{"id":53969,"date":"2020-01-14T18:36:49","date_gmt":"2020-01-14T17:36:49","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=53969"},"modified":"2020-01-14T18:36:49","modified_gmt":"2020-01-14T17:36:49","slug":"scalemp-demonstrates-best-in-class-large-system-results","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/53969-scalemp-demonstrates-best-in-class-large-system-results\/","title":{"rendered":"ScaleMP Demonstrates Best-in-Class Large-System Results"},"content":{"rendered":"

Company\u2019s soft\u00adware-defi\u00adned com\u00adposable ser\u00adver sys\u00adtems now lead SPEC<\/span> CPU<\/span> 2017 bench\u00admark results by gap of up to 96% vs. com\u00adpe\u00adting solutions<\/i><\/b><\/p>\n

FORT<\/span> LEE<\/span>, N.J.\u2013(BUSINESS<\/span> WIRE<\/span><\/a>)\u2013Scal\u00adeMP\u2122, a lea\u00adding pro\u00advi\u00adder of vir\u00adtua\u00adliza\u00adti\u00adon solu\u00adti\u00adons for high-end com\u00adpu\u00adting, today announ\u00adced that it has accom\u00adplished best-ever recor\u00added results for the SPEC<\/span> CPU<\/span>\u00ae 2017<\/a> bench\u00admark for any sys\u00adtem size of four pro\u00adces\u00adsors and hig\u00adher. ScaleMP\u2019s results for 32\u2011, 24- and 16-pro\u00adces\u00adsor sys\u00adtems are now hol\u00adding the top three posi\u00adti\u00adons for best results ever.<\/p>\n

In 2019, AMD<\/span> and Scal\u00adeMP announ\u00adced a stra\u00adte\u00adgic col\u00adla\u00adbo\u00adra\u00adti\u00adon<\/a> to enable AMD<\/span> ser\u00adver OEMs to crea\u00adte sys\u00adtems with four, eight and up to 128 pro\u00adces\u00adsor sockets, up to 8,192 CPUs and 256 tera\u00adbytes of shared memo\u00adry using ScaleMP\u2019s vSMP Foun\u00adda\u00adti\u00adon line of pro\u00adducts<\/a>.<\/p>\n

\u201c<\/span>We are extre\u00adme\u00adly plea\u00adsed with the\u00adse new bench\u00admark results, which show supe\u00adrio\u00adri\u00adty across the board,\u201d said Shai Fult\u00adheim, foun\u00adder and CEO<\/span> of Scal\u00adeMP. \u201cCus\u00adto\u00admers can now sca\u00adle their sys\u00adtems on demand for more com\u00adpu\u00adting or more memo\u00adry, pay only for what they need, and be sure that they are buy\u00ading the best and most cost-effec\u00adti\u00adve solu\u00adti\u00adon available that deli\u00advers the best performance.\u201d<\/p>\n

\u201c<\/span>ScaleMP\u2019s uni\u00adque soft\u00adware-defi\u00adned ser\u00adver tech\u00adno\u00adlo\u00adgy enables AMD<\/span> to cap\u00adtu\u00adre the lea\u00adding posi\u00adti\u00adons for the sys\u00adtem-sca\u00adle SPEC<\/span> CPU2017<\/span> bench\u00admark wit\u00adhout any cus\u00adtom engi\u00adnee\u00adring,\u201d said Rag\u00adhu Nam\u00adbi\u00adar, cor\u00adpo\u00adra\u00adte vice pre\u00adsi\u00addent, Dat\u00ada\u00adcen\u00adter Eco\u00adsys\u00adtems &<\/span> Appli\u00adca\u00adti\u00adon Engi\u00adnee\u00adring at AMD<\/span>. \u201cThrough our col\u00adla\u00adbo\u00adra\u00adti\u00adon with Scal\u00adeMP, we expect to con\u00adti\u00adnue to domi\u00adna\u00adte the sca\u00adle-up bench\u00admarks as this inno\u00adva\u00adti\u00adon continues.\u201d<\/p>\n

The fol\u00adlo\u00adwing table sum\u00adma\u00adri\u00adzes the cur\u00adrent best SPEC<\/span> CPU2017<\/span> Inte\u00adger Rates<\/a> and SPEC<\/span> CPU2017<\/span> Floa\u00adting Point Rates<\/a> for every sca\u00adle of ser\u00adver sys\u00adtem and con\u00adtrasts it with the pre\u00advious best system\/result for that scale.<\/p>\n

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
\n

SPEC<\/span> CPU2017<\/span> Inte\u00adger Point Rate<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

Pro\u00adces\u00adsor
\nSockets<\/b><\/p>\n<\/td>\n

\n

Best Sys\u00adtem \u2013 Janu\u00adary 14th<\/sup>, 2020<\/b><\/p>\n<\/td>\n

\n

Pre\u00advious System<\/b><\/p>\n<\/td>\n

\n

Impro\u00adve\u00adment<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

Ven\u00addor, Sys\u00adtem, Pro\u00adces\u00adsors &<\/span> Type<\/b><\/p>\n<\/td>\n

\n

Rank<\/b><\/p>\n<\/td>\n

\n

Rate<\/b><\/p>\n<\/td>\n

\n

Ven\u00addor, Sys\u00adtem, Pro\u00adces\u00adsors &<\/span> Type<\/b><\/p>\n<\/td>\n

\n

Rate<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

32<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#1<\/p>\n<\/td>\n

\n

7,100<\/p>\n<\/td>\n

\n

HPE<\/span> Super\u00addo\u00adme Flex:
\nIntel Xeon Pla\u00adti\u00adnum 8180<\/p>\n<\/td>\n

\n

3,920<\/p>\n<\/td>\n

\n

81%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

24<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#2<\/p>\n<\/td>\n

\n

5,920<\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n<\/tr>\n

\n

16<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#3<\/p>\n<\/td>\n

\n

4,290<\/p>\n<\/td>\n

\n

ATOS<\/span> Bull\u00adSe\u00adqua\u00adna S1600<\/span>:
\nIntel Xeon Pla\u00adti\u00adnum 8276M<\/span><\/p>\n<\/td>\n

\n

2,190<\/p>\n<\/td>\n

\n

96%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

8<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#6<\/p>\n<\/td>\n

\n

2,250<\/p>\n<\/td>\n

\n

Hua\u00adwei Kun\u00adlun 9008 V5<\/span>:
\nIntel Xeon Pla\u00adti\u00adnum 8280<\/p>\n<\/td>\n

\n

1,410<\/p>\n<\/td>\n

\n

60%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

4<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#35<\/p>\n<\/td>\n

\n

1,150<\/p>\n<\/td>\n

\n

Hua\u00adwei 2488H<\/span> V5<\/span>:
\nIntel Xeon Pla\u00adti\u00adnum 8280<\/p>\n<\/td>\n

\n

701<\/p>\n<\/td>\n

\n

64%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

SPEC<\/span> CPU2017<\/span> Floa\u00adting Point Rate<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

Pro\u00adces\u00adsor
\nSockets<\/b><\/p>\n<\/td>\n

\n

Best Sys\u00adtem \u2013 Janu\u00adary 14th<\/sup>, 2020<\/b><\/p>\n<\/td>\n

\n

Pre\u00advious System<\/b><\/p>\n<\/td>\n

\n

Impro\u00adve\u00adment<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

Ven\u00addor, Sys\u00adtem, Pro\u00adces\u00adsors &<\/span> Type<\/b><\/p>\n<\/td>\n

\n

Rank<\/b><\/p>\n<\/td>\n

\n

Rate<\/b><\/p>\n<\/td>\n

\n

Ven\u00addor, Sys\u00adtem, Pro\u00adces\u00adsors &<\/span> Type<\/b><\/p>\n<\/td>\n

\n

Rate<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

32<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#1<\/p>\n<\/td>\n

\n

5,870<\/p>\n<\/td>\n

\n

HPE<\/span> Super\u00addo\u00adme Flex:
\nIntel Xeon Pla\u00adti\u00adnum 8180<\/p>\n<\/td>\n

\n

3,290<\/p>\n<\/td>\n

\n

78%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

24<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#2<\/p>\n<\/td>\n

\n

4,510<\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n

\n

 <\/p>\n<\/td>\n<\/tr>\n

\n

16<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#4<\/p>\n<\/td>\n

\n

3,120<\/p>\n<\/td>\n

\n

HPE<\/span> Super\u00addo\u00adme Flex:
\nIntel Xeon Pla\u00adti\u00adnum 8180<\/p>\n<\/td>\n

\n

1,840<\/p>\n<\/td>\n

\n

70%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

8<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#8<\/p>\n<\/td>\n

\n

1,550<\/p>\n<\/td>\n

\n

Hua\u00adwei Kun\u00adlun 9008 V5<\/span>:
\nIntel Xeon Pla\u00adti\u00adnum 8280<\/p>\n<\/td>\n

\n

1,160<\/p>\n<\/td>\n

\n

34%<\/b><\/p>\n<\/td>\n<\/tr>\n

\n

4<\/b><\/p>\n<\/td>\n

\n

Scal\u00adeMP Super\u00admi\u00adcro A+ Ser\u00adver 2123BT-HNC0R<\/span>:
\nAMD<\/span> EPYC<\/span> 7702<\/p>\n<\/td>\n

\n

#68<\/p>\n<\/td>\n

\n

806<\/p>\n<\/td>\n

\n

Hua\u00adwei 2488H<\/span> V5<\/span>:
\nIntel Xeon Pla\u00adti\u00adnum 8280<\/p>\n<\/td>\n

\n

578<\/p>\n<\/td>\n

\n

39%<\/b><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Scal\u00adeMP soli\u00addi\u00adfies the lea\u00adding posi\u00adti\u00adon of its paten\u00adted vSMP Foun\u00adda\u00adti\u00adon tech\u00adno\u00adlo\u00adgy and line of pro\u00adducts as the best sca\u00adle-up com\u00adpu\u00adting archi\u00adtec\u00adtu\u00adre for com\u00adpu\u00adte-inten\u00adsi\u00adve workloads, with up to 96% bet\u00adter per\u00adfor\u00admance com\u00adpared to alter\u00adna\u00adti\u00adve sys\u00adtems. The results, made public by the SPEC<\/span>.org<\/a>, show that ScaleMP\u2019s on-demand com\u00adpo\u00adsed com\u00adpu\u00adting archi\u00adtec\u00adtu\u00adre pro\u00advi\u00addes bet\u00adter sca\u00adling than hard\u00adware-only approaches.<\/p>\n

Refe\u00adrence Results<\/b><\/p>\n