» »

Novi IBM-ov superračunalnik

SiliconValley.com - Podjetje IBM je oz. bo danes naznanilo rojstvo novega superračunalnika. Nova naprava z imenom Watson Blue Gene system je bila zgrajena v rekordnih dveh mesecih, uporabljalo pa jo bo 3.000 raziskovalcev v IBM-ovem Thomas Watson Research Center.

Maksimalna procesorska moč Watson Blue Gene je 115 teraflopov, kar pomeni 115 bilijonov kalkulacij na sekundo, njegova konstantna zmogljivost pa je 91,29 milijarde operacij na sekundo. Ti podatki postavljajo novonastali superračunalnik takoj za IBM-ov najzmogljivejši izdelek Blue Gene (maksimalno 135 bilijonov kalkulacij na sekundo), ki ga uporabljajo v Lawrence Livermore Laboratory.

Oba superračunalnika sta zgrajena modularno. Tako je sistem Watson sestavljen iz 20 modulov, v vsakem pa je po 2.000 PowerPC procesorjev. Kljub temu da so pri izgradnji uporabili nizkoporabniška vezja, naj bi Watson Blue Gene letno porabil za kar 400 do 500 tisoč dolarjev električne energije.

Če želite imeti omenjeni sistem doma, vam ga pri IBM izdelajo za pičlih 40 milijonov dolarjev, kar znese 2 milijona za en modul.

IBM očitno že ve kaj storiti z odvečnimi procesorji.

26 komentarjev

snow ::

kilo.
mega.
giga.
tera.

tisoč.
miljon.
miljarda.
biljon.

hm?
Random mutation plus nonrandom cumulative natural selection - Richard Dawkins

strictom ::

milijon
milijarda
...

:8)
"Violence is the last refuge of the incompetent" - Salvor Hardin

snow ::

:\
Random mutation plus nonrandom cumulative natural selection - Richard Dawkins

snow ::

Saj nebi blo tolk čudno, če bi vse 3 milijarde popravli. :) Ste pa sam dve.
Random mutation plus nonrandom cumulative natural selection - Richard Dawkins

M-man ::

Se oproščam za napako.
Trillion je prevod iz ameriške angleščine milijarda ali bilijon – v tem primeru bilijon.
Logično sklepanje me je pustilo na cedilu.:D
Alea Iacta Est. (Julius Caesar)

Keyser Soze ::

maljon
maljarda
....
:D

Hudiča so pa frajerji. Jaz imam probleme že z ENIM procesorjem. Tile pa kar par 10.000 procesorjev. Funny munny. Rama pa tudi najbrž da dol padeš v terabajte. Pri meni pa 768MB. Bom začel danes v prašička kovance metat. Morda kdaj našparam za eno tako zverino.

P.S.: Kaj mislite, koliko FPSjev pljune ven v DOOM3 na maks ločljivosti in z vsemi bonbončki. Al pa FarCry?:P

M-man ::

Čist možno da bi bila grafična ozko grlo.:\
Alea Iacta Est. (Julius Caesar)

snow ::

Če sploh ma grafično :)
Random mutation plus nonrandom cumulative natural selection - Richard Dawkins

Pyr0Beast ::

ene 2000 w software, kaj bi šele pri UT 98 >:D
Some nanoparticles are more equal than others

Good work: Any notion of sanity and critical thought is off-topic in this place

snow ::

Po drugi strani pa... takale zadeva najbrž simulira ene par PCjev z najhujšimi procesorji in grafičnimi karticami.
Random mutation plus nonrandom cumulative natural selection - Richard Dawkins

Thomas ::

> Po drugi strani pa... takale zadeva najbrž simulira ene par PCjev z najhujšimi procesorji in grafičnimi karticami.

Ja. Dovolj je enega. S 100 kratno hitrostjo? S 1000 kratno? I wonder.

PIPI ::

Grafika je software rendered.

pecorin ::

v racunalnistvu obstaja se kaj drugega kot igrice, overclocking in sminkanje ohisij..

Jeebs ::

In pol delaš celo življenje za račun za elektriko.:D
Press any key to continue... RESET. Didn't you say ANY key?!?!?!

B-D_ ::

v racunalnistvu obstaja se kaj drugega kot igrice, overclocking in sminkanje ohisij..

Včas tud crkne pa ga je treba tolčt... >:D

Drugač pa fensi mašinca... na njej pa ziher kakšen zanimivo razvejan PROLOG kraljuje :8)

Thomas ::

Fortran, IMO.

Malinovc ::

..pazi tole nevmesno..

apparently the playstation 3 does 2.18 teraflops (equivalent to about the 404th fastest super computer )
xbox 360 does 1.25 teraflops
a 3.6ghz P4 does 0.005 to 0.01 teraflops...
japans Earth Simulator with 5120 700mhz processors does 40 teraflops (3rd fastest super computer)
fastest is 135 Tflops...

..potem je tu se tole..

Well, if you watch the Linux Kernel development newsgroup, you can see a lively discussion going on, centered around an IBM programmer who is adding another architecture to the linux kernel, called the BPA. The Broadband Processor Architecture will shortly become yet another choice in the PPC processor family selection in the Linux Kernel config. BPA is the techie name for the Cell Processor.

Komantarji:

Keyser Soze ::

v racunalnistvu obstaja se kaj drugega kot igrice, overclocking in sminkanje ohisij..


Ne! A res!?8-O Like kaj? Avtomatično mazanje putra na spodnjo stran marmelade s kruhom? Al mogoče simulacija Coriolisove sile v WCju z vodo, ki vsebuje trdne delce?

P.S.: No shit Sherlock! Če sem malo bolj resen kot zgoraj.

xenon74 ::

Malinovc!
To je kao system FloatPoint performance in za PS3 znaša 2 teraflopsa. Pomeni pa seštevanje float point vrednosti CPU in GPU enot. Teoretični float point performance za PS3 znaša 218 gigaflopsov za CPU ter 1.8 teraflopsov za GPU. To je glede na podatke na njihovi strani.
Here is a rough analysis of the theoretical capabilities of the two processors.

Both processors achieve the bulk of their floating point performance by vector operations on 128-bit registers. Normally this means processing 4 32-bit floats at once; double the vector GFLOPs numbers for the (rare) case of 16-bit floats, halve them for the (somewhat more common) case of double-precision floats. The Altivec instruction set includes 'fused multiply-add', which is probably implemented by the vector units of all the processing cores in question. This doubles the theoretical peak GFLOPs, for the unrealistic case that every instruction is a multiply-add. In theory each processor will do a minimum of one instruction per cycle if the decoders are kept fed, but in practice the PowerPC architecture has multi-cycle latencies that delay serial execution of instructions that depend on each other. These are 6 cycles for FPU operations and 8 cycles for vector FP operations on the G5, probably similar on the two console CPUs (maybe less for the PS3's SPUs as they have a simpler design). Finally the CPUs in both consoles run at a clock speed of 3.2 GHz.

The Xbox 360 has 'one VMX unit per core', which if it resembles the G5 means one fully functional vector execution unit and one multiply/add unit, plus two floating point units. The G5 has a theoretical issue width of eight, so the Xbox 360 is unlikely to be decode limited. The theoretical streaming compute power (for single-precision values) is 2 x 3.2 x 4 = 25.6 GFLOPS for the vector units and 2 x 3.2 = 6.4 GFLOPS for the floating point units in each core. That gives a theoretical upper bound of 96 GigaFLOPS for the Xbox (172.8 with only fused multiply-adds or 16-bit data), less than 10% of the 'well over 1 Tflop'. In practice interleaving vector and scalar FPU code is hard, and IPC will be dragged down by branches and serial dependencies. The XBox 360's single threaded IPC will probably be comparable to an XServe G5 running scientific applications; perhaps a little over 1. Assuming that the SMT is quite a bit more efficient than Hyperthreading in the P4, a sustained IPC of 2 might not be unreasonable, giving a total system performance of about 75 GFlops for six threads of vector-heavy code.

The Playstation 3's cores are issue limited; they have a maximum sustained IPC of 2. For pure vector operations this again gives 25.6 GFLOPS, this time for each whole core. This gives a theoretical maximum system performance of 204.8 Gigaflops (409.6 with fused multiply-adds or 16-bit data). However this cannot be achieved even in principle because other program logic including load-stores must compete for decode slots with vector operations. Combined with the lack of the ability to reorder instructions, I would guess a single-threaded IPC a little lower than the Xbox, I would guess about 0.8. On the plus side this will allow SMT in the PPU to work a little better, achieving a joint IPC of perhaps 1.5. This gives a total system performance of about 90 GFlops for nine threads of vector-heavy code.

The results of this rough analysis; the PS3 has a little over twice the /theoretical/ compute power of the XBox360, but probably only about 0-50% more on realistic game code. However the PS3 also has a much faster memory interface, a very efficient intercore interconnect and probably a faster GPU (the exact details aren't available yet). As such 'twice as powerful as the Xbox 360' is probably fair as a rough description.

The 1 TeraFLOP Microsoft claims for the XBox 360 is an 'overall system performance' that most definitely includes the GPU. This is made clear on the official spec page on www.xbox.com. I haven't seen the power of the CPU alone stated anywhere. 200 gigaflops for the Cell sounds entirely reasonable as either (a) a theoretical peak value for normal operations or (b) a real benchmark using 16 bit values or fused multiply-adds. Sony claims 2 TeraFLOPs for total system performance, which implies a GPU twice as powerful as Microsoft's, but the relationship between theoretical peak FLOPs and actual performance is even more murky for GPUs than it is for CPUs. Regardless, 'the PS3 is twice as powerful as the XBox 360' still sounds right as an overall characterisation.

If anyone can see errors in the above analysis or has more specific information, please do correct me.


X74
http://setiweb.ssl.berkeley.edu/team_display.php?teamid=30319

neoto ::

theoretical capabilities :D

Teorija že, kaj pa praksa??

CCfly ::

Pravzaprav imamo v tem primeru marketing -> teorija -> praksa.
"My goodness, we forgot generics!" -- Danny Kalev

B-D_ ::

Ko bodo pognali isti izračun na vseh mašinah in izmerili čas bo pa dobljen realen rezultat. Vse ostalo so "teoretični" benchmarki znani tudi pod imenom: marketing.

Roadkill ::

2180 gigaFlops bi ps3 postavilo bolj okrog 230ega mesta. Me pa res zanima, koliko je povprečna računska moč...
Očitno rezultate preverjajo z enakim algoritmom.
Ü

Zgodovina sprememb…

  • spremenil: Roadkill ()

neoto ::

Pol se bodo pa našli freaki, ki bodo gor linux naložl pa kakšnih 100 ps3 dal v cluster.... :D

Malinovc ::

Nja...problem je tem zadevam ocenjevat hitrost, ko ni nobene tehnicne dokumentacije ipd. Govorimo o drugi arhitekturi, ki ima nov nabor ukazov ( zasledil sem nekje da celo 200+ ).
Sm pa zasledil, da ze obstajajo knjiznice, instruction seti, simulatorji... ki bodo "so" na netu dostopni vsem :)) . In seveda IBMovi razvijalci ze pridno delajo kernel patche :P . Za graficno pa se ze ve da bo nvidina in naj bila celo ~2x hitrejsa od NV40. Bomo videli .. je rekel slepi.
100xPS3 server....seveda ce bo poceni >:D

Monster ::

hehe če bo stvar sekala pa bo kolkotolko poceni .. zaka pa ne :) zadni cajt da se spremeni struktura procov lol :)
Ka zaboga...


Vredno ogleda ...

TemaSporočilaOglediZadnje sporočilo
TemaSporočilaOglediZadnje sporočilo
»

IBM gradi nov superračunalnik

Oddelek: Novice / Ostale najave
82509 (2184) DavidJ
»

Japonci rušijo rekorde

Oddelek: Novice / Znanost in tehnologija
283385 (2122) Pyr0Beast
»

IBM podira rekorde

Oddelek: Novice / Diski
232462 (1851) BigWhale
»

Filmska industrija gre po poteh glasbene

Oddelek: Novice / Ostale najave
221624 (1624) Bistri007
»

BlueGene/L je najhitrejši na svetu

Oddelek: Novice / Ostale najave
161596 (1596) noraguta

Več podobnih tem