My Q6600 saga has begun - feel free to chime in with suggestions

[MichaelD]

What's one more Q6600 thread, right?

I received my G0 yesterday and installed it tonight. After a bit of initial fidgeting w/the BIOS (been awhile since I've been in there) I'm sitting at 3.0GHz (333 x 9) with the voltage on AUTO (more on that in a sec). CPU-Z says my FSB/DRAM ratio is 2:3. Not the optimal 1:1 but good for now.

Unlike with my E6400 dual core, the OCing setup w/this Quad was different. With the dual, I set FSB to 400, memory multiplier (or whatever it's called to 2.0) and went. That didn't work with this Quad. I had to put the memory mult to AUTO and the FSB to 333 and it booted and ran. I'm sure I've got something screwy...but its running!!!

I would like to stress that this install and the Arctic Silver5 is less than 3 hours old...far from broken in.

With 4 instances of P95 running for about an hour now (no crashes or errors) and all my cores are at 61C (measuring w/CoreTemp) My room temp is 22C. Not bad at all, IMO!!! The cores idle at about 46C. I'll let it Prime for the next 5 hours or so, just to heat it up. Then I'll shut it off overnight to let the AS thermal-cycle.

3.0GHz...so far, so good. I haven't tried to go higher yet...wanna make sure it's stable at this speed.

The voltage is wacky. According to CPU-Z (v1.41) at idle, my voltage is 1.44v. With 4 instances of P95 running I'm at 1.36v. Weird, right?

Thoughts? Temps are not bad, considering the fresh install, right?

Here's my system:

Gigabyte P9650-DS3 Rev1.0 (F12 BIOS)
2GB Corsair PC6400
Freezer7 Pro
P180 with fans on Med
Corsair 620HX PS
8800GT

 

[imported_cambit69]

pump that thing to 400 x 9 like right now. don't wait

 

[Markfw]

380x9...I couldn't get 3.6 stable...Also, the vcore ? thats called vdroop. The only motherboard that I know of that doesn;t have the problem is the DQ6. And crank up the vcore a couple of notches to get more OC if you need to. 1.4 vcore under load is not a problem with decent cooling. Just don;t go over 65c under load using coretemp.

 

[4x4expy]

You shouldn't need that high of voltage to be stable at 333x9. I have my Q6600 on a P35-DS4 r2 with the same freezer 7 cooler. My VID is 1.25, and was stable P95 stable 4hrs at stock vCore at 333x9. I tinkered a a little and am at 356x9(3.2ghx) now at 1.26v(via CPU-z=1.246 idle and 1.232 under load) and never has a hiccup on P95 overnight. I had it up to 385x9 but took 1.39v, was ok on P95 x30 min but temps were getting to high for comfort, so I bailed, and went back down to 356. I'll push it harder when I get a better cooler.

When I first installed mine, I was getting spontaneous reboots and lockups at modest OC's. After I disabled C1E and EIST and turned the voltages to manual control, I had much better stability. I would suggest maximizing your BIOS first.

Very happy with my new system below:

Gigabyte P35-DS4 rev2
Q6600 @ 356x9
Gskill DDR2800 @ 890 4-4-4-15 w/1.95v
BFG 8800gt OC at 680/1000
Freezer 7pro cooler
Seagate 7200.11 500gb
PCP&C Silencer 610w
Lian Li AO5B

 

[MichaelD]

Originally posted by: 4x4expy
You shouldn't need that high of voltage to be stable at 333x9. I have my Q6600 on a P35-DS4 r2 with the same freezer 7 cooler. My VID is 1.25, and was stable P95 stable 4hrs at stock vCore at 333x9. I tinkered a a little and am at 356x9(3.2ghx) now at 1.26v(via CPU-z=1.246 idle and 1.232 under load) and never has a hiccup on P95 overnight. I had it up to 385x9 but took 1.39v, was ok on P95 x30 min but temps were getting to high for comfort, so I bailed, and went back down to 356. I'll push it harder when I get a better cooler.

When I first installed mine, I was getting spontaneous reboots and lockups at modest OC's. After I disabled C1E and EIST and turned the voltages to manual control, I had much better stability. I would suggest maximizing your BIOS first.

Very happy with my new system below:

Gigabyte P35-DS4 rev2
Q6600 @ 356x9
Gskill DDR2800 @ 890 4-4-4-15 w/1.95v
BFG 8800gt OC at 680/1000
Freezer 7pro cooler
Seagate 7200.11 500gb
PCP&C Silencer 610w
Lian Li AO5B

Thanks, 4x4expy. Interestingly enough, I have C1E enabled. It gave me zero problems with my last CPU (which I had for 20 months, overclocked!!!).

But...this is a different CPU...whole different ballgame. I'm still tinkering at this point. The next step is to try 400x8. I'd be very happy w/that.

BTW folks; still no errors in P95 (4 count) at this time. I'd almost call it stable.

*edit*

Holy crap, I don't understand this at ALL. I just stopped all 4 instances of P95. 4 hours, 1 minute with no errors or warnings. Great. And as soon as I stopped it, my CPU voltage jumped right back up to 1.44v!!! W_T_F??? Guess I need to disable all that underclocking nonsense in the BIOS...though it IS NICE to have...

 

[Markfw]

CPU voltage jumped right back up to 1.44v!!! W_T_F???

Like I said, that vdroop. Only really noticable on quads on almost all motherboards.

(except you get none on a DQ6)

Edit: and you need your vcore to be at 1.36 or better under full load, 100% on all 4 cores.

 

[MichaelD]

Originally posted by: Markfw900
380x9...I couldn't get 3.6 stable...Also, the vcore ? thats called vdroop. The only motherboard that I know of that doesn;t have the problem is the DQ6. And crank up the vcore a couple of notches to get more OC if you need to. 1.4 vcore under load is not a problem with decent cooling. Just don;t go over 65c under load using coretemp.

Thanks pal. You know; I've heard of the dreaded vdroop before, but never personally experienced it. I thought it was only a problem experienced by those "hardware challenged" people. With my PS I thought I was immune. Sad to say...I didn't perform tonight.

WTF? Hahahah! This is funny...

 

[MichaelD]

Well, the plot thickens.

After the previously mentioned P95 action, I'm now idling at 40C across all 4 cores with the same room temp of 22C. That's 6C lower than before, and I've not even shut the machine off yet!!!

I will assume that's due to shutting off C1E and all other automagic CPU/VID/multi controlling settings in the BIOS<--that's the only thing I changed.

But all's not well in Whoville. I have tried about 1,000 different settings and i cannot get it to 400x8 or above. My C2D ran at 400x8 for 20 damn months on this same board!!!

I've raised the CPU voltage to 1.475 and the MCH and FSB voltages to +.20 at the same time. It just will not boot Windows (but boots the BIOS and will show up to 3.2GHz) at anything above 333x9.

I don't think temps are the problem; I'm idling WAY lower than before. It's not my PSU (Corsair 620HX); is it the board? The CPU? My settings? GRRRR!

I'm not looking for 4GHz on air; 3.2GHz would be dandy.

 

[Markfw]

Michael, I PM'ed you my phone number. With all the gigbyte boards I have on C2Q's (4 of them) and none under 3.2, I am sure I can get you there. Call me 8 am-10 pm PST

 

[SickBeast]

I find that memory timings are often the cause of overclocking instability. What I would do would be to relax them as much as possible, and bump the memory voltage moderately. Once you have reached an overclock you are happy with, slowly tighten up the timings.

Also, I would avoid dropping the multiplier below stock until you've found out what your max overclock is.

Good luck. :beer:

 

[MichaelD]

Thanks very much, Mark; I will try to give you a call. I'm on a completely different schedule than the average normal guy...my day is your night, etc. But I'll give it a shot. There's always IM and email though.

SICKBEAST: Thank you too. I don't think it's my memory timings, and here's why. Nothing has changed since the last CPU (E6400). The memory has ran at SPD, at 400MHz since day one. Stock timings, stock voltage, etc. I've not tried OCing this memory; I've read that it doesn't have much headroom above 200MHz anyway.

It's weird that if I leave the CPU voltage on AUTO (but disable EIST/CIE) it runs great at 9*333 with CPUZ showing 1.328v. But if I try to run it at say, 8*400 with the voltage hard set at 1.328 or higher, Windows won't boot.

I already know my memory can run at 400MHz; it's DDR2-400! So WHY can't I get the CPU to run at 400FSB, even with 1.450v?

One thing I've NOT done is cleared the CMOS. That MIGHT be an issue...too much "garbage" left over from past OCing attempts...hmm.

 

[myocardia]

Why are you trying 8x400? Why not try 6x535? I mean, you're only trying to overclock twice as many cores, and four times as much cache. Yeah, 6x535 would be alot easier on your old motherboard than using the 9x multi.[/sacasm]

 

[MichaelD]

/taps sarcasm meter
/meter not moving

Maybe I'm not awake yet. You ARE saying that the probs I'm having are normal and I'm expecting too much, right?

Well, I look at it this way; I've got a known good OCing MB, good memory, great PS, good HSF and a G0 step that was packed less than a month ago. I'm not shooting for the moon, here.

I SHOULD be hitting 3.2 easily, based on everyone else's experiences and the fact that right now i'm Priming all 4 cores and sitting at 58C across the board.

 

[jmmtn4aj]

Originally posted by: Markfw900
380x9...I couldn't get 3.6 stable...Also, the vcore ? thats called vdroop. The only motherboard that I know of that doesn;t have the problem is the DQ6. And crank up the vcore a couple of notches to get more OC if you need to. 1.4 vcore under load is not a problem with decent cooling. Just don;t go over 65c under load using coretemp.

It's not a problem. It's an Intel design spec to prevent voltage overshoot when the load drops.

 

[myocardia]

For those with a broken sarcasm meter: 3.2 Ghz with a Q6600 = 9x356, not 8x400, 7x454, or 6x535.

edit: Oops, I meant 7x457.

 

[MichaelD]

Originally posted by: myocardia
For those with a broken sarcasm meter: 3.2 Ghz with a Q6600 = 9x356, not 8x400, 7x454, or 6x535.

edit: Oops, I meant 7x457.

I don't understand. 8*400=3200. Are you saying that 8*400 isn't possible w/aQ6600? Why not? I'm super confused right now.

Also,I prefer to run at nice, neat memory dividers like 333 or 400.

 

[Markfw]

Originally posted by: MichaelD

Originally posted by: myocardia
For those with a broken sarcasm meter: 3.2 Ghz with a Q6600 = 9x356, not 8x400, 7x454, or 6x535.

edit: Oops, I meant 7x457.

I don't understand. 8*400=3200. Are you saying that 8*400 isn't possible w/aQ6600? Why not? I'm super confused right now.

Also,I prefer to run at nice, neat memory dividers like 333 or 400.

One of my Q6600's runs 8x400, a B3 stepping. Michael, when I get time later, I will tell you all the settings, but did you check the memory multiplier ? It should be at 2.0

 

[MichaelD]

Originally posted by: Markfw900

Originally posted by: MichaelD

Originally posted by: myocardia
For those with a broken sarcasm meter: 3.2 Ghz with a Q6600 = 9x356, not 8x400, 7x454, or 6x535.

edit: Oops, I meant 7x457.

I don't understand. 8*400=3200. Are you saying that 8*400 isn't possible w/aQ6600? Why not? I'm super confused right now.

Also,I prefer to run at nice, neat memory dividers like 333 or 400.

One of my Q6600's runs 8x400, a B3 stepping. Michael, when I get time later, I will tell you all the settings, but did you check the memory multiplier ? It should be at 2.0

Hey buddy. :beer: I haven't been in the BIOS since last night, but IIRC I disabled it b/c Windows wouldn't post. I hard set the memory at 333 and now I'm running at 9*333 with a 1:1 FSBRAM ratio.

I will try for a higher clock tonight...I may just have to clear the CMOS and start from scratch.

 

[imported_LoKe]

Why are you guys dropping your multiplier? Just causes more heat...Run your memory 1:1 sync, timings on auto, or 5-5-5-15 if you can stabilize that way. 2.0-2.1v on the dimm (2.2 should be acceptable). Lock pcie to 100 or 102.

 

[MichaelD]

Originally posted by: LoKe
Why are you guys dropping your multiplier? Just causes more heat...Run your memory 1:1 sync, timings on auto, or 5-5-5-15 if you can stabilize that way. 2.0-2.1v on the dimm (2.2 should be acceptable). Lock pcie to 100 or 102.

Hello LoKe. I'm dropping to 8 b/c I don't know if my CPU can do 3.6GHz (9x400). I want the benefits of a high FSB, so I'm shooting for 400FSB.

I don't see a need to change the memory voltage, as I'm not OCing the memory. It's been running at 400MHz at the stock 1.9v since day one...20 months ago. Same board, same memory, NEW CPU.

I'm going to screw around w/the BIOS a bit more now and see what happens.

 

[Duvie]

Try 401fsb instead of 400fsb...known issue on some but usually isn't a make or break for booting...

also try...

1)make sure pci express is locked at 100
2) make sure pci clock is locked at 33
3) use a memory divider now to run ram at below stock speed.....remember "ISOLATE AND CONSOLIDATE" when ocing
4) use very conservative timings....IE stock is 4-4-4-12? use 5-5-5-15
5) see if raising NB vcore, vdimm or fsb termination helps...sometimes it is not juct vcore that needs a boost....

 

[Duvie]

My DS3 ran my QX6700 at 8x401 no problem.....It also ran as high at 8x425 but I experienced some throttling due to air cooling alone and 80c temps under load....no crashes though....It ran 3.3ghz at its highest, but ultimately it really liked to run 10x333.

Bottomline is the board can OC a quad core....

I would personally run the stock fsb multplier at 9x and keep the fsb lower....usually means less stress on the NB chipset

 

[GuitarDaddy]

Originally posted by: MichaelD
It's weird that if I leave the CPU voltage on AUTO (but disable EIST/CIE) it runs great at 9*333 with CPUZ showing 1.328v. But if I try to run it at say, 8*400 with the voltage hard set at 1.328 or higher, Windows won't boot.

You need more vcore, auto setting showing 1.328v in cpu-z is higher volts than hard set at 1.328 in bios. On P965 boards there is vdrop(idle vcore less than bios vcore) and vdroop (load vcore less than idle vcore) both work to decrease the vcore set in bios. On my board I have to set 1.45v in bios to get 1.36-1.37 under load in windows. Bump that sucker up to 1.45 in bios and see what it yeilds under load in windows. And I bet it will boot 401x8

 

[VirtualLarry]

Wasn't there some issue, where when you dropped your multiplier, the clock of the northbridge was still set to the higher multi, so effectively if you run your clock at 8x400, the northbridge is running at 9x400, or something like that? I remember reading something about that in relation to the 965 chipset. I don't know if the P35 behaves that way.