Yet, another :banana::banana::banana::banana: ... :ROTF:
http://i245.photobucket.com/albums/g...ere/xxx/W0.jpg
http://i245.photobucket.com/albums/g...ere/xxx/W1.jpg
...
Printable View
Yet, another :banana::banana::banana::banana: ... :ROTF:
http://i245.photobucket.com/albums/g...ere/xxx/W0.jpg
http://i245.photobucket.com/albums/g...ere/xxx/W1.jpg
...
pics not work for me :(
works for me, though, and I'm envious :D
OMFG I think I just wet myself.
Wait, that makes 12 or more Folding VM's with enough memory... :shocked:
When the task manager has to show two levels of cpu graphs, you know you've got something good under the hood.
So will the Westmeres run at lower clock speeds to accommodate the TDP requirement of 135W? Will 32nm get it all the way there? Or will 2.4ghz just have to be the new 2.66?
Sweet! Any major, or minor problem(s) running on the current X58?
With this kind of stock clocks I hope they at least treat it with an extensive turbo mode. Something like 2 bins per core-not-used would be satisfactory.
You ruin my post arrangement :rofl:
W3580 @ 4.2GHz
http://i245.photobucket.com/albums/g...x/Bloom_WP.jpg
W3580 @ 2.4GHz
http://i245.photobucket.com/albums/g.../Bloom_WPD.jpg
...
Very nice, seems like it might be some tweaks in there other than 2 extra cores, guess it's for making up for the slightly reduced clock speeds in apps that cannot utilize 4+ cores.
Are you gonna test overclocking next? :D
Man this is like Movieman gone wild :rofl:
This is where triple channel memory really comes in handy :)
Hopefully the supporting boards for these CPUs have better OC options than my Supermicro.
WTF is it running with 2.5GB of memory used? ;)
How soon should these be out to retail?
What the FLIP is with the NSFW images?
Perkam
Run 3D mark 06-s or Vantage.
How about "little" overclock? :D :D
Need full size non covered picture of your desktop, who gives a :banana::banana::banana::banana: about that I7...:D
Interesting...but useless because programmers are still stuck with most of the codes for dual core,so forget about 12threads...but looks so extreme :P
I can't fully agree with that. But for the most part, it's still true.
There's a "small", but substantial # of programmers who can use more cores and are begging for more - me included. :rofl:
More importantly, the increasing # of cores will force programmers to start getting their act together.
This is 3 chips on one board? How did you get a hold of gulftown chips this early?
both pics of the background and task manager = sexy almost too much for me to contain myself!
I can't say I fully agree with this, but we do still see quite a bit of laziness in the programming field. When I thread out code, I make it use as many cores as are available. Not everything can be processed in this fashion, however. Some times a process is just highly serial in nature, so it isn't fully fair to just blame the coders when programs seem to disappoint in this regard.
Ouch, me want!
Wanna see BOINC benchmark :D
thx for sharing! :D :toast:
hows the power consumption jc?
Ah jc, we missed you.
Glad your back, and even better with new stuff. :)
Yeap, this is the idea ...
And the different between them in the pics is obvious ... ;)
W5590 is this rite ...
http://i245.photobucket.com/albums/g...de/W5590_V.jpg
...
holy crap! Nice going JC!
when does westmere hit retail market, and does it work in current x58 1366?
ohhhhhh man... pr0n!
this is what i've been waiting for! i'll dive into 1366 for this cpu! :D
Don't forget that these type of systems are meant for workstation type work loads which often leverage their power not only through multithreaded applications but often many single threaded applications. Also people who work / have a hobby in the A/V field will obviously appreciate these cpus.
Now not everyone I've spoke to agrees with me on this but I for one find both Linux and Windows Vista / 7 to be more responsive and smoother on a system with more than 2 cores. Now the difference between 2-3 is a lot more than the difference between 3-4 in terms of general system performance but I still can notice it. I had been running my E8400 at 4Ghz for about 8 months and I'm back on my Q6600 which is at 3.2Ghz ( I didn't find the benefits of the higher clock dual core to be tangible enough to make up for the loss of 2 cores ) and the system has felt quite a bit better in comparison ( and this is based on everyday tasks to more strenuous activities )
I'm holding off on any platform upgrades until I see where these chips end up sitting price wise. Looking forwards to see how 32nm scales power wise.
I LIKE !!!! *drools*
I gotta ask an as*hole question. Will there be a different Mobo chipset for this or will it stay X58?
DONT SHOW US 3DMARK06 or Vantage :D
not sure if 06 would scale past 8 cores but damn it lets see :D
wow....6 cores 12 threads coming to desktop?
Me wanty!
I can't wait to replace my E5520's with some of these :drool:
WOW great CPU! Just curious, how did you get it? :D
Just curious as why they didn't raise the L3 cache up even more. Wouldn't that increase performance?
Don't think it would increase it much... Do you remember comparisons of Core 2 Duos with 2mb and 4mb of L2 cache? The difference wasn't big at all. Here they have 6x2mb basically, and 256kb of L2 for each core... If not all cores are equally loaded, then loaded ones get even more cache they can use. So don't think adding more would be beneficial. L3 at least.
JC what about that wallpaper :D
Yep, and Boinc bench :)
Intel's newest roadmap:
http://i279.photobucket.com/albums/k...3_intel_rm.png
It seems Gulftown will replace Core i7 975 Extreme and have a $999 price tag.
So this Core i9 is going to be out and about sometime early next year?
I must admit I am most impressed..especially considering you have an early ES Sample, no doubt things will improve come retail :)
Nice background too ;)
John
desktop roadmap is not very interesting for gulftown, i want to see the xeon roadmap.
It looks expensive.
I think it will be more practical if Intel release a i5-740 to replace Q8400 at the $176 price point.
They need more quad-core at the mainstream.
Only one 999$ priced 32 nm CPU with cores more than two, will be even in Q3 2010, a year after this moment? Where 32nm mainstream quads?
My guess is that your going to see Gulftown 6 cores for the single and dual socket boards in socket 1366 and then for the enterprise class stuff you'll see the 8 core Beckton EX but that should work in a dual socket board also but different socket than 1366..
That will be their flagship model to go up against AMD's server class systems
It will be interesting to see what happens.
Someone at Intel needs to lift the weight on the pendulum up a little bit. The tick-tock scheme is loosing time.
That's their Tick-Tock model.
And 32nm in Q1 '10 and Sandy Bridge in Q1 '11 is a little late compared with their initial promise, mind you. :p:
http://i279.photobucket.com/albums/k...l_ticktock.jpg
Want.
I think sooner or later, intel will release 32nm westmeres with only 4 cores from the salvaged 6 core chips. They could rebrand it as an i7 960 or something. What surprises me is why intel hasn't done this yet with the current failed i7 CPUs, but I guess dual/trip core on LGA1366 would tarnish the socket's 'high end' image.
X58s failed to boot ...
JC..so on whitch MB/socket you let the Gulftown boot and run your tests!
Great info (wallpaper is not bad too :D )
:slobber::slobber::slobber::slobber::slobber:
that's gulftown right?
A German site is claiming Gulftown isn't a native hexacore but 3 dualcore Westmere dies stuffed into a single package (with uncore on a separate die I presume).:shrug:
http://winfuture.de/news,48822.html
Can anyone confirm it or is this a complete fabrication on their part?
^
fabricated, since clarkdale has a PCIE controller on die and no external QPI link, but a DMI link. Gulftown needs at least one QPI link to communicate with the X58 northbridge or 2 for the DP version.
Though that doesn't mean that it couldn't be 3 dies, but it makes no sense. Since nehalem is modular and westmere ist just a shirnk they can add and remove cores as they like and make it all happen on one die.
Next rumor will be, that Beckton is only 2 Gainstown dies. :p:
Correct me if I'm wrong but I was under the impression Clarkdale's PCIe controller is a part of a separate uncore die together with the IGP and the IMC which are then linked to the core via QPI.
I'd like to believe Intel are sticking with their original plan of bringing a native hex to the desktop... but you know, with the sheer amount of transistors the Gulftown contains and the relative freshness (immaturity) of the production process, the projected yields could have been too low to make it worth their while. Or the gain in thermals from switching to 32nm was too low to make it work as a single die (Beckton? silly me:p:). There could have been a number of unforeseen circumstances (that we aren't privy to) that made them scramble and opt for a patch job.
Then again, they did publish die shots a while ago... Time for JC to pop the lid?:devil:
Just found this:
http://www.neoseeker.com/Articles/Ha...el_32nm/2.html
It looks like the 2C/4T die will be used to form the whole lineup by simple MCM process. Yeah,it still is not convincing evidence that Gulftown is not "monolithic"and "native" part,but still adds to the veracity of the German link.Quote:
The Gulftown variant of Westmere will package six cores (twelve threads) into one MCM (multi-chip module) - but other Westmere derived parts will instead package a dual core Westmere die with a GPU in the same package, thus bringing the GPU to the CPU. In fact, the two CPU core / four thread Westmere die will be used in various combinations, with and without GPU cores, to produce processors with 2/4/6 cores and 4/8/12 threads.
hehhee, somebody else paid attention :D
yeah i noticed that too :)
just think about it for a second...
so what, 3 dualcore cpus and a gmch on the same cpu package?
and how are they all connected to the gmch? via qpi?
that would mean the gmch needs 3 qpi links or they daisy chained the cpus, which is rather unlikely...
could intel do this? im sure, there might even be prototypes...
will they do this? unlikely... especially considering they showed some die shots of a monolithic die...
i think they and some others got confused...
the 6 and 8core parts are 1366 only and they are monolithic, most likely
they might be mcm, but surely not with gmch
there are 2 and 4 core parts for 1156 in 32nm, which will use a 45nm gmch, so yes, they will be mcm... whether the 4core igp parts will be 2 cpu dies and one gmch idk, but since gmch, as far as i know, only has 1 qpi link, thats unlikely...
so they probably misunderstood mcm and thought intel is just adding more and more cores on seperate dies. actually, as far as i know, its just the gmch and one cpu die, which is either 2 cores or 4 cores.
http://forums.techgage.com/showthread.php?t=4555
"After discussing the design with Intel though, we found out that like Dunnington, Gulftown will feature a similar design as Core i7, but will just be using Westmere cores at the base, rather than Nehalem cores. Simple explanation, really."
Westmere is the 32nm shrink of Nehalem. So in essence same arch just 45nm is Nehalem and 32nm is Westmere, and I can confirm as well that Westmere is native.
I think the wallpaper would be nice too ;)
John
Yes, Westemere is the codename of the new µarch as was/is nehalem for the current one.
Gulftown, Lynnfield, Clarkdale are codenames for the upcoming cpus in various core configuration base on either nehalem µarch (lynnfield) or westmere µarch (gulftown, clarkdale)