***Asus Rampage III Extreme Owners Thread***

Quote:

---

Originally Posted by [200]

On another note there must be a clockgen that controls the QPI PPL voltage. Does anyone know where on the board it is? Perhaps a variable resistor might be put in place to increase voltage like on video cards for example. I am not saying it would work but just an idea...

---

+1 :up: It would be good to know, if we can't get it in BIOS.

Quote:

---

Originally Posted by AceBaran

That is the thing Don, it is the lowest i can go with vtt, anything lower and i get bsod 124 (uncore = 4092) and raising RAM i need even more vtt.. my chips memcontroler sucks xmas balls >_<.. but im in a early state of testing this bios.. so, now i know R3E CAN do it.. its all in the settings.

---

Well, it's a pretty high UnCore clock (4096MHz), so it needs much VTT. IMO, it's too much for daily use. I would search for maximum UnCore clock while keeping VTT around 1.4-1.425V (or maybe up to 1.45V).

Quote:

---

Originally Posted by Brama

...and in my system it broke the S3 Suspend to RAM wake up. :(

As it seems you have a preferential link with Asus BIOS development, what's about the user requested QPI PLL voltage settable and separated from CPU PLL?

---

nothing @all @the moment

if this option comes in further bios


my opinion NO

Quote:

---

Originally Posted by Fire4Wire

nothing @all @the moment

if this option comes in further bios

my opinion NO

---

And what about this:

Quote:

---

Originally Posted by [200]

On another note there must be a clockgen that controls the QPI PPL voltage. Does anyone know where on the board it is? Perhaps a variable resistor might be put in place to increase voltage like on video cards for example. I am not saying it would work but just an idea...

---

Is there any way to know (from the inside sources) anything about measuring points for QPI PLL?

Quote:

---

Originally Posted by Brama

I had a negative experience with 0095.
Pushing an overclock a little over stability edge, caused the impossibility to reboot, even unplugging AC from mains. When you pushed the ON button on the PC, it started for about one second and after it shot down by itself.
Solved removing the cmos battery for a while and clearing cmos with the rear button.

Back to 1102 for now.

---

you do not test the 0011 bios?

to compare it with 0095 bios

How is this board with high baseclocks and 3d? My old r2e was pretty terrible, but I hear conflicting info about this board. Does it have the same 217 ceiling the r2e had?

Quote:

---

Originally Posted by cky2k6

How is this board with high baseclocks and 3d? My old r2e was pretty terrible, but I hear conflicting info about this board. Does it have the same 217 ceiling the r2e had?

---

Good REX2 has not these low BCLK wall.
I tesed 5 REX2, bad one faces BCLK wall under 210MHz, but good one over 235Mhz.
Mainly it depend on IOH quality, not mobo design.
But I think that REX3 average BCLK limit improve than REX2.
3D performance good same as REX2.

Anyway, I think UD7's average BCLK limit more better than REX3.
I bought 3 UD7(rev2) & these mobo BCLK limit are over 235MHz.
But UD7's memory performance is not good as REX3.
(3D performance is good as REX3.)
So you need consider what is more important for your OCing.

Only BCLK & 3D performance, I recommende UD7(rev2).
But you need memory & 3D perfprmance, I recommende REX3 or E760.
(Sorry, I had not test E770, yet.
so I do not know E770 is good mobo or not.)

I hope you will get good mobo.:up:

Quote:

---

Originally Posted by kurumi_Japan

Only BCLK & 3D performance, I recommende UD7(rev2).

---

The downside of UD7 rev2 is the number of phases on VRM that is lower than rev1.

A real headache to chose the right one. :)

Quote:

---

Originally Posted by Brama

The downside of UD7 rev2 is the number of phases on VRM that is lower than rev1.

A real headache to chose the right one. :)

---

I hear UD7 rev2 power module are re-designed & it has enough power.
So you do not need worried about phase number.

I tested UD7 rev2 under LN2, it's very stable & can pass over 6.0GHz many benchmarks.
I think the stability is almost same as REX3.
But a little bit slower memory speed than REX3.
So it's not good for Pifast & Super PI.
Especialy SuperPI 32M is very slow, almost 10 sec behind than REX3.
wPrime is no problem.

If you have chance, try UD7 rev2.:up:

Im really starting to hate R3E boards...

My second R3E just died after i good weekend of benching:mad:

There was no condensation on the board and it died anyway.

Hopefully the new Gigabyte UD7 board will do the trick:)

Quote:

---

Originally Posted by donmarkoni

+1 :up: It would be good to know, if we can't get it in BIOS.

---

I second this. I've got a ton of pots sitting around. I don't see why ASUS won't release information like this.

I don't see why they won't release information on how a BIOS setting changes other settings in the background either.

Quote:

---

Originally Posted by donmarkoni

Well, it's a pretty high UnCore clock (4096MHz), so it needs much VTT. IMO, it's too much for daily use. I would search for maximum UnCore clock while keeping VTT around 1.4-1.425V (or maybe up to 1.45V).

---

I'm running 12GB 2000MHz 9-9-9-25-1T with 3x4GB mushkin sticks. For stable operation at 143 BCLK I need 1.4875v vQPI.
However, for 200BCLK I do not have to increase vQPI at all for 2000MHz settings. 210BCLK my chip can't run uclk so high, but only needs 1.5125 vQPI with 1683MHz 7-8-7-25-1T.

Quote:

---

Originally Posted by Rasmus66

Im really starting to hate R3E boards...

---

Join the club:ROTF:

It seems to be a rather large club. I have some killer bench runs on this board at the expense of a full time shrink to baby sit its moody times.

QPI PLL = CPU PLL on R3E, same source
ie, you adjust 1.45v CPU PLL = 1.45v qpi pll as well, plane is tied together so there is no way to seperate.
It did not have an effect on oc for qpi pll last i tried ; what are you trying to accomplish with qpi pll? :)

@LardArse, Maybe you can work your magic with this board like you did the Classified. Or at least make sure the Rex4 doesnt suffer from some of the issues many had with this board. Mainly high bclk for 3D benching.

Quote:

---

Originally Posted by LardArse

QPI PLL = CPU PLL on R3E, same source
ie, you adjust 1.45v CPU PLL = 1.45v qpi pll as well, plane is tied together so there is no way to seperate.
It did not have an effect on oc for qpi pll last i tried ; what are you trying to accomplish with qpi pll? :)

---

So, you say that QPI PLL (which default is 1.1V) is set at 1.8V, when CPU PLL is left at default (1.8V). It would be insane IMO. :eek:

Quote:

---

Originally Posted by LardArse

QPI PLL = CPU PLL on R3E, same source
ie, you adjust 1.45v CPU PLL = 1.45v qpi pll as well, plane is tied together so there is no way to seperate.
It did not have an effect on oc for qpi pll last i tried ; what are you trying to accomplish with qpi pll? :)

---

I was hoping you would post :)

It's my understanding that it assists in higher bclk stability.

Quote:

---

Originally Posted by donmarkoni

So, you say that QPI PLL (which default is 1.1V) is set at 1.8V, when CPU PLL is left at default (1.8V). It would be insane IMO. :eek:

---

A terminology confusion

VTT = QPI/DRAM voltage = default 1.1v

QPI PLL and CPU PLL default is 1.80v.

Quote:

---

Originally Posted by [200]

I was hoping you would post :)

It's my understanding that it assists in higher bclk stability.

---

No, that is not what assists bclk stability

Interesting, thanks again for replying.

Would you know any possible reasons why a few owners of this board struggle to reach the same type of QPI link speed as compared to other manufacturer boards like the Classified or UD range in the Gigabyte line up? I've noticed a few people mention that the IOH chips might not be as binned as well as they could be but would like to think that it might be for other reasons such as voltage settings, cooling or even the CPU itself...

Very interested in getting your opinion & thanks once again.

Quote:

---

Originally Posted by LardArse

A terminology confusion

VTT = QPI/DRAM voltage = default 1.1v

QPI PLL and CPU PLL default is 1.80v.

---

I didn't misunderstood you, AFAIK QPI PLL default is 1.1V.

Quote:

---

Originally Posted by donmarkoni

I didn't misunderstood you, AFAIK QPI PLL default is 1.1V.

---

on which brand's naming convention?

Quote:

---

Originally Posted by LardArse

on which brand's naming convention?

---

ok Lardarse look at this guys postin the 4.5ghz stable club.... you can clearly see that the default for QPI PLL is 1.1V
http://img253.imageshack.us/img253/7221/linx204600.jpg

Uploaded with ImageShack.us

Quote:

---

Originally Posted by xdave

ok Lardarse look at this guys postin the 4.5ghz stable club.... you can clearly see that the default for QPI PLL is 1.1V
http://img253.imageshack.us/img253/7221/linx204600.jpg

Uploaded with ImageShack.us

---

ah ok, i dont know if its just a naming clash, i meant what is called 'qpi pll' on evga MBs default is 1.80v

On the EVGA board IOH PLL is set 1.80 default, not QPI PLL. That is default at 1.1 I believe.

hi default spec'ed by intel is 1.80v for at least what is called by the same name on ASUS. So if want to lower , have to lower cpu pll.

I think what is confusing is that on other boards, like EVGA or Gigabyte raising QPI PLL is needed to achieve high bclk. My UD7 or Classified 3 will not run stable at high bclk with QPI PLL set to low. The MSI XPower doesnt have a setting for QPI PLL but it doesnt seem to effect bclk when using fast mode. I think it goes back what was posted many pages back, if you get a board with a good chipset you will be able to do higher bclk, if not your bclk limit will be much lower in fast mode..

IMO, zoson's theory is most probably right, if not absolutely.

Quote:

---

Originally Posted by zoson

So everyone knows... In case it wasn't common knowledge.

QPI PLL is linked to the CPU PLL setting in BIOS on the R3E.

I asked how it was controlled to ASUS techsupport and that's the answer I got.

I've asked them for a table of values showing what QPI PLL voltage is set per different CPU PLL settings.

It seems 1.89v CPU PLL is probably between 1.14 and 1.16v QPI PLL. Raising my CPU PLL to this value gives me about the same stability/voltage settings that I got on my friend's x58a-ud3r with my chip.

---

I think that QPI PLL loops 2-3 times from lowest to highest in one loop of CPU PLL voltage from 1.2V to 2.1V.
One thing is sure, QPI PLL is 1.1V when CPU PLL is 1.8V and probably when CPU PLL is lowest 1.2V QPI PLL is 1.1V or even 1.0V.

Could be something like this:
CPU PLL | 1.2 | 1.3 | 1.4 | 1.5 | 1.6 | 1.7 | 1.8 | 1.9 | 2.0 | 2.1 |
QPI PLL | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 |

or maybe not so linear, IDK :shrug:

Tnx Donmarkoni for your guidance :)

More stable now @ lower QPI and volts

http://img208.imageshack.us/img208/6397/49899873.jpg

I'm getting a i7 970! Maybe I'll do better with 32nm.
:D

With 32nm chips, whether it was the 980x or the 970 I have had better success with dropping CPU PLL from 1.8v to 1.42v. I couldn't get my 980x stable at 4.2ghz until I did this.

Quote:

---

Originally Posted by donmarkoni

IMO, zoson's theory is most probably right, if not absolutely.



I think that QPI PLL loops 2-3 times from lowest to highest in one loop of CPU PLL voltage from 1.2V to 2.1V.
One thing is sure, QPI PLL is 1.1V when CPU PLL is 1.8V and probably when CPU PLL is lowest 1.2V QPI PLL is 1.1V or even 1.0V.

Could be something like this:
CPU PLL | 1.2 | 1.3 | 1.4 | 1.5 | 1.6 | 1.7 | 1.8 | 1.9 | 2.0 | 2.1 |
QPI PLL | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 |

or maybe not so linear, IDK :shrug:

---

A read point would be handy hey.

But having said that, I read the Gigabyte review posted in the reviews section on the forums (funny that). The reviewer stated,

Quote:

---

Natively the X58 IOH provides 36 lanes for the pci-e bandwidth. Not only do video cards operate off these lanes, but so does the integrated USB3, SATA6G controller, as well as many other peripherals. A brief history of the X58 chipset will reveal that there are three revisions of the chipset, the most recently being revision B3/13, before this revision the main one that hit markets was B2/12, it seems that revision B1/11 was never released to the public. Older boards that have rev12/B2 chipsets seem to have a hard time reaching high base clock frequencies, but base clock is only partially dependent on the chipset.

---

I just quoted the whole paragraph for context and put in bold the relevant information. Perhaps the chipset is old stock if I can use that term loosely. I know other boards have better results than others which might indicate the newer revision chipset.

And here is a read point for that particular board,

http://img507.imageshack.us/img507/9...smallsmall.jpg

I hope I am not breaking anyones balls by posting this but it would be good to get down to source of the problem.

Maybe those that can post a high bclk can take a pic of their IOH core to display the markings for reference?

Maybe i'll kill my board and hope I get a better one back?

Good bench session from last night, i use my win7 24-7 and no tweaks for the moment.

http://www.unknownteam.cl/wp-content...10/12/8611.png

http://www.unknownteam.cl/wp-content...0/12/27856.png

With 3DMark i have problems, my 950 isn't so good like my "oldie" 930 and requiere to much voltage for 4.5GHz, forget the QPI VTT that i'm using for that clocks...it's almost ridiculous.

Regards!

Wow! i7 6 core heatsinks are pretty nice! A little on the small side though.
My 970 just got here. Batch 3016B246.
Does 32nm setting UCLK to 1.5x make overclocking much easier with 12GB 2GHz mem?

Can anybody be so kind to give me a starting point on this board to run my 920 DO which i had on my classified (RIP) 760 at 4ghz...This is my first Asus board so the bios are a little strange...Iam running Gskill perfect Storm 2000 6gigs..


Iam running 1005 bios which it came with
I do remember I only required 1.3v vcore on the classified

Another question.
I used this guide to bring my overclocking knowledge up to spec from core2 to i7:
http://www.techpowerup.com/forums/sh...d.php?t=108526
They said that 32nm chips show damage above 1.35v QPI/VTT. I've been using 1.4875v QPI/VTT with my 930 and 950... Do the 32nm chips need less vQPI because of the smaller manufacturing process?
What should I know as far as the voltage limit differences between 45nm and 32nm?
Thanks!

Nevermind...
Wow, it's so easy...
http://valid.canardpc.com/cache/screenshot/1552465.png
http://valid.canardpc.com/show_oc.php?id=1552465
1.331v core
1.343v QPI
1.656v DDR
1.815v CPU PLL
1.166v IOH
1.113v ICH

Hi guys, may you help me overclocking my cpu?
Actually i've managed to overclock my i7-950 @ 3800Mhz stable in windows7.

My actual settings:
bclk=200Mhz
Pcie=101Mhz
Qpi=3200Mhz
Ram=1600Mhz 7-8-7-20-1N
Vcpu=1.25V
Vqpi=1.25V

All other settings to auto.
If i try to set the cpu multiplier to 20x, i must set Vcpu=1.35V for the windows login, but the system is unable to pass any stress test..

Quote:

---

Originally Posted by DrSlump

Hi guys, may you help me overclocking my cpu?
Actually i've managed to overclock my i7-950 @ 3800Mhz stable in windows7.

My actual settings:
bclk=200Mhz
Pcie=101Mhz
Qpi=3200Mhz
Ram=1600Mhz 7-8-7-20-1N
Vcpu=1.25V
Vqpi=1.25V

All other settings to auto.
If i try to set the cpu multiplier to 20x, i must set Vcpu=1.35V for the windows login, but the system is unable to pass any stress test..

---

Try with vqpi=1,35 and IOH 1,19 volt

Quote:

---

Originally Posted by zoson

Wow! i7 6 core heatsinks are pretty nice! A little on the small side though.
My 970 just got here. Batch 3016B246.
Does 32nm setting UCLK to 1.5x make overclocking much easier with 12GB 2GHz mem?

---

UnCore to RAM ratio of 1.5x does make it easier, but in my experience needs more VTT (QPI/DRAM) voltage then >1.5x and tend to show random instability issues. So, try to use at least one more step over the lowest.
And of course, higher UnCore speed = higher memory bandwidth. ;)

Quote:

---

Originally Posted by zoson

Another question.
I used this guide to bring my overclocking knowledge up to spec from core2 to i7:
http://www.techpowerup.com/forums/sh...d.php?t=108526
They said that 32nm chips show damage above 1.35v QPI/VTT. I've been using 1.4875v QPI/VTT with my 930 and 950... Do the 32nm chips need less vQPI because of the smaller manufacturing process?
What should I know as far as the voltage limit differences between 45nm and 32nm?
Thanks!

---

I see you got a grip of things pretty fast, but just take care about voltages/temperature. It's not the same using 1.6V vCore with 45nm compared to 32nm when pushing hard on air/water (what I do ;) )
Don't get offended or something, I don't take you for ignorant (on the contrary), just a friendly advice to not get carried away. There are many dead Gulftowns reported here on XS, but just a few Bloomfields.
May your CPU live long and prosperous. :up:

Quote:

---

Originally Posted by gmod

Can anybody be so kind to give me a starting point on this board to run my 920 DO which i had on my classified (RIP) 760 at 4ghz...This is my first Asus board so the bios are a little strange...Iam running Gskill perfect Storm 2000 6gigs..

Iam running 1005 bios which it came with
I do remember I only required 1.3v vcore on the classified

---

Why don't you just start from same settings you have been using, then if it gets hard with new BIOS type, just post settings using template and someone will help for sure. :)
And then, there is a thrill of overclocking on new MB, why would you skip that? ;)

Quote:

---

Originally Posted by DrSlump

Hi guys, may you help me overclocking my cpu?
Actually i've managed to overclock my i7-950 @ 3800Mhz stable in windows7.

My actual settings:
bclk=200Mhz
Pcie=101Mhz
Qpi=3200Mhz
Ram=1600Mhz 7-8-7-20-1N
Vcpu=1.25V
Vqpi=1.25V

All other settings to auto.
If i try to set the cpu multiplier to 20x, i must set Vcpu=1.35V for the windows login, but the system is unable to pass any stress test..

---

Just try using 21x CPU multi, you might get pleasant surprise. ;)
Try higher vCore (1.35-1.4V) and slightly higher VTT (QPI/DRAM) (~1.3V) to get it stable, then later you can lower it a bit if possible.

Just got myself a W3570 Batch# 3846A803 :woot:

It's a Christmas Eve miracle :p:

EDIT: zoson looking good brother :up: Can't wait till the day comes when I own a hexcore lol
Also are you replying using a mobile phone by any chance? Might explain the double posts..

Quote:

---

Originally Posted by donmarkoni

Just try using 21x CPU multi, you might get pleasant surprise. ;)
Try higher vCore (1.35-1.4V) and slightly higher VTT (QPI/DRAM) (~1.3V) to get it stable, then later you can lower it a bit if possible.

---

But, if i use x21 multiplier i would raise the speed to 4200Mhz. Are you sure i should raise to 21?
The main problem is that with such voltages, my temps raises a lot. If i try a linx test with my cpu @ 4Ghz, have a rapid raise even to 91°C :(

P.s. i'm using a corsair h70 as heatsink

Quote:

---

Originally Posted by zoson

Nevermind...
Wow, it's so easy...
http://valid.canardpc.com/cache/screenshot/1552465.png
http://valid.canardpc.com/show_oc.php?id=1552465
1.331v core
1.343v QPI
1.656v DDR
1.815v CPU PLL
1.166v IOH
1.113v ICH

---

by curiosity, you are how many GFlops with your 970 at this frequency (with LinX)?

this new 970 (and R3E, socket LOTES) is really strange

with my old 970, to be stable and 4.25GhZ and 2025MhZ for memory (Uncore 3645), it was

Vcore: 1.337v
vQPI: 1.30v

with the new, I am stable for the same frequency with:

Vcore: 1.311v
vQPI: 1.26v
cpu-pll: 1.81v

I have not yet tested under voltage ...
I tested bios 0011, 0095 and 1102 and it's the same thing

this new 970 heats a little more (~6° higher than the other) but requires less voltage

I have a question,

I'm not the NB LED that lights on the motherboard (those for DDR, CPU and SB are OK)
and the DDR LED is yellow (crazy) while VDIMM is 1.65V :confused:

with bios 1102

why?

thank you ;)

EDIT: for LED NB, it must be hidden by the Koolance waterblock on the motherboard, I think

Quote:

---

Originally Posted by donmarkoni

UnCore to RAM ratio of 1.5x does make it easier, but in my experience needs more VTT (QPI/DRAM) voltage then >1.5x and tend to show random instability issues. So, try to use at least one more step over the lowest.
And of course, higher UnCore speed = higher memory bandwidth. ;)

---

Thanks for the info! I will up the UnCore ratio x1 and re-test!

Quote:

---

Originally Posted by donmarkoni

I see you got a grip of things pretty fast, but just take care about voltages/temperature. It's not the same using 1.6V vCore with 45nm compared to 32nm when pushing hard on air/water (what I do ;) )
Don't get offended or something, I don't take you for ignorant (on the contrary), just a friendly advice to not get carried away. There are many dead Gulftowns reported here on XS, but just a few Bloomfields.
May your CPU live long and prosperous. :up:

---

I've been overclocking for many years! However, it is a completely new game from Core2/AthlonXP days. I feel fairly safe with the volts I'm using right now. :) I used the info from this thread to set up my BIOS! Very good info.

Quote:

---

Originally Posted by spoolindsm127

Just got myself a W3570 Batch# 3846A803 :woot:

It's a Christmas Eve miracle :p:

EDIT: zoson looking good brother :up: Can't wait till the day comes when I own a hexcore lol
Also are you replying using a mobile phone by any chance? Might explain the double posts..

---

I got a friend who works retail to sign up for Intel Retail Edge so I could buy the i7 970 for $215!!!! :D :D :D There's still time if you know someone!!!

I'm posting with Chrome... Not sure what's up with the double posts... It usually doesn't do it.

Quote:

---

Originally Posted by tistou77

by curiosity, you are how many GFlops with your 970 at this frequency (with LinX)?

---

82GFlops I have a bunch of services running through, so maybe a little bit more.

Quote:

---

Originally Posted by donmarkoni

IMO, zoson's theory is most probably right, if not absolutely.



I think that QPI PLL loops 2-3 times from lowest to highest in one loop of CPU PLL voltage from 1.2V to 2.1V.
One thing is sure, QPI PLL is 1.1V when CPU PLL is 1.8V and probably when CPU PLL is lowest 1.2V QPI PLL is 1.1V or even 1.0V.

Could be something like this:
CPU PLL | 1.2 | 1.3 | 1.4 | 1.5 | 1.6 | 1.7 | 1.8 | 1.9 | 2.0 | 2.1 |
QPI PLL | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 | 1.0 | 1.1 | 1.2 | 1.3 | 1.4 |

or maybe not so linear, IDK :shrug:

---

This makes sense, I believe 1.6 CPU_PLL was used for 266 BCLK on my board, meaning 1.4 QPI_PLL (It's highest)

I think I have two lazy temp sensors... My core #2 and core #5 are 10C less at both idle and load than all others... :(

Merry Christmast all friends :up: :)

Quote:

---

Originally Posted by zoson

82GFlops I have a bunch of services running through, so maybe a little bit more.

---

ok thanks ;)

I have the same thing

you have the CPU with watercooling? what temperature with LinX?
and what bios do you use?

thank you very much ;)

Well got it going and actually turned out with a better OC than the Classified 760...Now got my 920 @ 4418 210 x 21...1.34vcore....
Score 41174 in vantage with my 2 480's...NEEED to get my waterblocks for them SOOON..

That dam EK Chipset block was a pain to install..When is EK gonna start using standoffs???
But it does look good on the board..

Quote:

---

Originally Posted by tistou77

ok thanks ;)

I have the same thing

you have the CPU with watercooling? what temperature with LinX?
and what bios do you use?

thank you very much ;)

---

Yes my CPU is watercooled.
Maxtemp with linx is 84C
Bios 1102
See my sig for full system.

Quote:

---

Originally Posted by zoson

Yes my CPU is watercooled.
Maxtemp with linx is 84C
Bios 1102
See my sig for full system.

---

thank you for the info :up:

84° is not too much for a CPU watercooled?

I'm interrested by your voltages (vcore, vqpi, vdimm, cpu-pll, etc ...) and the frequency of uncore to ram @ 2100MHz

thank you ;)

PS: with a multi of 21 to the CPU, it'll be better for the OC ;)

Quote:

---

Originally Posted by PytonOrm

Merry Christmast all friends :up: :)

---

Merry Christmas! :up:



Can't wait to get this board under water and see what the QPI will do :shrug::shocked:

Quote:

---

Originally Posted by tistou77

thank you for the info :up:

84° is not too much for a CPU watercooled?

---

Intel says 100C, so that's what I go by. I've run all my CPU's very hot and have noticed no problems as long as voltages are reasonable.

Quote:

---

Originally Posted by tistou77

I'm interrested by your voltages (vcore, vqpi, vdimm, cpu-pll, etc ...) and the frequency of uncore to ram @ 2100MHz


thank you ;)

---

vcore: 1.34
vqpi: 1.35
vdimm: 1.656
cpu-pll: auto

I'm still using 1.5x uncore:mem ratio

Quote:

---

Originally Posted by tistou77

PS: with a multi of 21 to the CPU, it'll be better for the OC ;)

---

Not enough cooling! ;)
20x seems to work fine also...

84c is ridiculously hot on water dude....
My temps On my 920 Idle is 33c full load 57c...Your might as well be running air with those temps...

Hi, I believe this depends on the watercooling setup.

I also running my 950 @4.2Ghz in the same board (have note updated my sig yet) with Vcore of 1.32 under water and I am getting ~80C with LinX.

I am staying in Singapore so the ambient is hot (~33C) and am using a Feser triple with Typhoons with add GPU, RAM and Mobo block in the loop. I dont want to add another triple at the outside of my casing or mod my case, hence I am fine with it...

Just seems like a waste of money to me if you are at 80c load with water....I have never seen that high of temps even in the summer here in Texas....

which ambient temperature, you have, to have a temperature of 80° with LinX?

look at his specs lads : 3 x 120mm rads, cooling CPU, mobo and 2 275's in SLI... what do you expect... Also never judge anotehr CPU based on your temps... some CPU's are amasing hotheads... if I passed 4.2 on my 950 I went straight to 85°C, so I kept it at 4Ghz for daily usage... that CPU never ever got a chance to get into the 4.5 charts :p

Ambient temperature is a small factor.

I have cpu, mb, and two video cards all being cooled by 3x 120mm rads with 1200 rpm fans.

The point is that my cooler is SILENT, and gets better temps overclocked than a heatsink would.
gmod you're obviously a noob to not know that's why half the people that watercool do it.

better remove the n**b part mate, ain't gonna do any good in this constructive dialogue

Yay that 84C is hot for vCPU 1.3v

Ive got that temps at vCPU 1.57v

All in one loop. 4 v 6 cores cant make such a difference? HT On?

As Leeghoofd said, some chips are just hotheads. My 930 at 4.5ghz idles in the high 30's low 40's with HT disabled at 1.38v, load is very hot, without the use of AC I would have to back the oc down. My 950 idles at 32-34c at 4500mhz-1.51v and loaded out in the high 60's maybe 70c tops. I use the same WC setup for all my chips, Im pretty sure my cooling is up to par. The VID of the chip can also play a large role in how warm or cool a chip runs, my 930 is 1.175v default, my 950 is much higher.

First pic is no AC, second with AC, same water cooling.

My 920 does 30 idle 70 full load HT off

http://img148.imageshack.us/img148/4896/244h.jpg

Intel's spec is 100C, so no, 84C isn't 'too hot.'
HT off is a cop out. If I turn HT off, I lose 10C on my temps.

the temperature of the cores into "load" will not be the same depending on the ambient temperature (that is why I asked) ;)
if I put the PC in my garden, I lose 20° on cores :D

Me, I look ambient temperature and the cooling liquid to see if my integration (cooling) is adapted to my PC (delta air/water)

Quote:

---

Originally Posted by tistou77

the temperature of the cores into "load" will not be the same depending on the ambient temperature (that is why I asked) ;)

Me, I look ambient temperature and the cooling liquid to see if my integration (cooling) is adapted to my PC (delta air/water)

---

My ambient varies between 27 and 30C. The difference in load temp is maybe 1C. Insignificant and not worth worrying about.

Quote:

---

Originally Posted by zoson

My ambient varies between 27 and 30C. The difference in load temp is maybe 1C. Insignificant and not worth worrying about.

---

I did not say otherwise, I asked just ambient temperature, to get an idea ;)

Hi, here in Singapore my ambient temps are 30-33C (my "relaxation room" has no aircon installed...which I intend to change soon since if the PC is on long it will noticably heat the room further).

My water temps is at that level and when I play games it will raise to 35C or when I stress my CPU 100% to 36C. But take note that I am not running my fans at 100% always they are controlled with my aquaero such that if temps raise to 36C the RPM will ramp to 100% by then to keep it there.

BTW for LinX I never disable HT, if I do my temps drop by ~8-10C.

The ambient temps play a GREAT role in temps you will get since this is the temp of the air you are using to cool down the fins of your radiator...today morning my LinX temps are 74 since it was raining the whole night and the room temp is cooler.

Quote:

---

Originally Posted by hafiqb

Hi, here in Singapore my ambient temps are 30-33C (my "relaxation room" has no aircon installed...which I intend to change soon since if the PC is on long it will noticably heat the room further).

My water temps is at that level and when I play games it will raise to 35C or when I stress my CPU 100% to 36C. But take note that I am not running my fans at 100% always they are controlled with my aquaero such that if temps raise to 36C the RPM will ramp to 100% by then to keep it there.

BTW for LinX I never disable HT, if I do my temps drop by ~8-10C.

The ambient temps play a GREAT role in temps you will get since this is the temp of the air you are using to cool down the fins of your radiator...today morning my LinX temps are 74 since it was raining the whole night and the room temp is cooler.

---

If you're running a 24/7 overclock that isn't stable no matter what your ambient, you're doing it wrong.
Sorry, but it's true. a 100% stable 24/7 overclock doesn't have to rely on ambient temp.

^ So if someone doesnt run an oc your way they are wrong?

And if I disable HT it is a copout? HMM

Quote:

---

Originally Posted by OC Nub

^ So if someone doesnt run an oc your way they are wrong?

And if I disable HT it is a copout? HMM

---

If you have to shut down because of ambient and adjust your overclock to deal with it, then it's not a 24/7 overclock...

Quote:

---

Originally Posted by hafiqb

Hi, here in Singapore my ambient temps are 30-33C
...

My water temps is at that level and when I play games it will raise to 35C or when I stress my CPU 100% to 36C.

---

you have a great delta air/water (33°/36°)
your watercooling is very good ;)

And of course the ambient temps is important when OC
in watercooling is the ambient air cools the water (in radiators), more "air" is hot, less water is cooled (it really make sense)

Try OC in Africa (for example) with 45°, I'm sure that you OC much less than in your home :D

@zoson, I choose not to run a chip 24/7 with idle temps in the 40's and with high load temps, meaning high 70's-80's even if just in linx. Doesnt make sense for me to when I dont have to. Normally I switch chips to one that will run cooler in the months that my temps would be higher.

If someone does things different it doesnt make them "wrong". There is no set rule to this, just different choices and preferences. For me to tell someone they are doing things wrong because they choose to do it another way would be wrong, which is basically what you did in post #3870.

Quote:

---

Originally Posted by zoson

Ambient temperature is a small factor.

I have cpu, mb, and two video cards all being cooled by 3x 120mm rads with 1200 rpm fans.

The point is that my cooler is SILENT, and gets better temps overclocked than a heatsink would.
gmod you're obviously a noob to not know that's why half the people that watercool do it.

---

Iam the noob??? Iam not the one running 3 120mm rads...

Zozon, my overclock is 100percent stable but his was not the point...

Point was that someone suggested that 80c on water in linx is bad and I happen to hover around there but I believe one cannot just compare like this.

First it is the matter of water-cooling setup, ambient temps and then of course whether HT enabled
which will make a huge difference.

I agree wih you it is not correct to compare with ht disabled.

Quote:

---

Originally Posted by gmod

Iam the noob??? Iam not the one running 3 120mm rads...

---

http://sphotos.ak.fbcdn.net/hphotos-..._2060605_n.jpg
must be noob 'eh? or maybe i just use what is needed and nothing more because I put it all internal with case mods, instead of buying a case and throwing absurd amounts of money at it. I could easily do that, but then I'd just be another person like you. No skill, just money.

And yes you're the noob who didn't read my sig.

^ It is also a matter of your individual chip, some chips run cool, and some just run hot. If someone is running LinX at 1.3v and 80c+ that doesnt mean they have poor cooling, they may just have a hot chip. Or if someone runs LinX at 1.48v and has temps in the 70's, it doesnt mean they have superior cooling, they may have a cooler running chip.

@zoson, you should find another forum to troll on, there is no need to insult and call people names here.

I run my 970 at 4.2Ghz (200x21) at 1.35v, probably could go lower, but it's 24/7 stable. My VTT is at 1.31v and my ram voltage is at 1.68v for 2000Mhz 8-8-8-24. I am water cooling just the CPU with a Feser 360 and my temps hit 64c under OCCT Linpack load. This is with an ambient of 24c and water temps at 27c. Most of the time I do not hit those high temps even doing video conversions, just when I do an OCCT stability test. Hitting 80c doesn't seem quite right, but there are a lot of factors you have to take into account, such as fan speed and whether you are forcing air from the outside in or the inside out. As long as his system meets his needs, it shouldn't matter. I would double check a few things, but without the system in front of me I am just making some guesses as to why the temps are that high. If they were lower he could probably push the chip to 4.4Ghz safely, but unless you are benching there wouldn't be any point.

Seems almost noone reads how the guys system is setup. (must be the abundance of alcohol lol)

Quick summarize he's got his CPU, motherboard and 2 x 275's in SLI in the same loop with 3 seperate 120mm rads. If I put my 970 on a single rad unit, I 'll breach 80°C easy when doing Linx or co at 4.2 and beyond... He just wanted to have it all in his case and sacrificed performance for the build... There's no magic nor an issue there... can't ever imagined we could fill 2 pages with this build... amasing !

I'm not sure why this turned into some debate over my temps. I didn't complain about them. I was asked what my temps were. My cooler is performing exactly as I expect it to. I know I have barely enough cooling for the heat load I'm working with, but the point is - it is enough to get the job done without putting my parts in danger.

As UT said, there is no real-world load that can make my CPU heat up like linpack.

Right now at idle with ambient at 29C, my coolant temp is 38C, and my exhaust air is 38C.
My cores are idling between 36C and 44C. My core 2 and core 5 have lazy sensors(yes I realize their temp is below my coolant, which shouldn't be possible).

Quote:

---

Originally Posted by zoson

I'm not sure why this turned into some debate over my temps. I didn't complain about them. I was asked what my temps were. My cooler is performing exactly as I expect it to. I know I have barely enough cooling for the heat load I'm working with, but the point is - it is enough to get the job done without putting my parts in danger.

As UT said, there is no real-world load that can make my CPU heat up like linpack.

Right now at idle with ambient at 29C, my coolant temp is 38C, and my exhaust air is 38C.
My cores are idling between 36C and 44C. My core 2 and core 5 have lazy sensors(yes I realize their temp is below my coolant, which shouldn't be possible).

---

Iam so impressed I mean you are really awesome dude...I really like the blue kink coils...that just sets the whole system apart...

I think nobody here is a noob and none should be called so

We all do share and hopefully enjoy the same hobby though.

Zozon what are your settings to get the RAM at that speed?

Quote:

---

Originally Posted by hafiqb

I think nobody here is a noob and none should be called so

We all do share and hopefully enjoy the same hobby though.

Zozon what are your settings to get the RAM at that speed?

---

1.331v core 100% LLC
1.343v QPI LLC Jumper on
1.656v DDR
1.815v CPU PLL (auto)
1.166v IOH
1.113v ICH

Cas 10-10-10-27 tRFC 110 Read-to-Pre 8
I'm using uncore x15. I tried upping to x16 but it made me unstable. :(

Zoson, since you are running a 970, have you tried running the CPU PLL at 1.42v to see that allows you to lower your voltages? Also, why 20x210 and not 21x200? Just curious? I think your chip might be able to do more without raising the voltages just by raising the multi.

Quote:

---

Originally Posted by Utnorris

Zoson, since you are running a 970, have you tried running the CPU PLL at 1.42v to see that allows you to lower your voltages? Also, why 20x210 and not 21x200? Just curious? I think your chip might be able to do more without raising the voltages just by raising the multi.

---

Testing 21x200 right now with mem at 2000mhz 9-9-9-25-110-8 1T instead of 20x210 2100 10-10-10-27-110-8 1T. I've dropped my vqpi down to 1.325 and it's still stable. Didn't seem to make any difference with my temps though.

I haven't tried extensive testing with CPU PLL yet, but I am testing tonight with it. I'm hoping it'll help since dropping to 21x200 didn't allow me to lower my vcore any, just vqpi.

Honestly, I was doing 210 because I was hoping for 4.4ghz on the 21x multiplier. I run many virtualized environments for work, and they really, really like lots of memory bandwidth. I got it stable at 20x210 but saw that my temps were too high to push further. I'm optimizing for 21x200 now.

I've ordered some yate loon high speeds from petra's to replace the enermax magma fans so that I can get better performance out of my rads. We'll see how much I benefit going from ~1600rpm at full loads to ~1900rpm. I'm hoping Q-FAN can control the fans well enough to ramp them down to ~1200rpm when not fully loaded. The enermax ones sit at ~1100-1200 rpm at idle now.

zoson, with your configuration, I would advise you to put 2 rads 360 if possible (or added more 1 rad 360)
you will win to temps, I think ;)

with my config, 2 rads 360 (TFC Xchanger and Noctua P12 at ~800rpm), I have 63° on the hottest cores (with LinX)
21x202, memory @2025MhZ 7-9-7-20 and uncore to 3645MhZ

Quote:

---

Originally Posted by tistou77

zoson, with your configuration, I would advise you to put 2 rads 360 if possible (or added more 1 rad 360)
you will win to temps, I think ;)

with my config, 2 rads 360 (TFC Xchanger and Noctua P12 at ~800rpm), I have 63° on the hottest cores (with LinX)
21x202, memory @2025MhZ 7-9-7-20 and uncore to 3645MhZ

---

I do not want to put anything outside my case. It's big enough as it is!
I live in a very small studio apartment.
I think I need more vcore and vqpi because i have 3x4gb 2000mhz. I also think it's why my latencies on mem is higher.

So far 1.41775, 1.4375 cpu pll has not allowed a vcore drop. testing 1.44425 now. 1.4375 was better than 1.41775.

Quote:

---

Originally Posted by zoson

Testing 21x200 right now with mem at 2000mhz 9-9-9-25-110-8 1T instead of 20x210 2100 10-10-10-27-110-8 1T. I've dropped my vqpi down to 1.325 and it's still stable. Didn't seem to make any difference with my temps though.

I haven't tried extensive testing with CPU PLL yet, but I am testing tonight with it. I'm hoping it'll help since dropping to 21x200 didn't allow me to lower my vcore any, just vqpi.

Honestly, I was doing 210 because I was hoping for 4.4ghz on the 21x multiplier. I run many virtualized environments for work, and they really, really like lots of memory bandwidth. I got it stable at 20x210 but saw that my temps were too high to push further. I'm optimizing for 21x200 now.

I've ordered some yate loon high speeds from petra's to replace the enermax magma fans so that I can get better performance out of my rads. We'll see how much I benefit going from ~1600rpm at full loads to ~1900rpm. I'm hoping Q-FAN can control the fans well enough to ramp them down to ~1200rpm when not fully loaded. The enermax ones sit at ~1100-1200 rpm at idle now.

---

Yeah, I am still playing with mine. I seem to be close to 4.4Ghz, just not sure how much voltage I want to run through it. I do video conversions, but I use my GTX460's to do it, so it's more for just seeing what I can get out of it than anything else.

I like the Yates, but I would have done the Gentle Typhoons instead, although a bit more expensive. There is a great fan review by Martin in the water cooling section if you have not seen it yet. A lot of useful information in there.

Quote:

---

Originally Posted by Utnorris

Yeah, I am still playing with mine. I seem to be close to 4.4Ghz, just not sure how much voltage I want to run through it. I do video conversions, but I use my GTX460's to do it, so it's more for just seeing what I can get out of it than anything else.

I like the Yates, but I would have done the Gentle Typhoons instead, although a bit more expensive. There is a great fan review by Martin in the water cooling section if you have not seen it yet. A lot of useful information in there.

---

Just looked it over... Funny... Martin gave the fans I have now a 9 on his noise quality. I've got enermax magma's now. I like 'em, but at 1600rpm they don't get any favors from my gt stealth or gtx's.

I'm going to get me some gtx560's once they come out. :)

I just bought a rampage 3 extreme today. hooked it all up and went to set things up. system turned off on a windows install. next time I booted I got "IHO temperature warning" anyway Ive taken the heat sink off the board and then I noticed this..... Is my heatsink MISSING A HEATPIPE!

I never took a close look at my extreme heat sink when I took it off, but it looks like you are. Anyone else know the answer?

This is Correct
http://vr-zone.com/gallery_cache/485...fdd5df8f81.jpg

looking at that heatsink picture, it seems the thermal pad was removed by you ? , the missing heatpipe is another thing :p

Quote:

---

Originally Posted by Flo4ts

I just bought a rampage 3 extreme today. hooked it all up and went to set things up. system turned off on a windows install. next time I booted I got "IHO temperature warning" anyway Ive taken the heat sink off the board and then I noticed this..... Is my heatsink MISSING A HEATPIPE!

---

Like Leeghoofd notice there is no thermal interface or you clean it before take pic? The mising pipe is not a problem for IOH because it is coolest chip from the VRM-ICH-IOH circuit, the heatpipe is designed to take heat from VRM trought ICH to IOH for dispatch. Apply thermal interface and check for good contact and you can use it with higher temps on VRM and ICH. But better RMA it mate ;)

I've tested every setting pretty well from 1.4 - 1.5 CPU PLL. 1.45750 seems to be the sweet spot for me.
It allowed me to drop my core voltage *one* step from 1.331 to 1.325 in BIOS.
The combination of lowering vcore, cpu pll, and vqpi has dropped my load temps from 84 to 81C with a 28C ambient temp.

21x200
12GB (3x4Gb) 2000MHz Memory at 9-9-9-25-110-8-1T
UCLK 3000MHz(x15)

1.325v Core 100% LLC
1.325v QPI LLC Jumper on
1.656v DDR
1.45750v CPU PLL (auto)
1.166v IOH
1.113v ICH

I also tried upping UCLK to the 16 and 17 multipliers, but it always made me way unstable. :(

Also, how do you disable EIST on r3e? I can't seem to find the BIOS switch for it. I have C1E disabled, cpu TM disabled, and speedstep disabled, but according to CPU Tweaker, EIST is still enabled.

Found this on Asus R3E forums:
http://aod999.ca/R3E/BIOS
Bios 1203!

Quote:

---

Originally Posted by zoson

Found this on Asus R3E forums:
http://aod999.ca/R3E/BIOS
Bios 1203!

---

check page 151 post 3773 :up:

Quote:

---

Originally Posted by dejanbinladen

check page 151 post 3773 :up:

---

DOH!
Did you find any noticable difference between 0095, 0096, and 1203?

0096 so far best for me, didnt tryed 0095 on ln2 yet