Nehalem-EP......BLOOMFIELD

[Drwho?]

Something about signals between CPU/MEM/QPI being out of phase at higher multipliers? Or simply overworking the clock distribution circuits?

Less try to formulate it in simple words ... The statement below is true since the 486DX2, up to Core i7:

The frequency multipliers have their own limits, signal gets less clean when you get to very high ratio, this is physica laws... then, when the clock become "less square signal" , its propagation become less effective, detection of 1 or 0 become less accurate on the other side of the dice.
Over the years, Intel invented many many ways to avoid the side effects, the high frequencies of Pentium 4 generation pushed the limits. I am not allow to discuss those details here, but just understand that pushing the frequency of the base clock and increasing the Turbo ratio will give much better result than just increasing the Turbo ratio (plus TDP limit).

We got to keep the secret sauce of our processor secret ...

[Movieman]

Less try to formulate it in simple words ... The statement below is true since the 486DX2, up to Core i7:

The frequency multipliers have their own limits, signal gets less clean when you get to very high ratio, this is physica laws... then, when the clock become "less square signal" , its propagation become less effective, detection of 1 or 0 become less accurate on the other side of the dice.
Over the years, Intel invented many many ways to avoid the side effects, the high frequencies of Pentium 4 generation pushed the limits. I am not allow to discuss those details here, but just understand that pushing the frequency of the base clock and increasing the Turbo ratio will give much better result than just increasing the Turbo ratio (plus TDP limit).

We got to keep the secret sauce of our processor secret ...

Good explanation.
The question then becomes to what point will the turbo mode still allow increases assuming we cool the hell out of the cpu and I am still talking air and water.Let's face it, your cpu's are excellent but the stock HS leave something to be desired and I am not bashing you or intel, just stating a fact.
Lets assume for the sake of arguement that I can keep that cpu under 50C at full load. At that temp do you think or know if a sustained 4000mhz is possible?
I know, another tough question but the easy questions we already know the answers to!

[Booj]

DrWho? Can you say anything about cold bugging with Nehalems? Hopefully the guys from the overclocker mag asked you that hehe.. If you can't say anything, fair enough. Your posts in this thread have already revealed much. Thanks for stopping by

[Drwho?]

DrWho? Can you say anything about cold bugging with Nehalems? Hopefully the guys from the overclocker mag asked you that hehe.. If you can't say anything, fair enough. Your posts in this thread have already revealed much. Thanks for stopping by

answer is OC mag coming ...

[Movieman]

DrWho? Can you say anything about cold bugging with Nehalems? Hopefully the guys from the overclocker mag asked you that hehe.. If you can't say anything, fair enough. Your posts in this thread have already revealed much. Thanks for stopping by

The chips WILL run cold, very cold..

[Drwho?]

The chips WILL run cold, very cold..

it uses the same process technolgy as Penryn,plus some improvements ...
so, i let you guess

[EDIT]
We did make sure that the PCU does not get in the way of Extreme OC
[/EDIT]

[pajaa]

DrWho?, I understand that you are a busy guy at this time, so thanks for providing as much as information you are allowed to. Over at vr-zone is a preview of the Asus P6T Deluxe motherboard, which gives a strong warning not to raise the memory voltage higher than 1,65V, otherwise the CPU will be damaged permanently. Simple question: true or old information or can't you comment on this? (since the imc is in 45nm, it might be true, but i hope not).

Turbo mode is only for extreme editions, right?

[terrace215]

Turbo mode is only for extreme editions, right?

No.

[Shintai]

DrWho?, I understand that you are a busy guy at this time, so thanks for providing as much as information you are allowed to. Over at vr-zone is a preview of the Asus P6T Deluxe motherboard, which gives a strong warning not to raise the memory voltage higher than 1,65V, otherwise the CPU will be damaged permanently. Simple question: true or old information or can't you comment on this? (since the imc is in 45nm, it might be true, but i hope not).

Turbo mode is only for extreme editions, right?

Well..will should be maybe. And the same applies everytime you OC, increase vcore etc etc. Anything not running at stock=chance of CPU, memory, Board whatever to be permanently damaged.

And where is the strong warning in the link? I skimmed it over and I didnt see any warnings or anything about 1.65V.

And the only thing on that page with 65 in it is:

vNB-PCIe in 65-step chipset-PCIe bus voltage control

[T_Flight]

Is keeping it below 1.65v a bad thing?

I think what is getting people into trouble here is they are trying to go by what they've learned about older technology and don't know how this new technology works so they aere filling in the blanks with guesses.

Nobody can assume anything from that voltage level. That may very well be some rediculously high voltage that will never be needed. They probably put that there for the person who likes to come out the gate and just crank stuff up based on what they were "used to".

The only things we know about this CPU are what those that have worked on it, and worked with it have told us.

[pajaa]

No.

That's great news.

Well..will should be maybe. And the same applies everytime you OC, increase vcore etc etc. Anything not running at stock=chance of CPU, memory, Board whatever to be permanently damaged.

And where is the strong warning in the link? I skimmed it over and I didnt see any warnings or anything about 1.65V.

And the only thing on that page with 65 in it is:

Its in this picture here.

There are many DDR3-kits, which need 1,7V oder more to reach the advertised speeds and latencies or in other words: to run stock speeds. And we all want to get the best out of nehalem, or not?

[informal]

Its in this picture here.

There are many DDR3-kits, which need 1,7V oder more to reach the advertised speeds and latencies or in other words: to run stock speeds. And we all want to get the best out of nehalem, or not?

So it's official.It's Asus sticker alright and it lists tech paper from intel that states the 1.65V VDimm mark.Still it's a bit strange to have such a low limit knowing many stick need 1.7+V to work.
It's time for some low voltage DDR3 kits to arrive on the market.Jedec even approved the LowV DDR3 standard recently.

[Drwho?]

DrWho?, I understand that you are a busy guy at this time, so thanks for providing as much as information you are allowed to. Over at vr-zone is a preview of the Asus P6T Deluxe motherboard, which gives a strong warning not to raise the memory voltage higher than 1,65V, otherwise the CPU will be damaged permanently. Simple question: true or old information or can't you comment on this? (since the imc is in 45nm, it might be true, but i hope not).

Turbo mode is only for extreme editions, right?

I don't think it is accurate. Let's look at the logic behind memory controler...
Before, the memory controler was out of the CPU dice, so, if you burn a X48 ... it is only few 100 $ ... Now, the memory controler in on dice, so, if you burn, it will be expensive ... if using an extreme edition ...

So, electricly speaking, you want to pay attention to the laws of physics ... You got to make sure you align your memory voltage with the memory controler voltage ... I saw already people with illegitimate sample (grrrr) asking why their killed their CPU ... The reason is simple, if you put 1.5Volts on the memory controler, and 1.7 or 1.9V on the memory ... there is an obvious difference of electric potential ... and it will @#$!@#$*!#$

Any "non sunday overclocker" has a little understanding of physics laws, if you connect 2 wires with a different voltage ... you get electric courant.
In the case of Nehalem, you have to pay attention to this, don't connect 2 wires with different voltage together: Make sure your memory controler setting is not too far from the memory voltage.
On X48 , X35 and all the other, it was true too. People just burned chipset and did not really understood what was happening. This time, I come forward, and clarifie the laws of physics with the press, when I did meet them in Europe, USA and asia.

by the way, http://www.theinquirer.net/gb/inquir...s-qimonda-ddr3 should be enough for 99% of you guys

pay attention to physics laws

[edit]
I forgot to add, you all overclock at your own risk ....
[/edit]

[informal]

The difference in voltages between IMC and memory is a problem,but remember Agena core is also featuring IMC,although a DDR2 ,and voltage difference is a lot higher (1.15V for NB VS 1.8-2.4V for memory).That's 1+V difference right there.
Early B2 Agena cores did have problems with 2.2V+(some reports of dead CPUs) but B3 stepping has no such issues.

I guess with newer DDR3 memories this will become a non-issue on Nehalem platform.

[T_Flight]

Guys he's right. We're dealing with electricty, and that's what I've always studied the most. Things like loadline, and load balancing, voltage delta's between what is coming IN from the PSU and what is going to the socket (vDroop). I actually used to measure this stuff on my old IC7-G directly via a test point soldered to the board. I did vMods. Oh the horror...

Lemme tell 'ya, when i started doing those vMods, nothing will get the adrenaline flowing faster than when you push that power button the first time after doing a mod to your mobo!

I hope the warning he gave we all already are aware of. There are NO guarantees with OC'ing. It's gotten much easier and some take it for granted, but if you do it, you risk your hardware. It used to be said "If you don't have the money to replace your enetuire system don't do it at all." Some people used to go on to say "Do not come back here and blame me if you let the smoke out of your hardware. If you OC, it's YOUR risk. Take any info I post at your own risk."

The "feeling" I'm getting (and I know these guys are under strict NDA's) is that this CPU is something very different than we are used to. I also feel like there are mnay things yet to be revealed about it. They have stated point blank though that OC'ing has been taken into consideration at many levels, so I'm assuming it's gonna be another great CPU.

I'm glad I waiteds before buying memory now though, and am gonna follow those new JEDEC standards if that's what it takes. I love this design. It has alot of forward thinking to it. It's a completely different kind of running.

[pajaa]

@DrWho?
Merci beaucoup.
The only drawback with the qimonda memory is the green pcb. (and I haven't got a window in my computer case, but I know it's there)

T_Flight is right: It is a great CPU and a great platform, too.

And yes: overclocking is risky and have to be done carefully, and no one can blame someone else than himself for mistakes and blowing up his hardware, but it's fun. anyway

[Shintai]

That's great news.



Its in this picture here.

There are many DDR3-kits, which need 1,7V oder more to reach the advertised speeds and latencies or in other words: to run stock speeds. And we all want to get the best out of nehalem, or not?

So, it does say MAY and not WILL damage?

And How is Intels policy on OC? Overvolt? Precisely the same. This is just an extention of what we always got the last 10 years or so...

[Blauhung]

And also remember, I believe the on die memory controller is also on a separate power plane and can be adjusted separately from the cores and QPI controller. This allows you to tune your MC voltage to match your memory voltage.

This also has a unique end result for the good of the environment. With this platform, if you are forced with a choice between 2 sets of DDR3 operating at similar speeds, the more overclocking friendly memory will be the one that works at lower voltages as this allows the memory controller to relax the voltages. This will be a strong driving force in the enthusiast grade memory manufacturers to provide sticks that work at competitively lower power requirements.

[Hornet331]

On X48 , X35 and all the other, it was true too. People just burned chipset and did not really understood what was happening. This time, I come forward, and clarifie the laws of physics with the press, when I did meet them in Europe, USA and asia.

i never saw a chipset burned by high mem voltages, even beyond 2.3V i never heard of a dead X48/P45, fried memsticks on the other hean are far more common then dead chipsets.

[LowRun]

Lemme tell 'ya, when i started doing those vMods, nothing will get the adrenaline flowing faster than when you push that power button the first time after doing a mod to your mobo!

Time to shake that life a little dude

[Orthogonal]

All right, it's Q4 '08 now, where is Bloomfield!!!

[Hornet331]

All right, it's Q4 '08 now, where is Bloomfield!!!

Q4 lasts till 31.Dez.

[AuDioFreaK39]

Most important part here. Out of the box performance in more important than turning of stuff for theory and what ifs. It is far easier to work with something than to work against it. In this case it'd seem maxing out turbo tweaking it to the edge (over the edge = until its not stable anymore) would make all the sense in the world.

Go back to August 2006 (on this forum) and look at all of the features turned off or disabled to get better overclocks on the first C2D's? SO its not like folks aren't use to killing features to get better overclocks. IMHO, that's kind of counter productive

No matter how nicely said, some will always look for chinks and blemishes in the something shiny and new

So you're saying the best way to overclock Nehalems would be to max out turbo and "tweak it to the edge" (while maintaining good clock distribution over the dice area as Drwho? stated)? Makes sense.

[Revv23]

Guys he's right. We're dealing with electricty, and that's what I've always studied the most. Things like loadline, and load balancing, voltage delta's between what is coming IN from the PSU and what is going to the socket (vDroop). I actually used to measure this stuff on my old IC7-G directly via a test point soldered to the board. I did vMods. Oh the horror...

Lemme tell 'ya, when i started doing those vMods, nothing will get the adrenaline flowing faster than when you push that power button the first time after doing a mod to your mobo!

I hope the warning he gave we all already are aware of. There are NO guarantees with OC'ing. It's gotten much easier and some take it for granted, but if you do it, you risk your hardware. It used to be said "If you don't have the money to replace your enetuire system don't do it at all." Some people used to go on to say "Do not come back here and blame me if you let the smoke out of your hardware. If you OC, it's YOUR risk. Take any info I post at your own risk."

The "feeling" I'm getting (and I know these guys are under strict NDA's) is that this CPU is something very different than we are used to. I also feel like there are mnay things yet to be revealed about it. They have stated point blank though that OC'ing has been taken into consideration at many levels, so I'm assuming it's gonna be another great CPU.

I'm glad I waiteds before buying memory now though, and am gonna follow those new JEDEC standards if that's what it takes. I love this design. It has alot of forward thinking to it. It's a completely different kind of running.

Lol are you sure you are in the right forum buddy?

[Revv23]

i never saw a chipset burned by high mem voltages, even beyond 2.3V i never heard of a dead X48/P45, fried memsticks on the other hean are far more common then dead chipsets.

Ever blow an A64 mem controller? Was pretty easy back in the BH-5 days, i know i smoked two cpu's before i learned my lesson. I assume it's like that.

We usually don't blow our intel chipsets because we pile the volts on to those as we ramp up the memory usually as it helps the FSB go up... but it also protects the chipset from crazy differences between voltages...

I always make it a rule when im overclocking to clock everything at once. When i was learning overclocking I alway found the max of each component individually.. like i found my cpu max then my ram max then my fsb max etc.... Now adays I put everything to my max comfortable voltages and raise each component a few mhz between each testing cycle.