X5667 ES Q3QR @ LN2 # 3rd Vantage on HWBot

Printable View

03-22-2010, 02:55 PM
G.Foyle

X5667 ES Q3QR @ LN2 # 3rd Vantage on HWBot

Hi guys,
We just tried a Xeon X5667 ES on LN2. CPU SSPEC is Q3QR, batch is C936.
Unfortunately our Rampage II Extreme lost turbo multipliers, and we ran out of bclk. We didn't have much time for fine tuning performance, but it was enough to take world's 3rd vantage score:

Xeon X5667 ES
Asus Rampage II Extreme
GoodRAM PRO 2000 CL7 3x 2GB
Radeon HD5970 (thanks MSI!) + 2x Radeon HD5870 (Connect3D)
Antec TruePower Quattro 1200W (non-OC version)
Kingpin's F1 EE

http://profide.eu/files/results/gulf...rd_vantage.png
http://profide.eu/files/results/gulf...tage_47891.png
http://profide.eu/files/results/gulf_q3qr/rig.jpg

It'll be short lived as anyone with a semi-decent Gulftown on LN2 can beat that, but we're working on something more :)

Thanks & until next time!
03-22-2010, 02:59 PM
FatAlbert

nice :D

guys, use P6T7 and 4x5870 :D
03-22-2010, 03:01 PM
Xtreme Addict

We tried SuperComputer, but there is a problem with TURBO (it slows down during tests like on MSI motherboards) and BCLK ends at 229 MHz. We need MORE R2E's to find the good one :)
03-22-2010, 03:08 PM
FatAlbert

hell ya :D

you need RIIIE :D
03-22-2010, 03:12 PM
Xtreme Addict

Maybe new MSI BigBang Xpower will be better :)
03-22-2010, 03:12 PM
FatAlbert

will see :D
03-22-2010, 03:23 PM
saaya

asus still hasnt figured out tdp/tdc hard mods? LOL :lol:
check the intel whitepapers, there are pins where nehalem and westmere cpus meassure tdp and tdc from the pwm, just fake the reading with a resistor and those turbo multi issues should be solved :)
03-22-2010, 03:29 PM
Don_Dan

Good work Mateusz and Miravo! :up:
03-22-2010, 03:45 PM
saaya

jepp, good work, especially considering the multi issues! :toast:
03-22-2010, 04:07 PM
Xtreme Addict

@saaya

TURBO worked perfectly until a moment during BSOD and TURBO is lost forever on this motherboard, I tested it with other CPUs like i7 920. No multi and basta :lol:

I have screens from the beginning of the session.

http://img689.imageshack.us/img689/4927/gulfazopt.png

http://img411.imageshack.us/img411/2262/gulfazot3d.png

http://img694.imageshack.us/img694/6141/spi2323.png

Something is DEAD now on motherboard. We forced flashing bios and so on, nothing helps. I have to RMA it.
03-22-2010, 04:17 PM
Linuxfan

Sorry you guys had trouble with the RIIE!
Great results, gulffy is nice. ;)
03-22-2010, 04:21 PM
zalbard

That's a really strong chip!
Congrats, keep tweaking it! :up:
03-22-2010, 04:34 PM
Xtreme Addict

Thanks guys! :)

This chip is really good for a locked cpu. But unfortunately it has locked uncore multis :(

On single stage (idle -40*C, during vantage cpu test 2 temperature raise up to -2 :lol:) it does

http://img535.imageshack.us/img535/3869/spi.png

http://img534.imageshack.us/img534/8419/vantagefc.png

http://img263.imageshack.us/img263/2...rdpolski06.png
03-22-2010, 04:34 PM
Fitseries3

sweet.

cant wait to see what my x5677's can do :D
03-22-2010, 05:12 PM
tj_jackson

Xeon X5667 with 6 cores? I thought it was only 4 cores as shown in most of the links that have Internet

http://www.cpu-world.com/CPUs/Xeon/I...4005154AB.html

of the 6 are the ones who saw the:
X5650
X5660
X5670
X5680

4 of the new are
X5667
X5677
03-22-2010, 05:16 PM
Fitseries3

any x56XX chip that ends in a 0 is a 6core.

the rest are quads or dualies.
03-22-2010, 05:18 PM
Xtreme Addict

yes retails are quads, but this one is ES and has six cores
03-22-2010, 06:17 PM
Hondacity

excellent work xoqolatl and Miravo!

is 50k possible on vantage? :D
03-23-2010, 04:30 AM
zalbard

Looks like a good chip, congrats! :up:
03-23-2010, 05:34 AM
SoF

very nice multi-card action!
1.93V on just -40°C :eek: You are a brave man :D :D
03-23-2010, 05:59 AM
Sam_oslo

Quote:

Originally Posted by Miravo

Thanks guys! :)

This chip is really good for a locked cpu. But unfortunately it has locked uncore multis :(

On single stage (idle -40*C, during vantage cpu test 2 temperature raise up to -2 :lol:) it does

]http://img535.imageshack.us/img535/3869/spi.png[/IMG]

]http://img534.imageshack.us/img534/8419/vantagefc.png[/IMG]

]http://img263.imageshack.us/img263/2614/rekordpolski06.png[/IMG]

The locked x20 Uncore is a pain on my X5650 too, it needs too much VVT for higher BCLKs.

This chip is doing good by taking 1.93 Vcore, maybe these can take a good deal of torture anyways. How much VVT have you been pushing in this run?
03-23-2010, 06:18 AM
Xtreme Addict

if I remember correctly 1.65v - 1.7v vtt
03-23-2010, 06:25 AM
Sam_oslo

Quote:

Originally Posted by Miravo

if I remember correctly 1.65v - 1.7v vtt

That's good news :), it is a good indication on how much torture these Xeon DPs can take, at least for short runs.
I hope they can take extra VVT for 24/7 too, I need it badly.
03-23-2010, 06:59 AM
Xtreme Addict

1.65v-1.7v on AC 24/7 no problem with memory 2000 MHz CL7.

@SoF

According to xoqolatl I like high voltages :D
03-23-2010, 07:22 AM
tool_462

Quote:

Originally Posted by Sam_oslo

That's good news :), it is a good indication on how much torture these Xeon DPs can take, at least for short runs.
I hope they can take extra VVT for 24/7 too, I need it badly.

What are you giving yours for 24/7? I'm keeping it under 1.34v as a self imposed limit right now, more later when I get some balls :p:
03-23-2010, 07:47 AM
Sam_oslo

Quote:

Originally Posted by tool_462

What are you giving yours for 24/7? I'm keeping it under 1.34v as a self imposed limit right now, more later when I get some balls :p:

I've chickened out at VVT=1.32v and this give max 24/7 stable @3.8Ghz(172x22) 1.21v. Next option in my BIOS is VVT=1.34v and I'm really tempted to use it, but trying to learn more bout these chips, and see specially how much VVT they can take for 24/7 use, before I raise the VVT.

Quote:

Originally Posted by Miravo

1.65v-1.7v on AC 24/7 no problem with memory 2000 MHz CL7.

@SoF

According to xoqolatl I like high voltages :D

I belie VVT~=1.5v will get me up to @4.4GHz and I will be happy with that for 24/7 use, and my RAM can do 2000MHz CL9, but not CL7.

Can you please explain how 2000MHz CL7 can make those high VVT safe for 24/7?
03-23-2010, 07:50 AM
SoF

CL7 2000 can sometimes need past 1.5 vtt while CL8 same speed 1.35V is all you need - at least on my RIIE it is that way ;) CL8 and CL9 doesn't make a big difference ;)
03-23-2010, 08:02 AM
Sam_oslo

Quote:

Originally Posted by SoF

CL7 2000 can sometimes need past 1.5 vtt while CL8 same speed 1.35V is all you need - at least on my RIIE it is that way ;) CL8 and CL9 doesn't make a big difference ;)

Yeah, but does any of these make High VVT safer for 24/7?

The x20 Uncore of my X5650 needs a lot of VVT, almost independent of the Core-multi or RAM-timings. My concern is, what's the max-safe VVT for 24/7 without killing or degrading these 32nm CPUs?.
03-23-2010, 08:05 AM
Xtreme Addict

Quote:

Originally Posted by Sam_oslo

I've chickened out at VVT=1.32v and this give max 24/7 stable @3.8Ghz(172x22) 1.21v. Next option in my BIOS is VVT=1.34v and I'm really tempted to use it, but trying to learn more bout these chips, and see specially how much VVT they can take for 24/7 use, before I raise the VVT.

I belie VVT~=1.5v will get me up to @4.4GHz and I will be happy with that for 24/7 use, and my RAM can do 2000MHz CL9, but not CL7.

Can you please explain how 2000MHz CL7 can make those high VVT safe for 24/7?

I never bother about safe voltages with mem and vtt/qpi, I like good results and sometimes I have to help components raising voltage a bit ;)

I stopped thinking about all those theories about relation vtt - vmem like 0.5v etc. After launchtime of first i7 I bought i7 965XE retail and used it with R2E and 3x1 GB Goodram PC10600 D9 GTR, old good microns. On aircooling I put 1.8v qpi (max stable on 965XE on air cooling) and 2.5v mem (those microns loved voltage). I was working 24/7 on this setup during a month, on those voltages, Guess what was dead after a month? R2E :p: but I RMA'
ed it. CPU and memory are working till now, they didn't feel a degradation factor and so on. I killed about 30 E8xxx cpus but I never killed i7. And I just don't believe in too high vtt and too high vmem.
03-23-2010, 08:19 AM
Sam_oslo

Quote:

Originally Posted by Miravo

I never bother about safe voltages with mem and vtt/qpi, I like good results and sometimes I have to help components raising voltage a bit ;)

I stopped thinking about all those theories about relation vtt - vmem like 0.5v etc. After launchtime of first i7 I bought i7 965XE retail and used it with R2E and 3x1 GB Goodram PC10600 D9 GTR, old good microns. On aircooling I put 1.8v qpi (max stable on 965XE on air cooling) and 2.5v mem (those microns loved voltage). I was working 24/7 on this setup during a month, on those voltages, Guess what was dead after a month? R2E :p: but I RMA'
ed it. CPU and memory are working till now, they didn't feel a degradation factor and so on. I killed about 30 E8xxx cpus but I never killed i7. And I just don't believe in too high vtt and too high vmem.

I know what you mean. i7 45nm have been good at taking torture, including high VVT for 24/7. They have been almost indestructible. I'm not sure if the same applies to these new 32nm CPUs, but I hope they are as resilient too.

These kind of tests can give some indications too, even tho these are not the same as for 24/7. It is promising that this chip could take 1.65v-1.7v VVT in this test :) But I'm still a kind of chicken, and need to hear more tales like this.