ALL FOR SCI­ENCE, SCI­ENCE FOR ALL

HWM (Malaysia) - - SPECIAL -

If you’ve been fol­low­ing us on our tra­vails with NVIDIA, you’ll know that near­ing the end of 2015, we were given a chance to take a tour of the CAVE2 fa­cil­ity, a gi­ant dis­play built and pow­ered by NVIDIA’s own Quadro work­sta­tion GPUs. The fa­cil­ity, which is housed inside the Clay­ton cam­pus of Monash Univer­sity in Mel­bourne, Aus­tralia, was de­signed to aid aca­demic re­searchers in their pur­suit of sci­ence.

As one would ex­pect, the back­bone be­hind the CAVE2 is, of course, pow­ered by not one, but two su­per­com­put­ers, known sim­ply as MAS­SIVE (short for Mul­ti­modal Aus­tralian ScienceS Imag­ing and Vi­su­al­iza­tion En­vi­ron­ment). How­ever, back in Fe­bru­ary of this year, Monash Univer­sity un­veiled yet another su­per­com­puter – known sim­ply as the MAS­SIVE 3 (or M3), to fur­ther aid in their sci­en­tific en­deav­ors in the field of medicine, dis­eases and ill­nesses.

Just like its two other su­per­com­puter sibil­ings, the new M3 su­per­com­puter is a beast of a ma­chine. Where the CAVE2 was de­signed to cre­ate highly

HWM de­tailed ren­ders of sub­jects, places, and the odd anatom­i­cal struc­ture of a bi­o­log­i­cal be­ing, the M3 was pri­mar­ily de­signed for the pur­pose of High Per­for­mance Com­put­ing (HPC).

The M3 is no slouch ei­ther. Like the pre­vi­ous two su­per­com­put­ers, the M3 is pow­ered by NVIDIA graph­ics cards. More specif­i­cally, the M3 runs on a com­bi­na­tion of: • 1,700 In­tel Haswell CPU-cores, • 50 NVIDIA Tesla K80 GPU co­pro­ces­sors which are used for data pro­cess­ing and high-end vi­su­al­iza­tion, • Eight NVIDIA GRID K1 GPUs for medium-end vi­su­al­iza­tion that are made to sup­port up to 32 users con­cur­rently. To speed things up, the M3 also houses a 1.15 Petabyte Lus­tre par­al­lel file sys­tem, which is es­sen­tially a stor­age de­vice that is ca­pa­ble of read­ing data at 24 gi­ga­bytes per peak sec­ond. To put that into lay­man’s terms: that’s ap­prox­i­mately four times the peak trans­fer speed of the cur­rent M.2 medium, which most high-end note­books and gam­ing PCs use.

How­ever, even when we went back to Monash Univer­sity for the an­nounce­ment of its new M3 su­per­com­puter, we never re­ally got the chance to check out the ma­chine or its hous­ing fa­cil­ity, on the grounds that the in­sti­tu­tion did not al­low pho­tos or videos of it to be taken.

HPC and Deep Learn­ing in the realm of sci­ence has long been a driv­ing force for aca­demic in­sti­tu­tions. These highly-com­plex fa­cil­i­ties and GPU-pow­ered su­per­com­put­ers have been re­spon­si­ble in help­ing sci­en­tists in medicine and fore­cast­ing weather pat­terns, cre­at­ing and re­con­struct­ing el­e­ments within the hu­man body, as well as an­i­mals. These in­clude the abil­ity to repli­cate – with ab­so­lute in­tri­cacy – in­ti­mate de­tails of the hu­man ner­vous sys­tem, sy­nap­tic pat­terns of a brain, or even a com­plete ren­der of mouse’s head.

The CAVE2 fa­cil­ity at Monash Univer­sity just amassed a new su­per­com­puter to play with.

(From left to right) Allen Fin­kle, Chief Sci­en­tist of Aus­tralia; Ian Smith, Vice Provost, Monash Univer­sity; and Christina Mitchell, Dean of Fac­ulty Med­i­cal, Nurs­ing and Health Sciences, of­fi­ci­at­ing the launch of the new MAS­SIVE 3 fa­cil­ity.

Sure, NVIDIA’s GPUs are the driv­ing force of the M3, but look at the num­ber of those In­tel CPUs!

Newspapers in English

Newspapers from Malaysia

© PressReader. All rights reserved.