QX6700 detailed OC testing

4

4DoorGTZ

Limp Gawd
Joined
Nov 20, 2005
Messages
276
Ouline: Have a week off work, going to push my stable OC, starting very perliminary, then slowly tighten tweek for max output.

Hardware:
Qx6700es
Asus P5W DH Deluxe
Coolit Freezone Modified: 114cfm 120mm fan, duct tape fan shroud, removed aluminum side plates, "control" set to high
Noctura NH-U6 fanless chipset cooler on the NB with AS5
Stock southbridge heatsink (shiny asus tin cover removed, AS5 to be applied soon)
4GB Team Xtreme DDR2 800 [email protected] naked w/OCZ mem cooler
Ati AIW x1900
OCZ 600w Evostream PSU
74Gb Raptor sata
250GB WD sata

Case:
Coolermaster 830 silver
3x 45cfm 120mm front intake fans (plus 1 62cfm 140mm fan (at the back of the empty drive bays much closer the the mb, verticaly in the middle of the 9 5.25 bays)
1x 45cfm 120mm top exhaust
2x 62cfm 140mm side intake (swivel fan door) blowing at the bottom of the mb/AIW)
1x 62cfm 140mm side exhaust (swivel fan door) located right above the OCZ mem cooler blowing in the same direction
OCZ P/S with 2x80mm fans, 1 sucking from the void above the mb, one exhaust out the case)
*actually low db fans except for on the freezone, thats an airplane.

Software:
Asus 1707 Bios
Intel's TAT
Orthos beta (for C2d, running 2 instances, affinity's set core 0/1}aff1 2/3}aff2
Xp Pro SP2

Room setup:
This 14'x10' room has very poor ventilation, 28GHz including this folding, and a high ambient air temperature, therefore until I settle on a stable OC I'm trying to run this with the window open (Canadian Winter) hallway door shut (keep rest of the house warm) I have a old occilating fan under the window (nw corner), a wall clock with temp 3ft east of the window on the north wall. NE corner of the room houses a D805 @3.2G and my T7600 dell M2010 "laptop". Sw corner has my old P4 2.4 laptop, a Via 1Ghz, and a P4 3.4. I'm in the Se corner near the hallway door (closed) and have a temp readout on my Logitech keyboard. (simply put, cold air supply and me are in opposite corners of the room, with folding machines in the other 2 corners) All machines are on and folding (cept this while I tweak) I typed all that out so you can understand why I'm doing this with the window open, dont pay too much attention to the idel/load temps because unless you compute outside they're on the cool side, I'm mostly interested in what the chip will do at certain voltages/FSB/Multi's right now and will be watching the temp changes (delta's)

I dont know if there will be much interest in this, for a while at least this week I'll try to update this thread, feel free to bookmark it :)
Here are some pics I put in another thread, apologies for the shaky hand.
 
I'll list ambient temp as the average between my wall clock temp (near window) and keyboard readout (near pc) its just easier that way. mobo temp is with the case fan-door running, disconnecting these 3 fans had made the mobo temp readout jump 3-7c during regular folding operations. Idle is simply measured 3-5mins after windows and all startup items load (very minimal items running), load temp is only after 5min of Orthos (remember this is very preliminary, I just want to know what's kinda stable and what has no chance, will further stress test later) Note I've seen quite the voltage drop from bios settings (what I quote as my V) and displayed voltage, as bad as 1.525(bios) was 1.49(displayed @idle) and dropped to 1.455(displayed @load). I speak bios settings but actual voltage is always a bit lower, lower still under load.

_Bone stock, Memory at 4-4-4-10 @2.1
Amb:18
mb:24
idle:19
load:27
-So here at stock values my cooler is only 1c above ambient, and only a 8c temp delta.

_Exactly the same as above, multi @12x (3200)
amb:18.5
mb:23
idle:19
load:33
-Cooler is near the same @idle, load temps jump 6c, total delta 14c now.

_Tried a jump in fsb from 266-280, posted, started to load windows, rebooted before the login screen.

_upped Vcore to 1.350 (from stock 1.30) Windows loading fine, Orthos crashing in about 10s
amb:16
mb:23
idle:25
load:crash
-6c above ambient, cooler still seems fine here

_upped Vcore to 1.400 now, everything seemed fine, Taskmanager crashes after 4min load
amb:16
mb:22
idle:28
load:44
-9c above ambient at idle now, but boing temp delta of 16 at load (and suspiciously sorta unstable, orthos lasted the usual time tho so its almost stable for now)

_upped fsb to 290 (3480) and first time had to lower mem multi cuz I'm not testing that right now, Orthos crashes after 4s once then 1:25
amb:15.75
mb:22
idle:30
load:46 in 1:25 then crash
-looks like it needs more juice

_upped Vcore to 1.450, Orthos crashes again but lasts longer 33s/2:35
amb:15
mb:23 (ambient is falling, mobo is starting to warm up)_
idle:34
load:49 in 2:35 then crash
-more juice still I think

_upped Vcore to 1.500 Orthos survives
amb:15.75
mb:25
idle:44 (wow jump)
load:60
-starting to get to the top of my safe zone when considering future normal ambient temps in this room, but stable, time to bump fsb. You can see what people mean when they say this cooler cant handle a Kentsfield when OC'd with more voltage its running 46c above ambient under load you'd never have that with a good water setup and this is only 3.48Ghz @1.50v

_upped fsb to 300 (3600) Load all my utils and orthos, windows explorer crashes, try orthos anyway goes 1:35 crashes, then goes 30s and hard reboot
amb:16.5
mb:26
idle:46
load:57 in 1:35 then crash, reboot on 2nd try
-So with these higher fsb's and Vcore no longer does orthos just crash, if I persist the computer hard reboots, maybe more V?

_ upped Vcore to 1.525 now, Orthos reboots the pc, second try it goes 10s crashes, 3rd try 2:30 and hard reboot again, so much for stable.
amb:16.5
mb:26
idle:50
load:67 in 2:30 before 3rd atempt reboot
-Thats as high as I'm comfortable with Vcore, could gain stability with chipset V increase, perhaps later. for now less fsb less Vcore

_fsb 295 (3540) Vcore 1.500 Runs stable
amb:17
mb:27
idle:49
load:63
-good place to stop, stable but warm :)

Incidently I tried my 13x multiplier before all this and even with 1.525v it never even posted, I've heard others get their quads to run the 13x so maybe its my "ES" chip that is a bit supersticious.
---------
So day 1 I learn that my cooler is great for a stock Kentsfield (who buys a $350can cooler if they plan NOT to OC?????) But I guess it does moderatly well if you OC and dont raise the V too much. I can do 3200 14c above ambient 3360 28c above ambient, 3480 44c above ambient, or settling back to 3540 46c above ambient. This chip just needs the V for the speed, but the cooler lags behind. Perhaps tomorrow I'll go back to the "stable" results push them a few mhz and start playing with chipset voltages. For now back to my previously reliable folding speed of 3340 @.1.45 and 41c above ambient (which as I've typed all this out, shut the window and kept my door closed, damn its 23c near the window and 28c at the keyboard. 25.5c ambient if we're comparing to my above work) ahh there we go just opened the door no chance of freezing the house now. Edit: door open for a while, its actualy 38c above ambient now that the heat is spreading to the rest of the house like normal :)
 
son of a * its almost 5am, uhh time for bed. maybe if I'm gonna do a write up i should start my testing/tuning before 11pm :eek:
 
Well this afternoon I figured before I'll be able to get more refined OC's suitable for longer stress testing I should get to know my RAM :)

Running Memtest86+ v1.70, Vdimm is bios setting, shows .05 higher with the Asus AI booster in windows, no way to verify from within memtest, think of that as you want.

Rated for DDR800 4-4-4-10 @ 2.0-2.2V I'm running these all at 2.2v, only bumped to 2.25 couple times when 2.2 errored after time (not instantaneously) Didnt help anyways.

I'll save you the step by step like my first post, but so far 4-4-4-10 @2.2v is good (until the end of test #4) up to DDR997 (498mhz). Seems I over purchased ram for this OC, I dont see the need for DDR997 here, probably great ram to save for a future low multiplier/high fsb OC. Also runs 4-4-4-9 around DDR983 (491)

For the hell I tried a few other settings since CPUz seemed to think my ram was rated for DDR800 @ 3-4-3-9, Well it was wrong, 4-4-4-9 works at DDR800, gonna test the limit of that next, also tried 3-4-4-9 had instantaneous errors at DDR800, 4-3-4-9 seems a candidate for higher speed testing as well.

I'll add with an edit in a bit. I'm only testing until the end of Test #4 for time reasons, when I've got candidates for suitable mem speeds and ratings I'll test them longer to make sure they're good. (dont feel like waiting through multible 2-3hr tests, this way its 10-13min each)
 
Well for the end of the day I just did more vigerous testing on my "safe" OC levels, 3340([email protected] ddr2-927) ambient 24 was 48 Idle, as high as 69 load after 1/2 hour.(using OCCT now, it loads all 4 cores automaticaly, check it out)

I spent a good 2hrs with many different configs to try and get 3600 (12x300) stable. Went as high as 1.5625 in bios and realized this mobo cant do it.

1.5625(bios setting)=1.480(bios reporting, idle)=1.448(cpuz reporting, for the few sec i could load all 4 cores)
1.550(bios)=1.472/1.464(bios reporting, fluctuating)=1.456/1.448(cpuz briefly loaded)
1.5375(bios)=1.456(bios reporting, idle)=1.440(cpuz briefly loaded)

So huge voltage drops keep me from 3600, I might see if that the same result when using a 11x or 10x multi, but probably is.

Tonight before I go sleep I'll play some more with OCCT and my other preliminary "stable" OC's but no window open testing, the wind is blowing and it got as cold as 8c in my room this time.
 
makes me realise how good a cpu I managed to get but then it was cherry picked.
lucky you with only 24C ambients. more like 34C over here in Oz. Even at 1.42V(after droop) im loading cores at 72,71,69,70

hey do the vdroop pencil mod, anything between 2B and 8B pencil is fine, no HBs though. ;)
sorry for bad pic, taken from another thread(XS) but just add a few strokes of lead to the resistor thats got a white arrow pointing to it, will improve your o/c

p5wdhdroop.JPG
 
Had everything out of the case today, Reseated the CPU block with AS5, removed the tin cover from the southbridge heatsink and reseated it with AS5 as well. Went to the local office supply store and bought some graphite pencils and did the Vdroop mod. WOW did it ever help

First off Vcore setting and Vcore actual reading has narrowed by about .02 still .04 lower at 1.500. However No more Vdroop (well 0.008) got 3480 stable (12x290 4-4-4-9 1.500) just tried 3600 (12x300 4-4-4-9 1.500) OCCT failed after 6min, upped to 1.5125 (1.480actual) and here it drops .008 at load. OCCT crashed after 14mins..... hmm maybe more juice again :)

Spent some time trying for 3540 (12x295 DDR2-983/887) Seems ok in windows, OCCT crashed after about 15mins, temps get up to 80/85 1.500/1.525 Voltage drop only a consistant 0.008 idle/load after the pencil mod. That was done after leaving the window open longer then i planned, ambients from 10-14c, so that 80/85 is very high considering the low ambient, prob wouldnt boot windows with my normal 23-26c ambient. Went back to 3480 tested stable again, gonna leave it here to fold overnight and up the fsb 1 at a time tomorrow (only have a range of 5 to try) Unfortunatly I believe I may have hurt temps with my second seating of the waterblock on the cpu, I now get about 5c different between cpu1 and 2, before they were within 1 almost all the time, It might do for now, I think this chip will stay in this case with this mobo, but get a higher end normal watercooling setup in the future, while this Freezone will go to another folding rig after the next intel price drops :)

I would highly recommend anyone who is OC'ing a quad core (probably similar on the Q6600) with this P5W-DH board to do the pencil mod, "may" have helped my stability but the piece of mind that you're not fluctuating the Vcore on idle/load is nice to have. Probably has similar effects with a higher fsb OC on a dual core.
 
I can't get my Q6600 stable above 3.33 Ghz (9x367), but this may not even be the CPU, but the "quad with high FSB problem". I had my board running with a E6400 @8x450 FSB, but with the quad I can't even POST with 400, even when the multi down to 6... anyway, I'm impressed with your detailed quest for the holy overclock :D
 
Jadawin said:
I can't get my Q6600 stable above 3.33 Ghz (9x367), but this may not even be the CPU, but the "quad with high FSB problem". I had my board running with a E6400 @8x450 FSB, but with the quad I can't even POST with 400, even when the multi down to 6... anyway, I'm impressed with your detailed quest for the holy overclock :D
Click to expand...

I literally have nothing else to do but watch TV; house and grandparent sitting this week. He's mostly blind, doesnt understand hardly and english and my grandma who normally looks after him (and was suppose to 'sit this house) is in the hospital with a broken Femur (same weather cooling my room caused her to slip outside)
 
Sorry to hear about the problems of your grandparents. But then, at least you still have them, I don't. Even my parents already died... oh well.

Yeah, so you need something to kill time basically. You should write a nice article when you're done, with some graphs and all numbers nicely sorted :)
 
Well 3480 @1.500 gets mighty toasty at my normal ambients (26 right now, damn this room sucks) chip is up to 83 folding 2 smp's in VMware since last night, gonna have to shut er down and try for the same at lower V hope its stable and cooler.
 
Jadawin said:
Sorry to hear about the problems of your grandparents. But then, at least you still have them, I don't. Even my parents already died... oh well.

Yeah, so you need something to kill time basically. You should write a nice article when you're done, with some graphs and all numbers nicely sorted :)
Click to expand...

This is getting depressing:(
 
Nah, that's just life. You have to [H]ard enough to live through the bad times to have any chance of enjoying the good things.
 
3480 was my hovering point today. Tried it at various voltages around 1.500, the obvious happened but nothing rock solid stable in OCCT. Best I got was 17min of OCCT however I left the computer on folding (2 SMP's in VMware) most of the day and its good, of course VMware cant utilize as much as OCCT so this setting might be "stable enough"

I think that may be all for the major work on this setup, as the F@H contest is starting and I need the 4000ppd this box churns out, I might add mhz of fsb here or there.

SMP 1760 pointer @3300 11:30/fr
SMP 1760 pointer @3480 10:30/fr
So a 5.4% clock speed increase gave 9.5% SMP folding increase :) Makes me wanna get 3600 working, hmm maybe it does work for folding..... I had it 15min OCCT stable @1.525 and 65c above ambient, hmm makes for 87c right now as my abmient is 22, ouch. 3600 was also 14min OCCT stable @ 1.5125 and only 58c above ambient, time for a midnight tweak here or there and see how it fairs :)
 
Since the folding contest is under way I havent done a lot with this OC. However there were some windows updates, requireing a restart so I played a bit before starting up VMware again....

3480 @1.45v(bios) 1.416(actual) 1.408(load) [12x290] [DDR2-967] Passed the 30min OCCT stability test (I had only tried it at a higher V before and it failed) So I started up VMware, folding away on 2 SMP units and went to bed. Woke up to a windows login screen, guess it wasnt so stable after all :)

So I went back down to 3420 (all same voltage) [12x285] [ddr2-950] passed the OCCT test again, again left it run 2 SMP units all day and overnight. Was all good in the morning, so this morning I upped it to 3456 (samve voltages) [12x288] [ddr2-952] and it crunched SMP units most of the day no problem....

Leads me to believe this is pretty much the top, 3.456GHz.... I'm gonna try and submit it to the official conroe overclocking database (which I'd be #1 stable qx6700) as long as I can get it to pass orthos for 4hrs (i think it should)


And this evening I was cleanin up my workbench from when I modded my Freezone, thought why let this decent 92mm high cfm (relativly) fan go to waste? So i stuck it in my case, resting ontop of my video card pointing right at the rad of the Freezone.... Did all this with the computer on (tsk tsk bad boy) but it has dropped cpu temps by 8-10c (Currently running 2 VM's crunching SMP work units) hmm should have done that a long time ago.
 
Airflow is very important, I put my machine in a new case yesterday (Xion Stacker, nothing too special), but it improved my temps a lot. The old case was so small, I could hardly remove and reconnect any cables inside because there was no space left to just move the cables... airflow nada. Not I've got 2x 12cm fans at the front, one in the back and one on the side near the bottom, blowing air in the direction of the PCIe slot. 10C less temp on the 8800GTX. 10C less on the system sensor. 8C less peak temp of my Q6600.

I think around 3.4 Ghz is the max for the quads, anything else seems to be quite rare. There's the odd 3.7 OC somehwhere, but most seem to be in the 3-3.4 area.
 
4GB Team Xtreme DDR2 800 [email protected] naked w/OCZ mem cooler
....
Xp Pro SP2
Click to expand...

If you are going to stick with XP you should look at the x64 version. Right now at least 1gb of that ram cannot be used, maybe more. It makes no sense to use 4gb with the 32bit XP.

It won't mess up your OC and it will really work well with those 4 cores of yours. ~3.3ghz is pretty damn good though... just hate to see such good ram go to waste.
 
^ Yeah I'm aware of the XP 3GB limit, for the most part I bought 4gb because it was on sale after christmas. 1/2 this is going into my next Conroe box (probably a Q6600 after the price drop)
 
Well this is the eve of my QX6700's arrival from "The Egg" ..I am ditching my E6600 for a quadcore.. I've been on several forums and have read plenty on what to expect.. I just wanted to add to this forum, on what worked for me, as far as misc temps go..

I run a total of (11) 120mm fans on my setup right now. It's not near as loud as you might think.. I have (2) sunbeam rheostat 4 fan controllers and (1) silverstone 3 fan controller. (3) of my fans are mounted on the outside of my case, which are mounted on the bottom side of my radiator. Since I am running (2) 8800gts's in SLi, I have (2) fans mounted DIRECTLY under each 8800gts for a total of (4) fans dedicated to my 8800's. Next, I simply have (1) intake fan and (1) exhaust fan. That leaves me with the final (2) fans pointed DIRECTLY at my PWM's/NB/SB/RAM.. I have yet to see my case temp exceed 35c, even on the hottest of days here in Texas. I'm glad I found this thread as I can really relate to what's been said thus far.. I will pass on my temps when I get the QX6700 all mounted in. BTW, since the internal structure of the QX6700 is different than the regular C2Duo's, I can assume that the single thin line of AS5 down the center of the IHS is out of the question? How did everyone apply their thermal compound?
 
:eek:

Wow, That is one sweet ride. I am not a fan of LEDS, but still... oh and hey, do those bottom fans under the case match the others?
 
Good eye.. No they do not.. (Not yet atleast). I have (5) masscool 120's and (6) aerocool 120's ..In the beginning, I really liked the masscool fans because the are dead quiet. But the only draw back is, they only put out 56cfm. Well, that was fine and dandy because that was plenty of air to keep my cards cool. Well, when I went with my water setup, my particular setup benefits from higher CFM fans. So I looked around for a bit, trying to find HIGH CFM / LOWWWWW NOISE 120's. The only draw back is the price. So, eventually, I will get rid of my masscool fans and buy (5) more aerocool fans to keep with the theme.. LOL. Ouch, that's gonna hurt the pocketbook!

http://www.newegg.com/Product/Product.asp?Item=N82E16835129245
 
Cool deal... I am finally back up and running.. Less than 12 hours on my QX6700. My Abit IN9 32X-MAX motherboard finally came in! I went ahead and purchased an additional 6 120mm aerocool fans.. Now all of my setup is trimmed out in aerocools.. I plan on taking a few update pics and maybe a few screen shots once I get my quad up to speed..
 
4DoorGTZ

I want to ask you what size allen wrench did you use to remove the original fan from the freezone.

Thanks
Infospace
 
I've been tinkering with my setup a little.. I am pretty happy so far.. My very first run of 3dmark06 is promising. Over 1 million submitted scores on 3dmark06 and I am basically hovering around the top 1% ..not bad.


3dmark06-3.jpg



3dmark06-3-1.jpg



3dmark06-3-2.jpg
 
You must log in or register to reply here.
Back
Top