National Post (National Edition) - - CANADA - JOSEPH BREAN Na­tional Post jbrean@na­tion­al­post.com Twit­ter.com/JosephBrean

Anew su­per­com­put­ing part­ner­ship be­tween a Cana­dian pi­o­neer in quan­tum com­put­ers and a U.S. Depart­ment of En­ergy lab­o­ra­tory, which aims to reach “ex­as­cale” com­put­ing speeds within a few years, of­fers a glimpse of the fu­ture of ul­tra­fast com­pu­ta­tion, ac­cord­ing to the sci­en­tist lead­ing the project.

Rather than a whole­sale shift from clas­si­cal to quan­tum com­put­ing — as when in­ter­nal com­bus­tion en­gines re­placed steam power — the fu­ture of su­per­com­put­ing is likely to in­volve hy­brid strate­gies, with reg­u­lar dig­i­tal com­put­ers aug­mented by other more fancy kinds: quan­tum com­put­ers, graph­ics pro­ces­sors like the kind that run video games, and neu­ro­mor­phic ma­chines that mimic the be­hav­iour of the hu­man brain.

“It’s a per­fect world for DWave,” said Jeff Ni­chols, as­so­ciate lab­o­ra­tory di­rec­tor of com­put­ing and com­pu­ta­tional sciences at Oak Ridge Na­tional Lab­o­ra­tory in Ten­nessee.

“I do not be­lieve that you’ll ever re­place all of tra­di­tional, clas­si­cal com­put­ing with a quan­tum com­puter, nor will any of the other more ex­otic ap­proaches re­place clas­si­cal com­put­ers,” he said. “You’re not go­ing to carry around a quan­tum com­puter as your phone.”

Burn­aby, B.C.-based D-Wave Sys­tems’ new deal to pro­vide quan­tum com­put­ing power to ac­cel­er­ate Oak Ridge’s su­per­com­put­ers also marks a key strat­egy in the U.S. ef­fort to catch up to China, which has in­vested heav­ily in its push to build com­put­ers fast enough to reach the ex­as­cale, or a quin­til­lion cal­cu­la­tions per sec­ond. (A quin­til­lion is a bil­lion bil­lion, or 1 with 18 ze­ros.)

A ma­jor dif­fer­ence in the two coun­tries’ strate­gies has to do with the mas­sive en­ergy costs of run­ning such a fast com­puter, which in the case of Oak Ridge’s Ti­tan ma­chine is nine megawatts at its peak, at a cost of $9 mil­lion.

China has tried to start with the nec­es­sary hard­ware, then bring the en­ergy us­age and costs down. But Ni­chols said Oak Ridge is tak­ing the op­po­site ap­proach with the strate­gic place­ment of quan­tum “ac­cel­er­a­tors” to im­prove the ef­fi­ciency of cal­cu­la­tion.

In fu­ture, he said they might wish to have a quan­tum com­puter on site, “tightly cou­pled” to their su­per­com­puter but, for the mo­ment, D-Wave’s ser­vice will be pro­vided re­motely, over the in­ter­net from Canada.

The prize of an ex­as­cale com­puter, for both China and the U.S., would be a vastly im­proved abil­ity to solve some of science’s most com­plex prob­lems, such as those about cli­mate change, ge­netic anal­y­sis, pro­tein fold­ing, earth­quake pre­dic­tion, the per­for­mance of the elec­tric­ity grid, and cos­mol­ogy — prob­lems that are too big to sim­ply cal­cu­late by run­ning through all the pos­si­bil­i­ties.

Many of these are what math­e­ma­ti­cians call “op­ti­miza­tion prob­lems,” and these are what D-Wave’s quan­tum com­puter is best suited to solve, as they re­cently did, for ex­am­ple, in a study for Volk­swa­gen about how to op­ti­mize traf­fic flow in Bei­jing.

The clas­sic ex­am­ple of an op­ti­miza­tion prob­lem is of a trav­el­ling sales­man who needs to visit many towns and wants to know the short­est route.

Clas­si­cal com­put­ers, the kind made with sil­i­con chips, would just cal­cu­late each trip and choose the short­est. But in this kind of prob­lem, the num­ber of pos­si­bil­i­ties soon grows im­pos­si­bly large. To solve it clas­si­cally, you would need to be cal­cu­lat­ing for­ever with a com­puter as big as the uni­verse. To quan­tum com­put­ers, how­ever, the math and logic of op­ti­miza­tion prob­lems look very dif­fer­ent. They do not com­pute with the strict ones and ze­ros of bi­nary code, but rather with the strange quan­tum prop­er­ties of su­per­po­si­tion and en­tan­gle­ment.

A clas­si­cal com­puter cal­cu­lates with bits, which can be set two ways: one or zero. From this ba­sic bi­nary sys­tem, a com­puter can build up to all the com­plex­i­ties of mod­ern com­put­ing.

A quan­tum com­puter, how­ever, takes ad­van­tage of the strange prop­er­ties of mat­ter at the sub­atomic scale. Rather than bits, it cal­cu­lates with qubits, or quan­tum bits, which are tiny, frag­ile phys­i­cal sys­tems — some­times etched into a chip of metal cooled to near ab­so­lute zero, or a gas held in place by a mag­netic field, or a sliver of ar­ti­fi­cial di­a­mond — that can be in mul­ti­ple quan­tum states at the same time, known as su­per­po­si­tion. This prop­erty al­lows a qubit to be either one, zero, or a lit­tle bit of both at the same time, al­low­ing for a whole new style of logic and com­pu­ta­tion.

D-Wave’s de­vice uses a strat­egy known as quan­tum an­neal­ing to solve op­ti­miza­tion prob­lems not by brute cal­cu­la­tion, but by ex­ploit­ing quan­tum ef­fects to find the like­li­est can­di­dates for so­lu­tions.

Us­ing this style of com­put­ing to help a su­per­com­puter skip un­nec­es­sary cal­cu­la­tions helps Oak Ridge to keep its power costs down, while ac­cel­er­at­ing its per­for­mance, Ni­chols said.

The suc­cess of this ap­proach is a key rea­son that D-Wave pres­i­dent Bo Ewald thinks the fu­ture of quan­tum com­put­ing will look dif­fer­ent than the rapid ex­pan­sion and con­stant im­prove­ment of clas­si­cal com­put­ers since the mid-20th cen­tury.

He has a long his­tory in toplevel com­pu­ta­tion, for ex­am­ple at Los Alamos Na­tional Lab­o­ra­tory and as pres­i­dent of Cray Re­search, which once made su­per­com­put­ers that filled a room, cost mil­lions, and are now more or less matched by an off-theshelf lap­top.

“I al­ways knew that be­cause of Moore’s Law, things were go­ing to get faster and shrink,” he said. (Moore’s law says the num­ber of tran­sis­tors in a com­puter chip dou­bles ev­ery two years, and it has held true for decades.) “But in quan­tum, you’re us­ing more spe­cial­ized ma­te­ri­als, very cold su­per­con­duct­ing ma­te­ri­als in ex­treme vac­uum, shielded from ra­dio fre­quency.”

It is a more finicky hard­ware, he said, so he is skep­ti­cal that we will all carry quan­tum com­put­ers in our pock­ets in the fu­ture, as we do now with clas­si­cal com­put­ers such as iPhones.

“So I think there’s a lit­tle more chal­lenge to think we’ll have por­ta­ble quan­tum com­put­ers,” he said. “I don’t think we’ll need them be­cause I think they’ll be ubiq­ui­tous be­cause of the Cloud.”

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.