975 D0 Preview

[JAG87]

Originally posted by: dmens

fact is D0 is only a packaging revision, and not a die revision

D0 is an all-layer stepping, which is a die revision, i guess. the latter is not the term i would use.

also, a packaging revision (pin swap?) is even more drastic and costly than a die revision and is to be avoided at all costs.

You are familiar with this document right?

Originally posted by: chizow
My take on it is that engineers and manufacturer's goal is to produce the best possible chip, period. They're not telling electrons to work 75% as hard on one wafer and Extreme overtime on another, or to mail it in once they finish enough of the L3 to meet whatever chip bin they're shooting for, or letting their monkey go home early once its picked enough shiny heat spreaders.

Sure there's going to be binning and chip modification to meet market segment demands, but one of the fundamental principals in overclocking is that demand is going to be very low on the high-end parts with a supply that should theoretically all be the same until validated and binned otherwise. I'm clearly hedging my bets that there's more "high-end" parts that could be sold as such, but are down-binned to meet demand.

That's misinformation. Every cpu is binned, but they are not binned by voltage like you think, they are binned by wattage. If a cpu does not fit the TDP envelope at 3.2G, it becomes a 940, if it does not fit the envelope at 2.93G it becomes a 920. The same way, a 975 needs to fit the envelope at 3.33G. TDP is largely determined by voltage, so the lower the voltage required the lower the TDP will be, for a 975's TDP to equal that of a 965, the VID has to actually be lower on the 975, because the clock speed is higher so the current draw is higher.

Originally posted by: aigomorla
but my conclusions are coming from 2 ES samples.

C0/C1 ES
vs.
D0 ES.

And if the models behave simular, then i have great expitations on the D0.

Did you forget my 965 was also an ES?
http://i125.photobucket.com/al.../aigomorla/Final-1.jpg

Again, the chip you have now is a better bin. It has much more to do with that, than the fact that it's D0.

Originally posted by: harbin
Just want to know, is my C0 920 LinX 20 run @4.0g with 1.19V Vcore any good comparing with your D0 result? Or I need do more to approve D0 is nothing new?

With HT on? I doubt it, I'd love to see a screenshot.

Originally posted by: OCguy

I think you are the one who may be wrong.

Even woldfale C1 ---> E0 made a big difference.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.

 

[aigomorla]

Originally posted by: JAG87

Originally posted by: OCguy

I think you are the one who may be wrong.

Even woldfale C1 ---> E0 made a big difference.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.

uhhhh JAG... this is a very dangerous statement my friend.

We have enough valid proof to validate that the E0 is >>>>>>>>>>>>>>>>>>>>>>>>> C1, and not by a small margin either.


if you guys are wondering if the D0 runs cooler, no.... i dont see any difference in temperature except it gets hot @ 4.5ghz with HT on. Well im still not breaking 70's.

 

[ghost recon88]

Not breaking 70s @ 4.5GHz with HT on is an improvement to me

 

[dmens]

Originally posted by: JAG87

Originally posted by: dmens

fact is D0 is only a packaging revision, and not a die revision

D0 is an all-layer stepping, which is a die revision, i guess. the latter is not the term i would use.

also, a packaging revision (pin swap?) is even more drastic and costly than a die revision and is to be avoided at all costs.

You are familiar with this document right?

lol never heard of it.

 

[harbin]

Originally posted by: aigomorla

Originally posted by: IanY
Get it to 4.5 Ghz and over (without touching the chiller) and I'll be very impressed and I'll order a 965EE immediately upon release. Otherwise, it overclocks to the same extent as C0, so what's the point?

Ima make you eat those words pal..

http://i125.photobucket.com/al...igomorla/Capture-9.jpg

Offtopic, Iany is a really good friend of mine i dragged onto this forum.

i didnt have time for testing, this was what i took last night,

i'll test it later today when i get back from work.

Is this 4.98g prime stable? I can also show that my C0 920 do 5.0g in windows.

Edit: my 5.0g is HT off, so D0 might be better if retail 920 D0 can do the same with HT on.

 

[dmens]

I can also show that my C0 920 do 5.0g in windows

LOL what a load of shit. shens. you using LN2? that doesn't count.

 

[JAG87]

Originally posted by: aigomorla

Originally posted by: JAG87

Originally posted by: OCguy

I think you are the one who may be wrong.

Even woldfale C1 ---> E0 made a big difference.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.

uhhhh JAG... this is a very dangerous statement my friend.

We have enough valid proof to validate that the E0 is >>>>>>>>>>>>>>>>>>>>>>>>> C1, and not by a small margin either.


if you guys are wondering if the D0 runs cooler, no.... i dont see any difference in temperature except it gets hot @ 4.5ghz with HT on. Well im still not breaking 70's.

Why do you always seem to misinterpret me nowadays.. We used to communicate well, you and I.

I'm talking about C0 to D0, not C1 to E0. I have no argument against C1 to E0.

Originally posted by: harbin

Is this 4.98g prime stable? I can also show that my C0 920 do 5.0g in windows.

Edit: my 5.0g is HT off, so D0 might be better if retail 920 D0 can do the same with HT on.

yea sure, pics or bs.

 

[harbin]

Originally posted by: JAG87

Originally posted by: aigomorla

Originally posted by: JAG87

Originally posted by: OCguy

I think you are the one who may be wrong.

Even woldfale C1 ---> E0 made a big difference.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.

uhhhh JAG... this is a very dangerous statement my friend.

We have enough valid proof to validate that the E0 is >>>>>>>>>>>>>>>>>>>>>>>>> C1, and not by a small margin either.


if you guys are wondering if the D0 runs cooler, no.... i dont see any difference in temperature except it gets hot @ 4.5ghz with HT on. Well im still not breaking 70's.

Why do you always seem to misinterpret me nowadays.. We used to communicate well, you and I.

I'm talking about C0 to D0, not C1 to E0. I have no argument against C1 to E0.

Originally posted by: harbin

Is this 4.98g prime stable? I can also show that my C0 920 do 5.0g in windows.

Edit: my 5.0g is HT off, so D0 might be better if retail 920 D0 can do the same with HT on.

yea sure, pics or bs.

This is totally brain fart of me. I meant to say 4.489g and 4.5g on air, not 4.98g and 5.0g. I need serious coffee right now.

 

[chizow]

Originally posted by: JAG87
That's misinformation. Every cpu is binned, but they are not binned by voltage like you think, they are binned by wattage. If a cpu does not fit the TDP envelope at 3.2G, it becomes a 940, if it does not fit the envelope at 2.93G it becomes a 920. The same way, a 975 needs to fit the envelope at 3.33G. TDP is largely determined by voltage, so the lower the voltage required the lower the TDP will be, for a 975's TDP to equal that of a 965, the VID has to actually be lower on the 975, because the clock speed is higher so the current draw is higher.

What's misinformation? I never said every CPU wasn't binned, I said they are theoretically the same until binned and validated otherwise. I also didn't say anything about wattage or voltage, but I'm sure you considered TDP on all desktop i7 parts is the same with a TDP of 130W. You can use the search there for 920 and 940 to see their TDP is also 130W.

Intel white papers have stated TDP isn't a hard and fast indication of actual power consumption, its a maximum threshold. Its obvious every i7 is going to be within that 130W envelope, but you will never know by how much which is why we use VID as a gauge of how it was binned.

Again, I'm hedging my bets that the average VID and max frequency on a D0 is going to be as good or better than the best VID on a C0. As others have stated, there's only mounds of historical evidence showing later steppings and production dates tend to produce higher performing parts. I think the late-run Q6600 G0s may have been the only recent exception.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.

You do realize some of the forum members who have contributed in this thread are Intel employees right? I don't even frequent the CPU forum that often and I picked up on this long ago.

 

[JAG87]

Originally posted by: chizow

Originally posted by: JAG87
That's misinformation. Every cpu is binned, but they are not binned by voltage like you think, they are binned by wattage. If a cpu does not fit the TDP envelope at 3.2G, it becomes a 940, if it does not fit the envelope at 2.93G it becomes a 920. The same way, a 975 needs to fit the envelope at 3.33G. TDP is largely determined by voltage, so the lower the voltage required the lower the TDP will be, for a 975's TDP to equal that of a 965, the VID has to actually be lower on the 975, because the clock speed is higher so the current draw is higher.

What's misinformation? I never said every CPU wasn't binned, I said they are theoretically the same until binned and validated otherwise. I also didn't say anything about wattage or voltage, but I'm sure you considered TDP on all desktop i7 parts is the same with a TDP of 130W. You can use the search there for 920 and 940 to see their TDP is also 130W.

Intel white papers have stated TDP isn't a hard and fast indication of actual power consumption, its a maximum threshold. Its obvious every i7 is going to be within that 130W envelope, but you will never know by how much which is why we use VID as a gauge of how it was binned.

Again, I'm hedging my bets that the average VID and max frequency on a D0 is going to be as good or better than the best VID on a C0. As others have stated, there's only mounds of historical evidence showing later steppings and production dates tend to produce higher performing parts. I think the late-run Q6600 G0s may have been the only recent exception.

While most of this is true, it doesn't mean that D0 will set a new standard for 920s. All we have to gauge D0 on is highly binned 975 chips, that's all I'm trying to say.

With G0 there was a highly advertised TDP reduction, which meant less voltage cooler temps, and higher overclocks. With D0 there is no such thing, even Aigo says temps are the same as C0, which shows you there is no TDP reduction, therefore the 3.33G within TDP is achieved by binning CPUs that can be stable with significantly less voltage. That's not the case with 920s, which is why I think most people that have been waiting since January because of the D0 announcement will be disappointed.

Intel doesn't say so, and frankly I believe them more than a handful of forum members with carefully picked 975 chips. I guess you'll find out when 920 D0s hit retail.
You do realize some of the forum members who have contributed in this thread are Intel employees right? I don't even frequent the CPU forum that often and I picked up on this long ago.

Originally posted by: dmens

lol never heard of it.

Yea and they don't even know what goes on at their work place so what's your point?

 

[TuxDave]

Originally posted by: JAG87
Yea and they don't even know what goes on at their work place so what's your point?

Well it's more of "how much does the customer really need to know". The thermal envelope is the same. They moved the serial number around. Customer doesn't need to know that we sped up some silicon speed paths to improve yield.

 

[chizow]

Originally posted by: JAG87
While most of this is true, it doesn't mean that D0 will set a new standard for 920s. All we have to gauge D0 on is highly binned 975 chips, that's all I'm trying to say.

With G0 there was a highly advertised TDP reduction, which meant less voltage cooler temps, and higher overclocks. With D0 there is no such thing, even Aigo says temps are the same as C0, which shows you there is no TDP reduction, therefore the 3.33G within TDP is achieved by binning CPUs that can be stable with significantly less voltage. That's not the case with 920s, which is why I think most people that have been waiting since January because of the D0 announcement will be disappointed.

Well again, like I said, its a bet based on historical trends seen with semiconductors and overclocking. If I had the choice I would always choose the latest-run model or stepping over older stock, especially when we already know what is possible with the older steppings.

Yea and they don't even know what goes on at their work place so what's your point?

Sounds like they were being sarcastic. Their non-sarcastic comments tend to demonstrate intimate knowledge of what changes were made in this stepping and it certainly sounds like its worth waiting (a few days) for.

 

[IanY]

When's this 975 EE on sale again? In a few days?

 

[rge2]

Like IanY said, my only interest in D0 would be 4.5+ stable with reasonable volts or 5+ benching with room temp water. My i940 will boot at 4.6 with 1.49v, and can bench all day at that speed with 1.65v. Prime/linpack stable 4.2 with 1.38 load as is though mine i940. Only interest in 4 or 4.2 using low volts would be if that trend continues to allow 4.5+ ghz stable.

 

[harbin]

Originally posted by: rge2
Like IanY said, my only interest in D0 would be 4.5+ stable with reasonable volts or 5+ benching with room temp water. My i940 will boot at 4.6 with 1.49v, and can bench all day at that speed with 1.65v. Prime/linpack stable 4.2 with 1.38 load as is though mine i940. Only interest in 4 or 4.2 using low volts would be if that trend continues to allow 4.5+ ghz stable.

+1

 

[Rubycon]

The ipod is strong in this thread.

 

[aigomorla]

okey some more updates.

Iany for linpack 4.5ghz stable you need around 1.44-1.45vcore.

@ that vcore it does get hot. With coolants kept at 22C, i was pulling 68ish with 71 as my hottest core... i think i might need a remount.... (i kinda hurried the mount).

Average H2O setups i would say would hit 70ish... Oh this is with HT ON. With it off, im sure i could of done it with less voltage.

 

[rge2]

4.5 linpack stable at 1.45 would be pretty sweet....be interesting to see what the masses get...but that is tempting.

 

[rge2]

oops double post

 

[harbin]

Originally posted by: aigomorla
okey some more updates.

Iany for linpack 4.5ghz stable you need around 1.44-1.45vcore.

@ that vcore it does get hot. With coolants kept at 22C, i was pulling 68ish with 71 as my hottest core... i think i might need a remount.... (i kinda hurried the mount).

Average H2O setups i would say would hit 70ish... Oh this is with HT ON. With it off, im sure i could of done it with less voltage.

indeed, it's sweet on 4.5g with HT on stable!

 

[geokilla]

Not to spread hate or anything, but a member from another forum got 4Ghz with just 1.2V on a Core i7 920 D0. He's not stable though since he's still trying to find the limit.

4+Ghz @ 1.2Vcore

 

[daw123]

Originally posted by: IanY
When's this 975 EE on sale again? In a few days?

Ballicom.co.uk (UK online supplier) have said that they will have the 950 in stock in the second week of April. I'm guessing the 975 will be released about the same time.

Provided I get a decent chip, do you guys think a 4.2 - 4.4GHz stable overclock on the 950 is plausible?

btw it'll be on water.

P.S. Nice o/c Aigo, although I have a feeling that your WC system is more high end than mine.

RE: Enquiry from Ballicom UK?
From: Dai Dattani (Dai@ballicom.co.uk)
Sent: 03 April 2009 09:55:59
To: #####

Hi David

We were told that we should have these for the second week in April

Thanks

Dai

--------------------------------------------------------------------------------
From: ####
Sent: 02 April 2009 18:30
To: Dai Dattani
Subject: RE: Enquiry from Ballicom UK


Hello Dai,

The part number is: 425324

http://www.ballicom.co.uk/not-...ail-boxed.p425324.html

Kind regards,


David Wilson

 

[Rubycon]

As always, your mileage does and WILL vary.

http://www.evga.com/forums/tm....100577171&mpage=1&key=

4.5 w/HT ON is definitely good. I just ignore temps as long as it's stable and below TJM.

 

[IanY]

My HT and Turbo are always on.

The question is whether there is a need to buy a 975EE to reap the benefits.

I want to step up to 18 x 250 (with Turbo and HT) with some 2000 Mhz ram (maybe Corsair, maybe Mushkin). I also want to use the Classified mobo.

-------------------------------------
Offtopic

A/N, my quad videos are going water right now (when I have bloody time). Got the blocks and everything else here with me. It'll look might impressive with four of them up at 730/1700/2800. I haven't chosen which sound card I want... they all sound pretty awful (xonar, auzen, creative.. and I bought them all.. and returned all) through the analog ports... digitized crap.. honestly, the onboard sounded no worse, and in fact at least had a less compressed soundstage.. Haven't bought the Classified yet because I have too many projects going on and not enough time.
-------------------------------------

 

[Rubycon]

TRUE temps LINPAK loaded