[Computex 2008] - Intel Section

Nemesis11

Power Member
bloomfieldx58oz7.jpg


Another interesting hunt over at Computex is definitely the Intel Nehalem and we managed to get our hands on a 2.66GHz Bloomfield running on a X58 Tylersburg board. Its 3DMark Vantage CPU score is pretty remarkable at 16334 which is about 45% faster than a similarly clocked Yorkfield.

http://www.vr-zone.com/articles/Intel_Bloomfield_On_X58_Board_Benchmarked/5825.html

É apenas um benchmark, o que não mostra a realidade completa.
Mesmo assim e se a imagem não for fake, este cpu parece dar um grande salto em relação aos actuais cpus.
 
multiplicador de 20....

0.8 V.

isto aqui basta soprar para cima dele a cada meia hora, right ?
 
Última edição:
Aquela cache L2 é que me está a estranhar. De resto parecem valores normais.

EDIT: Vendo melhor aquele CPU-Z ainda não suporta o Nehalem em condições. Alias no Nehalem nem vai existir o FSB como normalmente chamamos.
 
Última edição:
Pois é. Estes já têm o controlador de memória integrado não é? Tinha de reconhecer algo semelhante ao hypertransport certo?
 
Mas vai haver um aumento de performance de 46% clock por clock??

O CPU score e como referiram bem deve estar a ser influênciado pelo CPU ser capaz de correr 8 threads ao mesmo tempo.

Salvo erro, o melhor resultado no 3DMark Vantage, foi obtido com um sistema Skulltraill (2 CPUs QX9775).

Mas 46% até soa a pouco, já que os testes de CPUs do Vantage até estão bem optimizados e escalam bem mais quando se passa de 4 para 8 threads.

Só se o HyperThreading não é tão eficaz em comparação à uma thread física. Ou fizeram correr o Vantage para correr apenas com 4 threads para mostrar a diferença.
 
We had an early opportunity to "play" with a Bloomfield CPU and X58 board yesterday. Our first impressions are quite positive based upon a quick run through of several major benchmarks. Our particular CPU sample was running at 2.93GHz and initial performance in video/audio applications was simply superb. The board design and BIOS were just finalized for qualification testing so it is too early to tell just how well this platform will eventually perform. There has been a lot of speculation about Intel limiting overclocking and we do not have a final answer yet if this is true.

On our platform that featured triple channel DDR3, the processor multiplier was locked at 22 with a QPI bus setting of 133 to generate the 2.93Ghz CPU speed. Memory was running at DDR3-1066 with ratios available for 1333, 1600, and 1866. Dues to the early BIOS revision, memory was only stable at 1066 in triple channel DDR3 operation but we were told 1333 memory speeds should be available shortly. Intel will allow a Turbo mode that increases the CPU multiplier to 23 and that was all our board/BIOS combo was capable of yesterday. We could not raise the QPI bus speed past 140 without incurring stability problems.

The board manufacturers informed us they are working on ways to increase the CPU multiplier pass the single Turbo speed step and that QPI bus speeds should improve greatly once BIOS tuning is completed. While we wanted to provide clarification to the overclocking question, it is just too early to tell what the enthusiast can expect in the way of overclocking at this time. We will follow up later this week with some additional information after we have more time with the platform.

http://www.anandtech.com/weblog/showpost.aspx?i=456

O bus speed no cpu-z, deve ser a velocidade do QPI.
Em relação à performance, parece-me ser bom sinal ser tão boa numa board "alpha".
 
Última edição:
Fake...

133x20=2660mhz

Ali aparece 2643mhz

Cumps
Pequenas variações podem existir,além de que todos dizem que é verdadeira;)
More computex Nehalem
Esta agrada-me muito:D
foxconnrenai.jpg

DP versions:D
tyantempests7012.jpg


Here, you'll notice many mobos with 12 or even 18 DIMM slots. Odd number? No, it is simply two CPUs times three channels times three DIMMs per channel.(...)Overall, the boards seem production ready, the PC and dual-socket " Skulltra il 2" follow ups will support XMP enthusiast memory timings as well, and the new heat sinks and cooling systems for the biggie LGA1366 sockets are right there too.(...)Chris "Fugger"(queiiim?Só conheço Charles Wirth "FUGGER" big typo:007:), Shamino, Kingpin and the rest of the top overclockers at the Foxconn "Quantum Force" booth also seem to have good words in for the Nehalem's clocks - but first take a look at their Penryn E-0 stepping overclocks tomorrow at the said location. Can anyone say 6.5 GHz quad-core?
In summary, Nehalem is there. As the clock speeds seem to have dramatically improved to the point that Intel can launch a 3.6 GHz Yorkfield or Harpertown Penryns without affecting the Nehalem benchmarketing, we can expect the initial rollout real soon now.
Bons OC's por parte dos top o'cers e parece que o Nehalem é bom o suficiente até contra 3.6Ghz dos Penryn(óbvio que o Nehalem está low-clocked) e também pelo que diz ali pode vir aí um 3.6Ghz York(QX9780 multi 9:p)
 
Última edição:
Back
Topo