gulftown had 50% more cores and was 20% faster. so the more cores increase the less advantage you have in clients. Most client applications have little benefit from > 4cores.Quote:
Originally Posted by zalbard
Printable View
gulftown had 50% more cores and was 20% faster. so the more cores increase the less advantage you have in clients. Most client applications have little benefit from > 4cores.Quote:
Originally Posted by zalbard
Quote from techpowerupQuote:
Originally Posted by SubZero.it
AMD claims doesn't equal to factsQuote:
Here, take some salt. AMD reportedly gave out performance figures in a presentation to its partners, performance figures seen by DonanimHaber. It is reported that an 8-core processor based on the "Bulldozer" high-performance CPU architecture is pitched by its makers to have 50% higher performance ........blabablaba
Didn't we hear crazy claims in the past about Agena being about 40% faster than intel quad core ?
Didn't we hear claims about Deneb being faster than yorkfield ?
I would never trust anything AMD says. If AMD confident about the performance of bulldozer, then they should let an independent reviewer to test it.
It's possible of course, albeit very unlikely. Keep in mind that this is all rumor. Don't go flaming AMD when you don't even know if they really said such a thing.
Then don't hang out in AMD threads then? You'll be happier.Quote:
Originally Posted by dartaz
No it was never Agena. AMD claimed they had 40% higher throughput in some select server workloads with Barcelona versus intel FSB based MCM server products.And it was true,in some select workloads that was the case.Quote:
Originally Posted by dartaz
Agena/desktop product was never mentioned in that context.
This time around AMD briefed some partners about Zambezi client performance.Totally different product,workloads and market segment.
I think Bulldozer's 50% in games/single threaded stuff doesn't come from cores, but rather turboing + a speed-oriented design.
Cant really compare them apples to apples anymore BD core to I7 core. They both take up aprox the same die area and both run 8 threads. Intel using HT to double the threads, AMD using there new modual shared cores.Quote:
Originally Posted by zalbard
Without more info on which benchmarks and actual results its hard to speculate over performance.
I still have a hard time to belive that, even for singelthreaded apps...Quote:
Originally Posted by Macadamia
In singel threaded apps the 950 has a freq of 3,3ghz and for everything else 3,16ghz.
If BD has the same IPC as nehalem it needs 50% more clock speed -> so 4,95Ghz... i cant belive that... If its IPC is 10% faster then it still needs 4,66ghz.
And if you look at SB which already had turbo up to 3.8ghz it even seems less likely...
This.Quote:
Originally Posted by EvilOne
I think everyone knows if you hand pick select benchmarks for each camp you can sway the performance differences greatly from side to side. Although any tidbits of information is always more than welcomed, we're still mostly playing the waiting game on determining ACTUAL real world performance numbers.
Amd's 4 modules vs Intel's 6 cores + 6 Ht threads , I think it will all come down to who can achieve more in a smaller die size. Let the battle begin :D
no. the i7 950 has 4 core + ht and the 4m bd is less than a 8-core cpu. so theyll meet somewhere inbetween.Quote:
Originally Posted by zalbard
i know what JF says. but one bd module |= dual-core.Quote:
Originally Posted by Hornet331
Each BD core is more potent than Thuban core,be it int or fp. It doesn't matter how you call BD core:full of half or whatever.It doesn't matter. IF performance is there it won't matter how people will refer to it. It will be branded as 8 core chip and will perform like an 8 core chip,beating Thuban and i7 in poorly and well threaded workloads bya big margin.It's all that matters.
Just what I said.. :p:
Have you heard anything about an 'AGLU'? There was a post on SA about it.Quote:
Originally Posted by informal
Yes. It's not public data so no discussion on that topic :).Quote:
Originally Posted by flippin_waffles
Yes I know.I should have just quoted you,it would be easier :D.Quote:
Originally Posted by Hornet331
This. Although if it outperforms a 980x I doubt it will be cheap. I am sure that AMD is tired of playing the budget CPU game.Quote:
Originally Posted by Hornet331
As before I remain cautiously optimistic about BD
Of course, fair enough. I was reluctant to even bring it up, but in the end my curiousity won out. :)Quote:
Originally Posted by informal
No problem we are all curious :D. It's still not clear what is the true nature of the units and whether or not they can be used for some(or any) ALU operations.There is a possibility they can be used. All speculation on my part of course :).Quote:
Originally Posted by flippin_waffles
It won't be long now,just a few months. AMD will finally have a core to carry the FX moniker again. It should be fun :).
Hey... a chance in the frozen wastelands for a serious mITX battle has jsut been rekindled. :D
heh, next months will be very interesting, I need April month, now :)
Show me where AMD made these claims.Quote:
Originally Posted by dartaz
Any rumors of Bulldozer 4x4 being at least considered? The SR-2 was pretty cool but I couldn't stand how unnecessarily large the form factor was. 7 PCIE slots was gratuitous.
If I needed to buy a special case just to fit a dual cpu overclocking board I would prefer they took a little inspiration from the Raven 2's reorientation of the board and make something like this.
http://img403.imageshack.us/img403/1...4bulldozer.png
Well one can dream...
WTF dude, nice microsoft paint job edit.... :down:Quote:
Originally Posted by mutantmagnet
what hes showing is interestingQuote:
Originally Posted by kuroikenshi
have the PCI slots point upward, and the ports still be in the back too. (even though your plugging things into 2 directions now
not everyone needs 7 expansion slots, so why not take away the first 3-4 and still let us throw in 2 GPUs and still have 2 CPUs