Gulftown (32nm 6-core 2nd gen nehalem corei9) photo & benchmark

[aigomorla]

Originally posted by: dmens
http://valid.canardpc.com/show_oc.php?id=745329

i *heard* it is on air...

im sorry that is not a 3066

And also @ 1.611vcore i call shins to air.

The 3066 EE aka i9 995 unless intel changes its code...
Dmens.. u should know this by now...

This is a 3066 Q3QR
http://img1.abload.de/img/4.8ghz-32mf204.png

4.8 @ 1.480Vcore.. Yummy!

nom nom nom... on air... and one will be mine soon....

 

[dmens]

^

for ES parts, the default frequency in cpu-z is often not indicative of anything, and the EE tag is added once all the bins are sorted out. in the early days, all parts are fused to a low frequency because the program is probably not ready.

that link was forwarded to me so i don't have any comments one way or the other.

 

[aigomorla]

Originally posted by: dmens
^
in the early days, all parts are fused to a low frequency because the program is probably not ready.

Oh but the Q3QR isnt that new..
Its just not widely known because everyone who got one was respecting the NDA.

EDIT: OMFG i got a timewarp...

 

[IntelUser2000]

The Turbo Mode on Gulftown isn't supposed to be impressive compared to Bloomfield because of the TDP. It'll be somewhat higher than Bloomfield, but nowhere near even Lynnfield levels.

For example, the 8 core Beckton has Turbo Mode of only +3 grades.

 

[imported_Shaq]

Come on that 4.8 isn't stable. lol That is a suicide run. What can it do on air and pass LinX and Prime?

 

[DisgruntledVirus]

Wow.....

So, it looks like it's *only* the EE version chips and there won't be any "mainstream" varients (even in the $500 realm).

 

[drizek]

Originally posted by: Idontcare

Originally posted by: nyker96
i think the whole idea of x58 is to be able to take in the 6core upgrade in the future, that was understood from the beginning. Gultown is the future upgrade path, i bet people will burn Intel down if gulftown require a new board. i wonder how far you can push a 6core considering the more you have, the harder to push the cores up. I think in the future there should be individual clockings on each core, so we can truthly get the max OC from a multi core cpu.

Isn't that what you can do with AMD cpu's already? Clock each cpu to its own multiplier?

Yes, and AMD Overdrive/K10Stat let you set individual clocks per core. This is useful if you are reaching the limits of the CPU, but it doesn't help too much with TDP.

If you want to control TDP, you need to have per core voltage settings. From what I'm hearing about Turbo on Clarkdale, it gets close to doing this by "turning off" one of hte cores, although I don't think it will let you run one at 800mhz or whatever.

 

[aigomorla]

Originally posted by: Shaq
Come on that 4.8 isn't stable. lol That is a suicide run. What can it do on air and pass LinX and Prime?

1.48vcore sucide run?

Since when???

My 975 does 4.4 /w HT ON @ 1.37vcore...

The i9 is 32nm not 45nm also... Keep note, we expect better OC potential from the die shrinkage...

Also im hearing about 20-30% FASTER clock per clock vs a 45nm.
(rumor... i'll probably test it when i get my gulftown)

Originally posted by: DisgruntledVirus
Wow.....

So, it looks like it's *only* the EE version chips and there won't be any "mainstream" varients (even in the $500 realm).

As i was telling dmens.. no...

Currently there are 2 variants.

A 2.4ghz version which you see everyone spaming..
And the ellusive EE edition i just showed.

I'll preview this chip when i get it soon and show it off...

Ummmm but 500 dollar hexcore from intel?
My response... PUAHAHAHAHHAHAHAHAHHAHAHAHAHHAHAHAHA

Sorry... its dreaming if u can get a 32nm hexcore for 500 dollars retail.

 

[imported_Shaq]

1.48v is really high for 32nm. I used 1.54v for the q6600(65nm) and now 1.408v for the 920(45nm). 1.408v is considered high for a 920. So 1.48v on 32nm is pretty darn high. I would expect 1.3v to be on the high side for Gulftown and it is probably about 1.1-1.15v stock.

 

[potato28]

:Q I might need to grab an X58 board and one of those when they release for a new editing PC.

 

[aigomorla]

Originally posted by: Shaq
1.48v is really high for 32nm. I used 1.54v for the q6600(65nm) and now 1.408v for the 920(45nm). 1.408v is considered high for a 920. So 1.48v on 32nm is pretty darn high. I would expect 1.3v to be on the high side for Gulftown and it is probably about 1.1-1.15v stock.

I bet you on my classy i could probably do it with sub 1.45Vcore.

As i said i dont like ASUS.... :X
(not gonna argue.. its a personal taste)

I also run enough cooling on my cpu loop to cool a nuclear reactor. :X
(kidding, but its the highest tier in watercooling you can get).

Also as i said if my 45nm 975 can do this:
http://i125.photobucket.com/al...enchmarks/newstats.jpg

i dont see why a 32nm gulftown wont get as close if not better.

 

[Shmee]

while six cores do sound nice, how about something that would be cheaper than gulftown, but be cheaper, like a 920, but 32 nm, making it a great OC'er? any plans for another 1366 quad core of that kind in 32nm?

8 threads seems plenty even for the future for me, i only use 4 right now, I just could use a higher overclock and faster chip. lower voltage will be nice of course too.

 

[imported_Lothar]

Originally posted by: Shmee
while six cores do sound nice, how about something that would be cheaper than gulftown, but be cheaper, like a 920, but 32 nm, making it a great OC'er? any plans for another 1366 quad core of that kind in 32nm?

8 threads seems plenty even for the future for me, i only use 4 right now, I just could use a higher overclock and faster chip. lower voltage will be nice of course too.

http://forums.anandtech.com/me...AR_FORUMVIEWTMP=Linear
Read that thread to find the shocking answer.

 

[Shmee]

interesting. Seems like by the time they are out, I should just wait for next gen. I may end up just getting a new mobo with a hydra first. Perhaps just forget about a proc upgrade for a few years.