Asus/NCIX Competiton Entry - dejanh

[1MaNArmY]

Nice OC, hopefully my core i7 rig will be up shortly and I can get some tips from ya

[dejanh]

Very nice oc man!
I have 2000MHz memorys but it is tricky or even impossible to run them stock (9-9-9-27, 2000MHz). Do I have to increase QPI/Dram over 1.5V and Dram above 1.65V to manage it? It seems to be freq. who stop me not timings.....
ATM I run them 1800MHz 8-8-8-20, QPI/Dram 1.4V, Dram 1.65V

Running your memory so high also forces your Uncore (UCLK) to run 4GHz+. Some boards require high QPI/DRAM voltage to achieve this so what you are seeing can be considered "quite normal". I think you should try increasing the QPI/DRAM first to 1.45V with 1.65V DRAM and see how that goes. If it does not work, increaste the QPI/DRAM to 1.5V, then to 1.55V. Some 2000MHz DRAM modules actually set the QPI/DRAM to 1.65V when using the X.M.P. profile as far as I know. While this is technically outside Intel specifications there does not seem to be any apparent affect at least not over few months of use. Keep your QPI/DRAM voltage within 0.5V of your DRAM voltage.

Let me know what happens

Looking good dejanh, the micro res is one of my w/c staple components

Yeah, without it bleeding the system and refilling is very difficult. Also, the added volume of coolant does wonders in my setup. Without it I just do not get the same cooling effect.

Nice OC, hopefully my core i7 rig will be up shortly and I can get some tips from ya

No worries. Drop me a line anytime

[ReverendMaynard]

word. If you have a compact loop having that extra volume is clutch. For me, it's the bleeding aspect. I eff around with my loop often (this weekend again in fact) so the res really comes in handy.

[dejanh]

word. If you have a compact loop having that extra volume is clutch. For me, it's the bleeding aspect. I eff around with my loop often (this weekend again in fact) so the res really comes in handy.

Yeah, I actually went for the compact loop idea particularly because I did not want any unsightly components outside of the case or the need to mod anything on the case itself. Everything is fully internally contained and like you said fairly compact so the res really does wonders.

[scaramonga]

Nice pics dejanh

Did you fix the 80mm fan on with anything e.g. cable ties or something else?

Keep up the good work, lookin good buddy

[dejanh]

No, it is actually wedged in between the video card and my block mounting mechanism. Can't move at all I just need it to blow over the NB, nothing more. It really keeps the temps low though which is great. I was thinking to go with a 50mm which fits that heatsink but I decided not to since the CFM is very low on those fans.

[Helli]

Wooahh...4,73 GHz

It looks i'm to shy to push my 920 over 1,6 VCore . I'm able to boot into
Vista with 4,551 GHz and 1.57 VCore. For my daily distributed Computing needs
i'm running rock stable at 21x200 at 1,45 VCore.

Helli

[r1rhyder]

That's a decent overclock indeed, I guess living in VC that be some chilly water.

[ReverendMaynard]

That's a decent overclock indeed, I guess living in VC that be some chilly water.

He has the melting mountain water running directly into his loop That...be some chilly water

[Pisklink]

@ Dejanh

Thanks for your extensive explanation on your cooling setup. I never heard of such a cooling block with two TEC's attached to it. Me and my friend wanted to use the TEC to cool the CPU directly and use a watercooling block to cool down the TEC. Since this will only be an experimental rig we can use two power supplies in order to provide the TEC and other system components with enough juice.

But I was still wondering if you didn't had to isolate anything around the TEC's? Because I think we have to isolate the socket when we will cool the CPU directly with the TEC.

[dejanh]

Wooahh...4,73 GHz

It looks i'm to shy to push my 920 over 1,6 VCore . I'm able to boot into
Vista with 4,551 GHz and 1.57 VCore. For my daily distributed Computing needs
i'm running rock stable at 21x200 at 1,45 VCore.

Helli

Yeah, I'd say that over about 4.7GHz you need Vcore in excess of 1.6V unless you are running the 965 XE on Phase or LN2.

That's a decent overclock indeed, I guess living in VC that be some chilly water.

He has the melting mountain water running directly into his loop That...be some chilly water

Damn, that's a good idea! Why cool my rad with air when I can cool it with mountain water Don't be surprised if I follow up soon with some crazy cooling lol!

@ Dejanh

Thanks for your extensive explanation on your cooling setup. I never heard of such a cooling block with two TEC's attached to it. Me and my friend wanted to use the TEC to cool the CPU directly and use a watercooling block to cool down the TEC. Since this will only be an experimental rig we can use two power supplies in order to provide the TEC and other system components with enough juice.

But I was still wondering if you didn't had to isolate anything around the TEC's? Because I think we have to isolate the socket when we will cool the CPU directly with the TEC.

Yeah, you can do that. Typical application of TECs would be done in that way. Now moisture buildup can be a problem that way though it does largely depend on the power of your TEC and your ambient temps. Technically a powerful enough TEC can cause even frost to build up. In any circumstance, I do recommend insulating the socket and other electronics around it if you are planning on playing with more powerful TECs directly on the chip.

[dejanh]

I need some advice re: Uncore. I am trying to go for the memory OC record but even though I can get my hands on some nice DRAM with Samsung ICs that should be able to hit as high as 2300MHz or very close to I cannot seem to get Uncore higher than about 4.4GHz - 4.45GHz. My QPI/DRAM voltage does not want to go higher than 1.6V even if I set it higher (I tried up to 1.65V but it always reads out 1.60V). I also increased the DRAM voltage to 1.9V, and IOH/ICH to 1.41V each to no effect. Looks like the board won't run higher Uncore which basically seals my fate in terms of the memory competition

Any advice? I appreciate anything anyone can share...

[Godmyster]

Thanks dejanh for the info im going to try some of the settings, and where i can go

[Twist360]

Really interesting thread and results

[Pisklink]

Good question what to do to get your uncore stable at higher clocks. To be honest I haven't had my uncore above 3800Mhz yet. Therefore I'm totally not familiar with what settings you should use then. However, you could take a look in the R2E thread or on the Asus forum, although there are many noobs present while here are more experienced people available.

[Master_G]

Impressive results from chilled liquid, nice job getting that Coolit system really working (definitely think you should continue with the mods get make it reallly cold!)
As for the higher uncore, i think pisklink is right with checking out with other people with the board, got to be some settings and things you can try. Fingers crossed you haven't hit the board's limit.

G

[scaramonga]

No, it is actually wedged in between the video card and my block mounting mechanism. Can't move at all I just need it to blow over the NB, nothing more. It really keeps the temps low though which is great. I was thinking to go with a 50mm which fits that heatsink but I decided not to since the CFM is very low on those fans.

Thx for that buddy , I'll see if I can do the same as I'm pretty sure I have a few of those kicking around

[Helli]

....
My QPI/DRAM voltage does not want to go higher than 1.6V even if I set it higher (I tried up to 1.65V but it always reads out 1.60V).
...

Are you sure? Have you measure this with an accurate DMM?

After i read this, i tried it for myself and i can set QPI to 1,8V and my
very accurate DMM (0.05%) reads 1,73V. However - i'm unable to
boot into Windows with an higher Uncore of 4311 (needs 1,69V).
By the way - with an QPI VCore over 1,8V my Rig won't start and
one time i crashed my Bios with an QPI VCore of 1,9V

Helli

[dejanh]

Thx for that buddy , I'll see if I can do the same as I'm pretty sure I have a few of those kicking around

No worries. Let me know how the temps look like after you place the fan

[dejanh]

Good question what to do to get your uncore stable at higher clocks. To be honest I haven't had my uncore above 3800Mhz yet. Therefore I'm totally not familiar with what settings you should use then. However, you could take a look in the R2E thread or on the Asus forum, although there are many noobs present while here are more experienced people available.

Impressive results from chilled liquid, nice job getting that Coolit system really working (definitely think you should continue with the mods get make it reallly cold!)
As for the higher uncore, i think pisklink is right with checking out with other people with the board, got to be some settings and things you can try. Fingers crossed you haven't hit the board's limit.

G

Yeah, I looked around a bit, dug into Saaya's VTT Graveyard as well. To be honest I am not even comfortable trying to push the QPI/DRAM past 1.7V even if I could (which I appear not to be able to do anyway ) since there has definitely been cases when the chip just dies. Whether it is some oscillations in the board voltages or the high VTT directly affecting the chip I am not sure, but I do not think that I want to find out on a $800 CAD chip.

Are you sure? Have you measure this with an accurate DMM?

After i read this, i tried it for myself and i can set QPI to 1,8V and my
very accurate DMM (0.05%) reads 1,73V. However - i'm unable to
boot into Windows with an higher Uncore of 4311 (needs 1,69V).
By the way - with an QPI VCore over 1,8V my Rig won't start and
one time i crashed my Bios with an QPI VCore of 1,9V

Helli

I have done DMM measurements at several different QPI/DRAM voltages up to 1.56V and the DMM measurement was always at most 0.02V different than the reported QPI/DRAM voltage. I did not try measuring it past 1.6V though but I really have a feeling that they are no increasing past 1.6V since no tool reports the voltages higher than 1.6V once I cross the 1.6V mark. My board also has this strange problem with IOH voltage where it is effectively stuck at 1.21V measured until you bump the IOH voltage to 1.35V at which poing the voltage changes to 1.31V with a hefty 0.04V droop. Maybe the board has some electrical issues?

I will do a few more tests and report back on this tomorrow most likely. Maybe today if I have time.

[Pisklink]

You said that you can't push your QPI/DRAM any higher than 1.6 - 1.65v. But doesn't a R2E have some jumper you can switch to get the availability over higher voltage settings?

On my P6T Deluxe I know I have some jumpers I can switch to open up a higher voltage range for extreme overclocking of some components. Problem is I can't recall exactly which jumpers / components these are. Maybe only for increasing vcore and vmem. However a R2E has much wider range of settings, I guess so maybe your lucky and you have a jumper for increasing the QPI/DRAM voltage.

[Helli]

@dejanh

By the way - have you set "Extreme OV" to "Enabled"? Disabled prevent you from setting extrem Voltage Values.

Helli

[dejanh]

You said that you can't push your QPI/DRAM any higher than 1.6 - 1.65v. But doesn't a R2E have some jumper you can switch to get the availability over higher voltage settings?

On my P6T Deluxe I know I have some jumpers I can switch to open up a higher voltage range for extreme overclocking of some components. Problem is I can't recall exactly which jumpers / components these are. Maybe only for increasing vcore and vmem. However a R2E has much wider range of settings, I guess so maybe your lucky and you have a jumper for increasing the QPI/DRAM voltage.

@dejanh

By the way - have you set "Extreme OV" to "Enabled"? Disabled prevent you from setting extrem Voltage Values.

Helli

Wow, this is a good point. I will check this. I have completly forgotten that I can enable Exteme Voltages. I did not think though that setting 1.7V on QPI/DRAM would be extreme and I thought this applied only to trully excessive voltages that are like 1.9V+ or something but looking into this is definitely worth a shot. I'll keep you posted.

[dejanh]

Ok well, I tried what you guys suggested and it did not have any effect. Like my IOH voltage around 1.21V the QPI/DRAM voltage seems to get "stuck" around 1.607V. However, if I bump up QPI/DRAM to 1.7V or higher the actual QPI/DRAM voltage does increase though it has a massive 0.05V droop. At 1.8V on QPI/DRAM this droop turns into 0.08V droop. Strange behavior indeed. I did all the measurements with a DMM this time around.

I managed to POST with 4.611GHz Uncore only once but could not post again afterwards. I am thinking that in my testing this is a combination of DRAM issue as well as the chip and the board. One other thing that I noticed while testing Uncore is that the 1.65V triple-channel Corsair Dominator sticks that I have do not seem to do well with increased DRAM voltage and will infact cause channels to be lost. If I drop the DRAM voltage back down to 1.65681V then I get back all of the channels. However, in order to properly test max Uncore I need to increase both DRAM and QPI/DRAM voltages. Quite a conundrum if you ask me...

I don't want to go buy some high powered sticks only to end up still not being able to clock the Uncore higher and plain wasting money on them...

Finally, I noticed I can POST higher Uncore if my BCLK is higher which was opposite of what I was expecting. I redid the tests several times with BCLK at 207MHz and at 162MHz and using 207MHz I can post higher Uncore clocks than with 162MHz.

Bah! I'm at my wits end here...at least for the moment...any further ideas?

[Pisklink]

Ok well, I tried what you guys suggested and it did not have any effect. Like my IOH voltage around 1.21V the QPI/DRAM voltage seems to get "stuck" around 1.607V. However, if I bump up QPI/DRAM to 1.7V or higher the actual QPI/DRAM voltage does increase though it has a massive 0.05V droop. At 1.8V on QPI/DRAM this droop turns into 0.08V droop. Strange behavior indeed. I did all the measurements with a DMM this time around.

I think these high vdroops are becoming larger with each step because the electrical circuit gets less capacity to deliver the set voltage. Maybe you could try a v-mod?!?

I managed to POST with 4.611GHz Uncore only once but could not post again afterwards. I am thinking that in my testing this is a combination of DRAM issue as well as the chip and the board. One other thing that I noticed while testing Uncore is that the 1.65V triple-channel Corsair Dominator sticks that I have do not seem to do well with increased DRAM voltage and will infact cause channels to be lost. If I drop the DRAM voltage back down to 1.65681V then I get back all of the channels. However, in order to properly test max Uncore I need to increase both DRAM and QPI/DRAM voltages. Quite a conundrum if you ask me...

Maybe you should try to post with only one stick of memory if you didn't try yet.

Finally, I noticed I can POST higher Uncore if my BCLK is higher which was opposite of what I was expecting. I redid the tests several times with BCLK at 207MHz and at 162MHz and using 207MHz I can post higher Uncore clocks than with 162MHz.

I'm totally clueless at this point as well. I would expect that lower bclk would put less stress on every component connected to it and therefore providing more stability.

Did you already do a google search for extreme OC results on a R2E, performed with nearly the same cooling solution as you are using?