu­al machi­nes in the cloud, all the way to exas­ca­le super­com­pu­ters,” said For­rest Nor­rod, seni­or vice pre­si­dent and gene­ral mana­ger, Data Cen­ter and Embed­ded Solu­ti­ons Busi­ness Group, AMD. “Com­bi­ning AMD EPYC pro­ces­sors and Instinct acce­le­ra­tors with cri­ti­cal app­li­ca­ti­on soft­ware and deve­lo­p­ment tools enab­les AMD to deli­ver lea­ders­hip per­for­mance for HPC workloads.”

AMD and Micro­soft Azu­re Power HPC In the Cloud

Azu­re is using 2nd Gen AMD EPYC pro­ces­sors to power its HBv2 vir­tu­al machi­nes (VMs) for HPC workloads. The­se VMs offer up to 2x the per­for­mance of first-genera­ti­on HB-seri­es vir­tu­al machi­nesi, can sup­port up to 80,000 cores for MPI jobsii, and take advan­ta­ge of 2nd Gen AMD EPYC pro­ces­sors’ up to 45% more memo­ry band­width than com­pa­ra­ble x86 alter­na­ti­vesiii.

HBv2 VMs are used by nume­rous cus­to­mers inclu­ding The Uni­ver­si­ty of Illi­nois at Urbana-Champaign’s Beck­man Insti­tu­te for Advan­ced Sci­ence & Tech­no­lo­gy which used 86,400 cores to model a plant virus that pre­vious­ly requi­red a lea­ders­hip class super­com­pu­ter and the U.S. Navy which rapidly deploys and sca­les enhan­ced wea­ther and oce­an pat­tern pre­dic­tions on demand. HBv2 powe­red by 2nd Gen AMD EPYC pro­ces­sors also pro­vi­des the bulk of the CPU com­pu­te power for theOpe­nAI envi­ron­ment Micro­soft announ­ced ear­lier this year. 

AMD EPYC pro­ces­sors have also hel­ped HBv2 reach new cloud HPC mile­stones, such as a new record for Cloud MPI sca­ling results with NAMD, Top 20 results on the Graph500, and the first 1 terabyte/sec cloud HPC par­al­lel file­sys­tem. Across the­se and other app­li­ca­ti­on bench­marks, HBv2 is deli­vering 12x hig­her sca­ling than found else­whe­re on the public cloud.

Adding on to its exis­ting HBv2 HPC vir­tu­al machi­ne powe­red by 2nd Gen AMD EPYC pro­ces­sors, Azu­re announ­ced it will uti­li­ze next genera­ti­on AMD EPYC pro­ces­sors, code­n­a­med ‘Milan’, for future HB-seri­es VM pro­ducts for HPC

You can see more about the AMD and Azu­re col­la­bo­ra­ti­on in this video with Jason Zan­der of Azu­re and Lisa Su of AMD

AMD Con­ti­nues to Be the Choice for HPC 

AMD EPYC pro­ces­sors and Instinct acce­le­ra­tors have the per­for­mance and capa­bi­li­ties to sup­port nume­rous HPC workloads across a varie­ty of imple­men­ta­ti­ons. From small clus­ters at rese­arch cen­ters, to com­mer­cial HPC, to off pre­mi­se and in the cloud, to exas­ca­le com­pu­ting, AMD con­ti­nues to pro­vi­de per­for­mance and choice for HPC solutions. 

Hew­lett Packard Enter­pri­se (HPE), CSC Fin­land and EuroHPC recent­ly intro­du­ced a new pre-exas­ca­le sys­tem, LUMI. Based on the HPE Cray EX super­com­pu­ter archi­tec­tu­re, LUMI will use next genera­ti­on AMD EPYC CPUs and Instinct acce­le­ra­tors and is expec­ted to pro­vi­de a peak per­for­mance of 552 peta­flops when it comes online in 2021, making it one of the fas­test super­com­pu­ters in the world.

Bey­ond LUMI, AMD powe­red HPC sys­tems con­ti­nue to grow in volu­me. Sin­ce SC19, the­re have been more than 15 super­com­pu­ting sys­tems announ­ced using AMD EPYC CPUs, Instinct GPUs, or both. A high­light of the sys­tems includes 

  • Chi­co­ma – Los Ala­mos Natio­nal Labo­ra­to­ry– this sys­tem is based on the HPE Cray EX super­com­pu­ter archi­tec­tu­re and uses 2nd Gen AMD EPYC CPUs, com­bi­ned with 300 tera­bytes of sys­tem memo­ry for COVID-19 research, 
  • Coro­na – Law­rence Liver­mo­re Natio­nal Labo­ra­to­ry– this sys­tem was recent­ly upgraded with fun­ding from the Coro­na­vi­rus Aid, Reli­ef and Eco­no­mic Secu­ri­ty (CARES) Act, adding near­ly 1,000 AMD Instinct MI50 acce­le­ra­tors, pushing peak per­for­mance to more than 11 petaFLOPS,
  • Mam­mo­th – Law­rence Liver­mo­re Natio­nal Labo­ra­to­ry– the “big memo­ry” clus­ter uses 2nd Gen AMD EPYC Pro­ces­sors to per­form geno­mics ana­ly­sis and graph ana­ly­tics requi­red by sci­en­tists working on COVID-19.
  • Nort­hern Data – a dis­tri­bu­t­ed com­pu­ting sys­tem in Euro­pe that is using AMD EPYC CPUs and Instinct acce­le­ra­tors for lar­ge sca­le HPC app­li­ca­ti­ons such as ren­de­ring, arti­fi­cial intel­li­gence and deep learning, 
  • Paw­sey Super­com­pu­ting Cent­re – Using the HPE Cray EX super­com­pu­ter archi­tec­tu­re and future AMD EPYC CPUs and AMD Instinct acce­le­ra­tors, the super­com­pu­ter at Paw­sey will be Australia’s most power­ful supercomputer. 


In addi­ti­on, AMD is also powe­ring the fol­lowing super­com­pu­ters: Anvil and Bell – Pur­due Uni­ver­si­ty, Big Red 200 – India­na Uni­ver­si­ty, Brid­ges 2 – Pitts­burgh Super­com­pu­ting Cen­ter, CERN, Euro­pean Cent­re for Medi­um-Ran­ge Wea­ther Fore­casts, Expan­se – San Die­go Super­com­pu­ter Cen­ter, Goe­the Uni­ver­si­ty Frank­furt, IT4Innovations Natio­nal Super­com­pu­ting Cen­ter, Jet­stream 2 – India­na Uni­ver­si­ty, Mah­tiCSC, Man­gi – Uni­ver­si­ty of Min­ne­so­ta, Natio­nal Ocea­nic and Atmo­s­phe­ric Admi­nis­tra­ti­on, Red Rai­der – Texas Tech Uni­ver­si­ty, Tin­ker­Cliffs– Vir­gi­nia Tech.

With the Expan­se super­com­pu­ter, our goal is to give sci­en­tists and rese­ar­chers cloud-like access to a high-per­for­mance machi­ne that can hand­le ever­ything from astro­phy­sics to zoo­lo­gy,” said Micha­el Nor­man, Direc­tor of the San Die­go Super­com­pu­ter Cen­ter. “The 2nd Gen AMD EPYC pro­ces­sors have hel­ped us achie­ve fan­tastic per­for­mance with Expan­se, enab­ling our rese­ar­chers to do more sci­ence than befo­re. We also have a gre­at col­la­bo­ra­ti­on with AMD and have worked tog­e­ther to crea­te a forum for AMD HPC cus­to­mers to share expe­ri­en­ces, infor­ma­ti­on and more, to bet­ter bene­fit HPC research.”

Paving the Path to Exas­ca­le Computing 

To help rese­ar­chers start on the path to exas­ca­le, AMD has pro­vi­ded Oak Ridge Natio­nal Labs access to the new AMD Instinct MI100 acce­le­ra­tor, which deli­vers a giant leap in com­pu­te and inter­con­nect per­for­mance. The Instinct MI100 acce­le­ra­tor enab­les a new class of acce­le­ra­ted sys­tems and deli­vers true hete­ro­ge­ne­ous com­pu­te capa­bi­li­ties from AMD for HPC and AI. Desi­gned to com­ple­ment the 2nd Gen AMD EPYC pro­ces­sors, and built on the AMD Infi­ni­ty Archi­tec­tu­re, the AMD Instinct MI100 deli­vers true hete­ro­ge­ne­ous com­pu­te capa­bi­li­ties from AMD for HPC and AI

Fron­tier, powe­red by AMD, repres­ents a huge incre­a­se in com­pu­ta­tio­nal power com­pa­red to today’s sys­tems. It’s going to allow sci­en­tists to ans­wer ques­ti­ons that we did­n’t have the ans­wer to befo­re,” said Bron­son Mes­ser, direc­tor of sci­ence, Oak Ridge Lea­ders­hip Com­pu­ting Faci­li­ty. “The abi­li­ty to run mole­cu­lar simu­la­ti­ons that aren’t just a few mil­li­on atoms, but a few bil­li­on atoms, pro­vi­des a more rea­listic repre­sen­ta­ti­on of the sci­ence, and to be able to do that as a mat­ter of cour­se and over and over again will lead to a signi­fi­cant amount of important discoveries.”

AMD con­ti­nues to pro­vi­de the per­for­mance, capa­bi­li­ties and sca­le nee­ded to power cur­rent and future HPC workloads, no mat­ter if they are hel­ping stu­dents at a rese­arch cen­ter, impro­ving aero­dy­na­mic effi­ci­en­cy for an auto manu­fac­tu­rer, or pro­vi­ding valu­able insights for cri­ti­cal medi­cal bre­akthroughs. Read more about the AMD pre­sence at SC20 and its HPC capa­bi­li­ties here.

Sup­por­ting Resources

About AMD

For more than 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: AMDweb­siteblogFace­book and Twit­ter pages.


This press release con­tains for­ward-loo­king state­ments con­cer­ning Advan­ced Micro Devices, Inc. (AMD) such as the fea­tures, func­tio­n­a­li­ty, per­for­mance, avai­la­bi­li­ty, timing and expec­ted bene­fits of AMD pro­ducts inclu­ding the next genera­ti­on AMD EPYCTM Pro­ces­sors and AMD InstinctTM MI100 Acce­le­ra­tor and the expec­ted timing and bene­fits of AMD part­ners­hips, 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,” “may,” “expects,” “belie­ves,” “plans,” “intends,” “pro­jects” and other terms with simi­lar mea­ning. Inves­tors are cau­tio­ned that the for­ward-loo­king state­ments in this press release are based on cur­rent beliefs, assump­ti­ons and expec­ta­ti­ons, speak only as of the date of this press release 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; the abi­li­ty of third par­ty manu­fac­tu­rers to manu­fac­tu­re AMD’s pro­ducts on a time­ly basis in suf­fi­ci­ent quan­ti­ties and using com­pe­ti­ti­ve tech­no­lo­gies; expec­ted manu­fac­tu­ring yiel­ds for AMD’s pro­ducts; the avai­la­bi­li­ty of essen­ti­al equip­ment, mate­ri­als or manu­fac­tu­ring pro­ces­ses; AMD’s abi­li­ty to intro­du­ce pro­ducts on a time­ly basis with fea­tures and per­for­mance levels that pro­vi­de value to its cus­to­mers; glo­bal eco­no­mic uncer­tain­ty; the loss of a signi­fi­cant cus­to­mer; AMD’s abi­li­ty to gene­ra­te reve­nue from its semi-cus­tom SoC pro­ducts;  the impact of the COVID-19 pan­de­mic on AMD’s busi­ness, finan­cial con­di­ti­on and results of ope­ra­ti­ons; poli­ti­cal, legal, eco­no­mic risks and natu­ral dis­as­ters; the impact of government actions and regu­la­ti­ons such as export admi­nis­tra­ti­on regu­la­ti­ons, tariffs and tra­de pro­tec­tion mea­su­res; the impact of acqui­si­ti­ons, joint ven­tures and/or invest­ments on AMD’s busi­ness, inclu­ding the announ­ced acqui­si­ti­on of Xilinx, and the fail­u­re to inte­gra­te acqui­red busi­nes­ses; AMD’s abi­li­ty to com­ple­te the Xilinx mer­ger; the impact of the announ­ce­ment and pen­den­cy of the Xilinx mer­ger on AMD’s busi­ness; poten­ti­al secu­ri­ty vul­nera­bi­li­ties; poten­ti­al IT outa­ges, data loss, data breaches and cyber-attacks; uncer­tain­ties invol­ving the orde­ring and ship­ment of AMD’s pro­ducts; quar­ter­ly and sea­so­nal sales pat­terns; the restric­tions impo­sed by agree­ments gover­ning AMD’s notes and the revol­ving credit faci­li­ty; the com­pe­ti­ti­ve mar­kets in which AMD’s pro­ducts are sold; mar­ket con­di­ti­ons of the indus­tries in which AMD pro­ducts are sold; AMD’s reli­an­ce on third-par­ty intel­lec­tu­al pro­per­ty to design and intro­du­ce new pro­ducts in a time­ly man­ner; AMD’s reli­an­ce on third-par­ty com­pa­nies for the design, manu­fac­tu­re and sup­ply of mother­boards, soft­ware and other com­pu­ter plat­form com­pon­ents; AMD’s reli­an­ce on

Micro­soft Cor­po­ra­ti­on and other soft­ware ven­dors’ sup­port to design and deve­lop soft­ware to run on AMD’s pro­ducts; AMD’s reli­an­ce on third-par­ty dis­tri­bu­tors and add-in-board part­ners; the poten­ti­al dilu­ti­ve effect if the 2.125% Con­ver­ti­ble Seni­or Notes due 2026 are con­ver­ted; future impairments of good­will and tech­no­lo­gy licen­se purcha­ses; AMD’s abi­li­ty to attract and retain qua­li­fied per­son­nel; AMD’s abi­li­ty to gene­ra­te suf­fi­ci­ent reve­nue and ope­ra­ting cash flow or obtain exter­nal finan­cing for rese­arch and deve­lo­p­ment or other stra­te­gic invest­ments; AMD’s indeb­ted­ness; AMD’s abi­li­ty to gene­ra­te suf­fi­ci­ent cash to ser­vice its debt obli­ga­ti­ons or meet its working capi­tal requi­re­ments; AMD’s abi­li­ty to repurcha­se its out­stan­ding debt in the event of a chan­ge of con­trol; the cycli­cal natu­re of the semi­con­duc­tor indus­try; the impact of modi­fi­ca­ti­on or inter­rup­ti­on of AMD’s inter­nal busi­ness pro­ces­ses and infor­ma­ti­on sys­tems; com­pa­ti­bi­li­ty of AMD’s pro­ducts with some or all indus­try-stan­dard soft­ware and hard­ware; cos­ts rela­ted to defec­ti­ve pro­ducts; the effi­ci­en­cy of AMD’s sup­ply chain; AMD’s abi­li­ty to rely on third par­ty sup­ply-chain logistics func­tions; AMD’s stock pri­ce vola­ti­li­ty; world­wi­de poli­ti­cal con­di­ti­ons; unfa­vor­able cur­ren­cy exchan­ge rate fluc­tua­tions; AMD’s abi­li­ty to effec­tively con­trol the sales of its pro­ducts on the gray mar­ket; AMD’s abi­li­ty to ade­qua­te­ly pro­tect its tech­no­lo­gy or other intel­lec­tu­al pro­per­ty; cur­rent and future claims and liti­ga­ti­on; poten­ti­al tax lia­bi­li­ties; and the impact of envi­ron­men­tal laws, con­flict mine­rals-rela­ted pro­vi­si­ons and other laws or regu­la­ti­ons. 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 26, 2020.


©2020 Advan­ced Micro Devices, Inc. All rights reser­ved. AMD, the AMD Arrow logo, EPYC, AMD Instinct, and com­bi­na­ti­ons the­re­of are trade­marks of Advan­ced Micro Devices, Inc. PCIe is a regis­tered trade­mark of PCI-SIG Cor­po­ra­ti­on. Other pro­duct names used in this publi­ca­ti­on are for iden­ti­fi­ca­ti­on pur­po­ses only and may be trade­marks of their respec­ti­ve companies.


i Cal­cu­la­ti­ons con­duc­ted by AMD Per­for­mance Labs as of Sep 18, 2020 for the AMD Instinct™ MI100 (32GB HBM2 PCIe® card) acce­le­ra­tor at 1,502 MHz peak boost engi­ne clock resul­ted in 11.54 TFLOPS peak dou­ble pre­cisi­on (FP64), 46.1 TFLOPS peak sin­gle pre­cisi­on matrix (FP32), 23.1 TFLOPS peak sin­gle pre­cisi­on (FP32), 184.6 TFLOPS peak half pre­cisi­on (FP16) peak theo­re­ti­cal, floa­ting-point per­for­mance. Publis­hed results on the NVi­dia Ampere A100 (40GB) GPU acce­le­ra­tor resul­ted in 9.7 TFLOPS peak dou­ble pre­cisi­on (FP64). 19.5 TFLOPS peak sin­gle pre­cisi­on (FP32), 78 TFLOPS peak half pre­cisi­on (FP16) theo­re­ti­cal, floa­ting-point per­for­mance. Ser­ver manu­fac­tu­rers may vary con­fi­gu­ra­ti­on offe­rings yiel­ding dif­fe­rent results. MI100-03

ii Source:

iii Source:

iv AMD EPYC™ 7002 Seri­es pro­ces­sors have 45% more memo­ry band­width than Intel Scala­b­le pro­ces­sors in the same class.

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.