How many people are actually running Q66/700s at or above 3.6ghz?

[Idontcare]

Originally posted by: Syzygies

Originally posted by: rbk123
load it peaked at 58C but is around 56C; running Orthos large for 2.5hrs

My hunch reading this was that Orthos isn't stressing all four cores. A quick google suggests (I don't know for sure) that it stresses two cores, one needs two instances to stress all four cores:

Overclock.net

This is true, Orthos will only load up 2 cores. You can run multiple instances of orthos though to ensure all 4 cores are loaded.

Or you can just get Prime95 (same underlying program as Orthos) and it will automatically recognize all the cores and load them appropriately.

 

[rbk123]

Originally posted by: bryanW1995
what bios version are you running?

I'm running the original 11 bios which came with it. The other bios's actually slow the board down some so I haven't switched as the double boot issue doesn't bother me in the least.


For the questions on running Orthos Large, I only ran 1 instance as I thought that stressed the 4 cores (all 4 did go up in temp). However I will run 2 instances and report back with my temp update. Again, keep in mind my voltage under load was 1.38V which is relatively low, plus my ambients are pretty cool in my basement. Nonetheless, I'll report back when I do further testing.

 

[Ciber]

I've been running at 3.6 ghz for the past 3-4 days without an issue. I've managed to go higher, but i need better cooling.

 

[Rubycon]

Originally posted by: lopri
Probably the board tries to default to tRD=5 or 6 @400FSB and 7 or 8 @420FSB.

At lower multi's 400 is fine. Also higher multi's with lower FSB (ex 333x12) still have issues.

 

[rbk123]

So here's my latest update -

- I ran 2 instance of Orthos and it peaked at 57-58C; did this for almost an hour
- stopped that then proceeded to run 4 instances of Orthos (more is better, right?) and it also wouldn't break 58C. Ran this for almost 2 hours before quitting (while surfing, MP3 rearranging, etc..)
- I was getting the same reading Core Temp as I was in Real Temp factoring in the 5C delta between the 2.

Not sure what else I need to do to prove but this chip works really well for me. Again my ambients are right around 60F so no doubt that helps.

 

[bryanW1995]

that's great temps. are you using a lapped true?

 

[rbk123]

The Zalman is clearly lapped by Zalman (not by me) and I did not lap the heatsink on the chip itself.

 

[Tweakin]

Originally posted by: rbk123
Here's my G0 update:

- running stable at 3.6G on my IP35-E
- had to set the bios voltage to 1.465 but this board has big v-droop issues
- CPU-Z says the voltage is 1.432-1.440V at idle
- at load it says the voltage is 1.38V
- this chip runs uber-cool
- on air (Zalman 9500) I am getting idle temps at 28C
- load it peaked at 58C but is around 56C; running Orthos large for 2.5hrs

This chip rocks as if I didn't have such v-droop, I could set the bios voltage lower since it only needs 1.38V for load stability. Don't get me wrong, though, I do love the IP35-E.

I'd post screenshots but I don't know how to do that. I have plenty.

Those temps seem a little low for a Zaleman on top of a Quad at that vcore. What program are you using to check these?
EDIT - I'm late on this post...

 

[Syzygies]

Originally posted by: EvilSponge
Man I'm really getting interested in water cooling. I think what keeps me away is the cost, matience and the fear of water near all my expensive components.

I would love to get into water cooling. Review after review makes it clear that state-of-the-art air cooling can overtake average water cooling kits. Spec'ing a respectable custom setup, I just priced the parts I would need for a Thermochill PA 120.3 based T-line cpu-only loop, looking for guidance to AnandTech motherboard review test setups, and I got to $500 very easily.

My present rig is almost yours. I need to lap my TRUE and Q6600, and I could replace my push/pull fans with stronger fans. Let's say this only gets me from 3.2 Ghz to 3.4 Ghz at my temperature comfort levels (24/7 computations on all four cores < 60 C). Water cooling would remove temperature as an obstacle, then I'd get to 3.6 Ghz before exceeding my voltage comfort levels. There's no point. Within reason, processors don't mind running hot; they do it all the time in laptops which last years.

So EvilSponge, you're my hero, I can screw up lapping my Q6600 twice on the budget I'd save by sticking to air. This whole thread has been a very useful reality check.

 

[jeeka715]

To debunk some earlier posts about people only posting their bootable settings in those enthusiast forums, I am a member of XtremeSystems (jeeka715) and I run my q6600 24/7 at 3800Mhz, 8x475fsb, 1.5v.

This is not my "bootable" only setting, this is my HD/DVD Video crunching, Game Playing, E-Mail Checking, ORTHOS stable (2 instances running Affinities on cores 0-1 and 2-3 to properly run it ragged) machine. I got a lucky CPU I guess.

Most air cooled q6600's I have read about like to sit at 3600Mhz on average, but mine, for some reason likes 3800Mhz, and at 475x8 on top of that. There is a general consensus that Quad Cores don't like high fsb. I would consider mine a "high" fsb Quad Core. Most Dual cores run higher fsb, though. Again this is in general.

I had reservations about getting a q6600, expecting to settle at 3.2Ghz over my e6600 at 3.6Ghz. Not only did I hit 3.8Ghz on the q6600, it runs cooler and uses less volts, and gets higher fsb than my e6600! So I guess I won out on the clock, cooling, volts and fsb.

Granted I have a Thermalright Ultra 120 Extreme, but my temps never go over 60C on 100% (all cores) load.

I guess its luck of the draw.

I would not run at that speed if I didn't trust it to do my Multimedia work, the work that pays the bills.

Hope this adds to the discussion!

 

[aigomorla]

Originally posted by: Syzygies

Originally posted by: EvilSponge
Man I'm really getting interested in water cooling. I think what keeps me away is the cost, matience and the fear of water near all my expensive components.

I would love to get into water cooling. Review after review makes it clear that state-of-the-art air cooling can overtake average water cooling kits. Spec'ing a respectable custom setup, I just priced the parts I would need for a Thermochill PA 120.3 based T-line cpu-only loop, looking for guidance to AnandTech motherboard review test setups, and I got to $500 very easily.

MCR320 + DDC-3.2 + XSPC Top + D-tek Fusion + 15 feet of 7/16ID" Masterkleer + 3 yates + swiftech radmount if you lack moding skills. does not = 500. More like somewhere near 290-320.

And yes this would stomp air.

And please tell me who it was that, told you that you require a PA120.3 for a cpu loop only. I would love to go 1 on 1 with him.

An MCR220 with decient fans could give you great temps better then air on a cpu loop only. Depending on overclocking and heat load, the 220 would hold fine. The 320 would be border line xtreme, the PA is only if your aiming at near silence while pulling good temps.

[/quote]

 

[Gillbot]

Uh oh, Aigo's mad! 😉

 

[bryanW1995]

aigo, what's the noise like on the 320? I'd like to get into water cooling but I've (nearly) resigned myself to waiting until nehalem and/or westmere for it...

also, would an antec 900 be adequate for a water-cooling rig?

 

[Syzygies]

Originally posted by: aigomorla
MCR320 + DDC-3.2 + XSPC Top + D-tek Fusion + 15 feet of 7/16ID" Masterkleer + 3 yates + swiftech radmount if you lack moding skills. does not = 500. More like somewhere near 290-320.

Parsing that line sent me back to hours of research, and I thought I'd broken the spell water had on me!

Martin's Liquid Lab is a crucial link that helps to explain your astute recs, far more going on in the above line than personal preference, e.g. the XSPC top rules, to a large extent well-made triple rads are in fact interchangeable, and there are at least four competing water blocks now for an overclocked Q6600. One needs to take into account the issues posed by each component choice, but many combinations will work nicely.

A quick back-of-the envelope power calculation gives

95 watts * (3.6/2.4) * (1.5/1.275)^2 = 197 watts

...which seems to agree with anecdote that one could be moving 200 watts to cool a Q6600 overclocked to the title speed of this thread. And 200 watts is what one remembers from all the double 120 radiator graphs, with a 10 C water rise. Doable, but expecting the system as a whole to be less efficient, a double radiator won't be in its peak game here, it will run a little hot. Your triple radiator rec is only a few bucks more; it seems the way to go. Yes, this prices out more like $300 than $500, with no sense of compromise. It's inevitable I'll do this with at ieast one machine.

Originally posted by: EvilSponge
My rig has ran prime95 4 threads 16 hours stable @ 3.6 ghz [...] My Q6600 is a GO stepping with a 1.2875VID so to run prime stable I need to have 1.408V, 1.50v set in bios [...] My idle temps are between 34-35c and my loaded temps range between 61c - 63c Now I'm using a lapped TRUE with two 120mm fans in a push/pull configuration. That's with living in very hot south Florida.

My PC: http://www.bluekarmann.com/web%20pages/espc.html

I took the side off my P182 and stared as it ran, for a long time. A slow but sure way to fix things. (After a "rain man" stare in the early 90's at a flagship box that Power Computing couldn't fix, I realized that the case was mildly warped, sending a barely visible ripple across the motherboard to unseat the memory ever so slightly on the other side. A two cent washer fixed it for good.)

Doing the same F*V^2 power calculation to explain my core temps at various overclocks wasn't making sense at all, unless I assumed that the air going into my push/pull TRUE cooler was much warmer than "ambient". Sure enough, the cores run 5 C cooler with the side panel off.

There's a 2" gap between the pull TRUE fan and the rear P182 case fan, and they line up nicely, with only a tricky-to-reroute power cable crossing the space. One could reverse the direction of all these fans, mod the top of the case to mount more fans for exhaust, perhaps duct the rear fan straight into the TRUE, for the same 5 C drop and better cooling overall? I now appreciate that depending on setup, the Antec 900 offers the chance to run cooler.

An easier fix: Rig another fan in 3 bays, ducted to blow a fairly straight shot into the push/pull TRUE:

Scythe Kama Bay HDD Cooler
FrozenCPU 120mm Fan Duct

(Looks like I got the last one.) I have to try this with lapping, to give air a fair chance against water.

Originally posted by: rbk123
I ran 2 instance of Orthos and it peaked at 57-58C; did this for almost an hour

My friends were avoiding the server version of Ubuntu, till I pointed out how easy it was to get the GUI back at installation, and that our computations run a few percent faster than on the Ubuntu desktop kernel. Either way, they run significantly faster on Linux than on Mac OS. Surfing for water parts last night, I saw a lament where testers couldn't achieve the power they expected from Windows, and the other shoe dropped: There's no reason to assume that Windows is better than OS X at supporting computations like these prime tests, and a reasonable guess would be it's worse, not based on a Unix kernel. My students use "Tiny XP" to run their games faster, a bittorrent stripped version of XP that runs faster.

So I may be running hotter, testing using Linux.

I'll end with my last musing: if water is so much better than air, is the humid air of south FL better at cooling a TRUE?

 

[v8envy]

Originally posted by: Syzygies

So I may be running hotter, testing using Linux.

You may be, unless you're setting CPU affinities. I noticed I get a bit more work out of my quad with a server OS (I'm sure the same goes for Windows variants) because the OS doesn't insist on round robin migrating processes across all four cores every few milliseconds, destroying any sort of cache coherency.

Also, just like you can install the desktop on a server Ubuntu you can just install the linux-blah-server (or even -realtime!) kernel packages on the desktop distro.

If you need the absolute maximum performance from your apps I recommend the gentoo approach of recompiling your kernel and libc packages from source with cpu-optimized flags (e.g., -O2 -march=core2 or -march=native, with or without -msse3). There's a goldmine of info at http://www.thinkwiki.org/wiki/Intel_Core_2_Duo and friends. It's a lot easier to get cpu-optimized binaries w/ Ubuntu than it sounds. You'd be amazed what a giant difference doing this makes. The lowest common denominator (read: 386) binaries aren't all that on a pentium-m

Thanks for the pointer to tiny-xp, I'll check it out.

 

[EvilSponge]

Hey Syzygies, Well it's not humid inside my house. It is very humid outside though 🙂. My ambient temprature ranges between 75f and 80f with little humidity due to the air conditioner. Hope this helps BTW water cooling is the way to go for better than true cooling.

-sponge

 

[Rubycon]

Dewpoint, enthalpy, and RH are related (see psychrometrics) however they effect the comfort index for animals (people) which depend on evaporative cooling of perspiration to control bodily temperature in hotter climates. Reducing humidity will do nothing to improve computer cooling UNLESS you're using a cooling tower (see bong cooler).

 

[Syzygies]

Originally posted by: Rubycon
Reducing humidity will do nothing to improve computer cooling UNLESS you're using a cooling tower (see bong cooler).

I'm arguing that increasing humidity will make air a more effective heat transfer medium, because the water vapor increases the thermal mass of the air. I was wondering whether the effect would be noticeable.

A commercial airline pilot takes into account temperature, humidity, and altitude at the airports at both ends, in making various calculations for planning a flight. The density of the air is crucial to liftoff, landing parameters.

Imagine that one could turn a dial, and twisting to the left, air acted like bone dry, mile high desert air, and twisting to the right, air had exactly the thermal mass and thermal transfer rate of water. Overclockers on air would twist to the right.

We're debating what happens as we twist the dial back to the left. Does this advantage suddenly die out, or does it fade away smoothly? My intuition is that such effects are generally smooth, but they can become negligible.

The commercial pilot analogy suggests there is still some advantage to humid air, for air cooling. Whether it's more that a degree C, I couldn't say. But ever try to cook pasta at 10,000 feet? You can't, without a pressure cooker. These effects are real, as pilots know.

 

[bryanW1995]

good thing that the italians came up with the idea of pasta then! If they'd first thought of it in nepal then I wouldn't get to eat mom's spaghetti and meatballs tonight...yum!

 

[EvilSponge]

Actually the italians did not come up with pasta they got pasta when Marco Polo the venecian trade who followed the silk road brought it back to Italy from China.
Now with that being said pasta is my fav dish!

-sponge

 

[Rubycon]

Originally posted by: Syzygies
I'm arguing that increasing humidity will make air a more effective heat transfer medium, because the water vapor increases the thermal mass of the air. I was wondering whether the effect would be noticeable.

A commercial airline pilot takes into account temperature, humidity, and altitude at the airports at both ends, in making various calculations for planning a flight. The density of the air is crucial to liftoff, landing parameters.

Imagine that one could turn a dial, and twisting to the left, air acted like bone dry, mile high desert air, and twisting to the right, air had exactly the thermal mass and thermal transfer rate of water. Overclockers on air would twist to the right.

We're debating what happens as we twist the dial back to the left. Does this advantage suddenly die out, or does it fade away smoothly? My intuition is that such effects are generally smooth, but they can become negligible.

The commercial pilot analogy suggests there is still some advantage to humid air, for air cooling. Whether it's more that a degree C, I couldn't say. But ever try to cook pasta at 10,000 feet? You can't, without a pressure cooker. These effects are real, as pilots know.

These effects are real and I'm not arguing that however they have no bearing on measurable difference in air-air or air-water heat exchange!

The reason why one has a hard time cooking at higher altitude is due to the lower boiling point of water. Hotter water cooks faster but if it boils away you lose the cooking (transfer) medium! A pressure cooker creates its own atmosphere to allow higher temperatures and faster cooking.

Increased humidity is bad for computer parts. Computer rooms are kept at 45% RH - low enough to avoid condensation and high enough to limit the buildup of potentially destructive electrostatic discharge.

 

[spinejam]

i've had 3 q6600's (all G0) on air-cooling.

one o/c to 3.4 on an ip35-e
one o/c to 3.5 on an ip35
and the one in my sig! 🙂

 

[kb2114]

I think I got lucky on my 1.25 VID Q6600. I run 24/7 stable at 3.6 ghz currently with the vcore at 1.4, and I can get 3.825 ghz 3dmark stable at 1.45v, and I can get into windows and idle with no crashes at 3.915 ghz with 1.5v. Temps at 1.45v and above start to hit the high 60's at load, so I don't leave it there for long.