— ESnet6 leverages AMD adaptive networking technology for packet monitoring at extremely high transfer rates and to allow for continuous innovation in the network —

SANTA CLARA, Calif., Oct. 11, 2022 (GLOBE NEWSWIRE) — Today AMD (NASDAQ: AMD) announ­ced its col­la­bo­ra­ti­on with the Ener­gy Sci­en­ces Net­work (ESnet) on the launch of ESnet6, the newest gene­ra­ti­on of the U.S. Depart­ment of Energy’s (DOE’s) high-per­for­mance net­work dedi­ca­ted to science.

AMD work­ed clo­se­ly with ESnet sin­ce 2018 to inte­gra­te powerful adap­ti­ve com­pu­ting for the smart and pro­gramma­ble net­work nodes of ESnet6. ESnet6’s extre­me sca­le packet moni­to­ring sys­tem uses AMD Alveo™ U280 FPGA-based net­work-atta­ched acce­le­ra­tor cards at the core net­work swit­ching nodes. This will enable high-touch packet pro­ces­sing and help impro­ve the accu­ra­cy of net­work moni­to­ring and manage­ment to enhan­ce per­for­mance. The pro­gramma­ble hard­ware allows for new capa­bi­li­ties to be added for con­ti­nuous inno­va­ti­on in the network.

In order to cus­to­mi­ze AMD Alveo U280 2x100Gb/s acce­le­ra­tors as net­work inter­face cards (NIC) for ESnet6, the Open­NIC over­lay – deve­lo­ped by AMD – was used to pro­vi­de stan­dard net­work-inter­face and host-attach­ment hard­ware, allo­wing novel and expe­ri­men­tal net­wor­king func­tions to be imple­men­ted easi­ly on the Alveo card. Open­NIC has sin­ce been open-sourced after being suc­cessful­ly used and evol­ved by ESnet, as well as various lea­ding aca­de­mic rese­arch groups. Also important for rapid inno­va­ti­on by non-hard­ware experts was the use of the AMD VitisNetP4 deve­lo­p­ment tools for com­pi­ling the P4 packet pro­ces­sing lan­guage to FPGA hardware.

AMD Alveo U280 cards, using Open­NIC and VitisNetP4, are being deploy­ed on every node of the ESnet6 net­work. The high-touch approach based on FPGA-acce­le­ra­ted pro­ces­sing allows every packet in the ESnet6 net­work to be moni­to­red at extre­me­ly high trans­fer rates to enable deep insights into the beha­vi­or of the net­work, as well as hel­ping to rapidly detect and cor­rect pro­blems and hot spots in the net­work. The Alveo U280 card with Open­NIC plat­form also sup­pli­es the adap­ta­bi­li­ty to allow the con­ti­nuous roll-out of new capa­bi­li­ties to the end user com­mu­ni­ty as needs evol­ve over the life­time of ESnet6.

The bene­fits of AMD adap­ti­ve com­pu­ting at deman­ding net­work line rates are also appli­ed to the edges of the ESnet6 net­work. Glo­bal­ly dis­tri­bu­ted sci­en­ti­fic instru­ments are the pri­ma­ry sources of the “big data” sci­ence flows that ESnet must sup­port. The Alveo card with Open­NIC can be used for cus­to­mi­zed data fil­te­ring and sha­ping befo­re trans­mis­si­on over ESnet. For super­com­pu­ting faci­li­ties, which are among the desti­na­ti­ons for the­se lar­ge data sets, the Alveo card with Open­NIC can be used for load balan­cing across com­pu­te and sto­rage servers.

“AMD is hono­red to pro­vi­de ESnet with lea­ding-edge tech­no­lo­gy and exper­ti­se nee­ded to help bring the new net­work into ser­vice,” said Ivo Bol­sens, seni­or vice pre­si­dent and CTO of Adap­ti­ve and Embedded Com­pu­ting Group, at AMD. “AMD FPGA-based adap­ti­ve net­wor­king capa­bi­li­ties make up a core buil­ding block of the ESnet6 extre­me-sca­le packet pro­ces­sing sys­tem and pro­vi­de the under­pin­ning for the DOE’s future visi­on of dis­tri­bu­ted e‑science.”

High-capa­ci­ty, high-per­for­mance networking
Head­quar­te­red at Law­rence Ber­ke­ley Natio­nal Labo­ra­to­ry (Ber­ke­ley Lab), ESnet ser­ves as the “data cir­cu­la­to­ry sys­tem” for the DOE, con­nec­ting all of its natio­nal labo­ra­to­ries, and tens of thou­sands of DOE-fun­ded rese­ar­chers. This includes DOE’s pre­mier sci­en­ti­fic instru­ments and super­com­pu­ter cen­ters – inclu­ding the world’s first exas­ca­le sys­tem, the Fron­tier super­com­pu­ter powered by AMD EPYC™ CPUs and AMD Instinct™ Accelerators.

“ESnet6 is built to sup­port the DOE’s mul­ti-bil­li­on-dol­lar invest­ment in sci­en­ti­fic rese­arch that leads to breakth­roughs that impact our ever­y­day lives. It repres­ents a trans­for­ma­tio­nal chan­ge in the way net­works are built for sci­en­ti­fic rese­arch,” said Inder Mon­ga, exe­cu­ti­ve direc­tor of ESnet. “AMD’s tech­no­lo­gy helps us enhan­ce the resi­li­en­cy, fle­xi­bi­li­ty and per­for­mance of the network.”

With over 46 Tera­bits per second of band­width, ESnet6 fea­tures a signi­fi­cant increase in band­width over pri­or gene­ra­ti­ons to sup­port uni­que needs of sci­en­ti­fic rese­arch. With this boost in capa­ci­ty, along with new ser­vice capa­bi­li­ties, sci­en­tists can more quick­ly pro­cess, ana­ly­ze, visua­li­ze, share, and store the moun­ta­ins of rese­arch data pro­du­ced by expe­ri­ments, mode­ling and simu­la­ti­ons. Data rapidly moves bet­ween the inter­con­nec­ted sites and col­la­bo­ra­tors, acce­le­ra­ting time to discovery.

Sup­port­ing Resour­ces 

• Learn more about AMD Alveo Data Cen­ter Acce­le­ra­tor Cards
• Read about AMD Exas­ca­le Com­pu­ting Tech­no­lo­gies and AMD HPC Solutions
• Learn more about AMD EPYC Pro­ces­sors and AMD Instinct Accelerators
• Beco­me a fan of AMD on Face­book
• Fol­low AMD on Twit­ter

About AMD 
For more than 50 years AMD has dri­ven inno­va­ti­on in high-per­for­mance com­pu­ting, gra­phics and visua­liza­ti­on tech­no­lo­gies. Bil­li­ons of peo­p­le, lea­ding For­tu­ne 500 busi­nesses and cut­ting-edge sci­en­ti­fic rese­arch insti­tu­ti­ons around the world rely on AMD tech­no­lo­gy dai­ly to impro­ve how they live, work and play. AMD employees are focu­sed on buil­ding lea­der­ship high-per­for­mance and adap­ti­ve 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, Lin­ke­dIn and Twit­ter pages.

AMD, the AMD Arrow logo, Alveo, EPYC, AMD Instinct, Xilinx 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