First time overclocker, Q6600 Gigabyte EP45-UD3R

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
Hey!

So first time I attempt this. I have a hard time finding my ways around the BIOS on this motherboard since it's different from others.
I managed to change the four essential things:

Mutliplier: 8 ---> 9
Cpu Frequency (FSB) : 266 ---> 333
Vcore: 1.28750v ----> 1.3150v
PCI express: Auto ----> 100

When I booted after this, my computer powered up and shut down about 3 times before It went on and I could access the BIOS again. When I opened the BIOS I saw a red screen that mentioned the issue and said it could be due to overclocking or other system enhancements (duh..).
My changes were still the same in the BIOS though, although CPUZ in windows is showing the previous, default, settings still.

What went wrong? How can I fix this?
 

lostnkonfused

Limp Gawd
Joined
Feb 23, 2008
Messages
133
The overclock was unstable and wouldn't post for you. It automatically reset itself to default settings but didn't change the previous settings you set from the overclock so you can make adjustments and save it again. It boots at the default settings but still retains your previous data saved in its memory without utilizing them--which was what you saw in CPU-Z. The solution is to go in the bios and either tweak your previous settings to make it stable, or set the bios settings to default manually.
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
lostnkonfused, Thanks man, I went back and set everything back to default manually. But I did it one step at the time. And it seems as if it was the FSB than needed to be reset (from 333 to 266). After that everything was back to normal again.
The motherboard (Gigabyte EP45-UD3R),the CPU (Q6600 with Zalman 9500 HS) and the Memory (G.Skill 4GB 1066) are all overclock friendly, highly overclock friendly. I must be doing something(s) wrong in the BIOS settings cause I think my build is pretty much setup for overclocking.

Any tips and tricks would be highly appreciated.

Thanks!
 

lostnkonfused

Limp Gawd
Joined
Feb 23, 2008
Messages
133
For most first time overclockers, I notice that most people do not set the correct RAM divider and RAM voltage. The GSkill 1066, I assume requires 2.0-2.1V to run above DDR2-800--I would take the voltage settings off auto and set it manually to this. Also, make sure you have the correct timings for your RAM set.

I've never personally used the Gigabyte EP45-UD3R before but there should be some options to adjust the RAM divider in correlation with your FSB (usually they just call it DDR frequency). The key to adjusting the divider is to initially set it to stock speed (around 1066 for you) and the goal is to keep it at or below this frequency setting as you raise your FSB (the DDR frequency is calculated using ratios which means that raising the FSB raises the DDR frequency). As you raise your FSB higher and higher, the gap between selectable frequencies will get larger and larger, but just remember to stay at or below your stock speed.

Remember that overclocking is a long and tedious process. What I recommend is to start small and limit the variables that you have to deal with. I would first start out with a low multiplier for your processor and set it to 6. Take every voltage setting off auto and set it to their minimum stock settings. Keep increasing the FSB (5MHz at a time)until you get to the point where you need to start adding voltage to make it stable (record the previous setting that you used that didn't need an increase in voltage, stress test after every 15 FSB and after every voltage increase as well). I would run orthos on small fft for 1 hour intervals for quick assessments after 15 FSB increments. When you start adding voltage, be very conservative about it and only do one component at a time. Keep doing this process until you find the maximum FSB you can run stable at (record this and all your voltage settings). After you've found the maximum FSB, you can then start concentrating on the processor by raising the multiplier--the key is to find a good mix with a high FSB and processor speed (I would raise the multiplier to 8). If you find that you can't get the processor stable with the high FSB and multiplier, start lowering the FSB 2MHz at a time until you find a stable setting. When you think you've reach the max limit of your OC, run 24 hours on small ffts with a final test of 16 hours large ffts after that.

Stay within reasonable temps, vcore (I wouldn't exceed 1.5), and other voltages.
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
Excellent walkthrough lostnkonfused!

I'm sorry it took so long to reply but I've been OC'ing back and forth for the past 3 days, with stress tests overnights (Prime95).
I think I am stable at 3.0 GHz (25% oc) with 334 FSB and 9x multiplier. The Memory frequency is 1069 Mhz(1066 max), which is tricky, cause I do not know what's ideal for that?
I tried at first to put the DRAM as close to 1066 as possible, and the system ran really stable, even stayed stable for 12 hours of Blend Torture test in Prime95. But when I did the small FFT test (which is more CPU demanding) it crashed in the middle of the night.

For more info here are some screens of what CPU-Z and Core Temp is showing right now,whilst Prime95 is running torture test:

cpuzvr7.jpg


memoryuq1.jpg


coretemphq6.jpg


Is there anywhere I can fetch the BIOS settings in Windows Xp ? So that I can show what I set everything to. Think that'd be more clearer.

Thanks!
 

lostnkonfused

Limp Gawd
Joined
Feb 23, 2008
Messages
133
Excellent walkthrough lostnkonfused!

I'm sorry it took so long to reply but I've been OC'ing back and forth for the past 3 days, with stress tests overnights (Prime95).
I think I am stable at 3.0 GHz (25% oc) with 334 FSB and 9x multiplier. The Memory frequency is 1069 Mhz(1066 max), which is tricky, cause I do not know what's ideal for that?
I tried at first to put the DRAM as close to 1066 as possible, and the system ran really stable, even stayed stable for 12 hours of Blend Torture test in Prime95. But when I did the small FFT test (which is more CPU demanding) it crashed in the middle of the night.

For more info here are some screens of what CPU-Z and Core Temp is showing right now,whilst Prime95 is running torture test:

cpuzvr7.jpg


memoryuq1.jpg


coretemphq6.jpg


Is there anywhere I can fetch the BIOS settings in Windows Xp ? So that I can show what I set everything to. Think that'd be more clearer.

Thanks!

That looks very good on such low vcore. Perhaps a little more vcore will make it pass small ffts. Small increases in vcore (even a tiny bit) can help it become stable.The goal is to achieve a high speed while being conservative on voltages--the uppermost limit is 1.5 vcore. Also remember to enable Load Line Calibration to prevent vdroop on voltages.

These are things you can probably worry about later:
I wouldn't touch GTL until later on when I find that additional vcore won't help stability, but 0.65 seems like an adequate reference value to start out with. Also, in my experience, the Q6600 doesn't seem to overclock well with a high PLL voltage (for me, 1.6V seems to be a good spot but it may be different for you).
 
Joined
Dec 25, 2008
Messages
61
Looks like you have a very low vid Q6600 - should be able to achieve some good results with that mobo/ram combo.

My G0 Q6600 passes small FTTs @ 3.015ghz at 1.15v, 1.12 after vdroop (even with load line calibration). For the sake of testing though I think it might be worth bumping it to 1.2 or higher, find a speed you are happy with stability/temperature wise and then start dropping it down until you find the lowest stable voltage.

Some ideas:

1. You're running decent ram. Check on the manufacturer's website or on the ram itself for a voltage setting and make sure your vdimm corresponds to this. 1066 ram often runs at voltages of 2.1-2.2v and most motherboards will default it to 1.9. This can be a cause of instability. Don't take it further than 2.2v though unless you know what you're doing.
2. I note you're using a 9x multiplier. If you're set on using this, make sure Speedstep is disabled, leaving it enabled whilst overclocking can cause instability sometimes as it tries to step down to the 6x multiplier and change voltage while it does so.
3. Your ram timings are 5-7-7-20. I'm not sure what the Gskill kit you have is like but you should find out whether they're specified to run at 5-5-5-15, as are many 1066 kits.

If you have good case cooling and are just searching for a 100% stable 24/7 overclock, I would recommend running a 400mhz fsb with an 8 multiplier. Run your ram at a 1:1 ratio, which will leave it underclocked at 800mhz. Considering this you may well be able to drop your timings to around 4-4-4-12.

With a similar setup at the moment I'm running 430FSB with overclocked pc6400 patriots running at around 1040ish mhz with good results. I'm running it like this because despite the conjecture surrounding the issue I believe a faster FSB and mem speed has real world, not just synthetic benchmark, benefits.

Perhaps it would be a good idea for you to try the above recommendation, acheive stability and then move on from there.
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
Nice suggestions! I'm getting closer to stability!

remmeh, lostnkonfused and Bipolarbear ---> I took all your advices into account. So first I checked out the website remmeh suggested (great help!) and fixed my GTL, which is now around 66.66% (67%). Ran Prime95 for 2.5 hours on Blend mode and it went fine.
And now, like lostnkonfused suggested, I pumped up the vcore a couple of steps, and it's up at 1.248v, under load. The thing is, before it showed 1.168v, which is NOT what I had in the BIOS, I had 1.2500v. Yet now I pumped it up to 1.2750v, and thus it now shows 1.248v. What could've been the cause for that great leap?
I Enabled Load Line Calibration. Also, when I am NOT running Prime95, the Core Speed is at 2000 MHz and multiplier 6x, but when I do run Prime, it's at what you see above (3000 MHz and 9x), is that because of the SpeedStep that Bipolarbear mentioned?
Cause I can't find SpeedStep in the BIOS, my motherboards BIOS (Gigabyte EP45-UD3R F5) is quiet different from others so it may be named differently. (Isn't this speedstep good for power and temperature savings by the way?)

And yes, Bipolarebear, I have my Ram voltage set to 2.0v which is what is recommended by manufacturer, although I forgot to change the timings 5-7-7-20 which should be 5-5-5-15 like you mentioned. Now it's set to the accurate timings though (see here for the G.skill I own). Thanks a heap for that, totally forgot about it!

Is there anyway I can post my BIOS settings here, through Windows XP ? Writing it all down and retyping it here seems a bit tedious, I could do it, but I figured I'd save myself some time by finding a more appropriate path.

You guys rock, thanks again!
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
And oh yeah, Bipolarbear, you're right, I'm gonna try your first tips and the others suggestions first, then I might try the 400 FSB and underclock the ram to 800 mhz with 4-4-4-12 timing. Right now it would just confuse me :)
 
Joined
Dec 25, 2008
Messages
61
Cause I can't find SpeedStep in the BIOS, my motherboards BIOS (Gigabyte EP45-UD3R F5) is quiet different from others so it may be named differently. (Isn't this speedstep good for power and temperature savings by the way?)

Speedstep might be found in your BIOS under 'EIST'. As a general rule yes, Speedstep is good for power and temperature savings. If you're running stock than of course you should leave it enabled.

On a standard Q6600 your processor speeds will be 2400 (load) and 1600 (idle). Speedstep/EIST simply reduces your multiplier from stock 9x to 6x. Because you've changed your FSB to 333mhz, you are now getting idle 2000mhz and load 3000mhz.

If you're set on using speedstep it's a good to find your overclock without it first then fiddle around with it. Like most things, your mileage will vary. Remember also that since you're going to be upping your FSB, possibly as high as 400mhz+, your 'speedstep on' speed is going to get higher.

However, once you start overclocking, fiddling with FSB and mucking around with voltages, the power saving features can be a main cause of overclock-instability.
 
Joined
May 26, 2007
Messages
713
I'm on a p35-ds3r, q6600(3.0) with default vcore(1.2), 4gigs pc-6400. Temps on cores are about 40-36c idle and 58-60 load and cpu temps is about 42c with prime95.
 

lostnkonfused

Limp Gawd
Joined
Feb 23, 2008
Messages
133
Also, sometimes they might label speedstep as "C1E Halt State" as well.

An easy 3.6GHz would be 450 x 8 multiplier.
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
After all the changes everything seems to be working fine. I did 2 overnight tests, one of 9 hours and one of 12 hours, blend and small FFTs torture testing. Went smooth.
My temperature went all the way up to 55C at one point though...maybe reducing of voltage again should fix that?

Thanks though, you guys truly educated me on this thread :)
 

sirmonkey1985

[H]ard|DCer of the Month - July 2010
Joined
Sep 13, 2008
Messages
22,230
its c1e halt state, lostnconfused.. i just helped a friend overclock with this board..


verklighet if you want to get higher then 3ghz id say set your memory multiplier to 2.00b (if your using the default bios that came with the board) then work on your overclock.. once your at what you want and its stable, start raising the multiplier til your as close to 1066 as possible and make sure the memory is at 2.1v.. the gskill has a lot of problems running correctly at 1066.. so keep an eye on it.. my buddy is using gskill 1066 as well.. and its been rock stable at 1069 at 2.1v so far.. odds are the idiots complaining about the memory dying on newegg never bothered to set their memory voltages..

the one thing i did notice with that board is that it has an insane amount of vdroop with load line calibration off.. even with it on it has a crap load of vdroop.. off it was .15v vdroop, on it was .03v vdroop.. (at full load) so keep an eye on that.. also i would not recommend updating the bios if you are use to the stock bios (unless you already have).. i noticed it changes a lot of settings/features with the newer bios...

also 55c isnt that bad.. id actually work on trying to raise your overclock at the current voltages you have.. you should still end up staying under 60C.. remember stress test temps are unreliable because there is no possible way with current multithreaded applications that you would ever see those same temps.. i doubt you would even hit 50C running any application that claims its running at full load over all 4 cores..

but like i stated in the beginning of this post.. when overclocking always drop your multiplier down on your memory before doing anything so that you remove it as a variable..
 

verklighet

Weaksauce
Joined
Dec 31, 2008
Messages
84
simonkey1985, Hey man, thanks for the thorough input! I'd like to stay with 3.0ghz for now, just to have that in my luggage, you know, my first stable overclock. The experience of any instability on this level will make me aware of any other variables (or current), and how they work. Yet, I'm sure my thirst for higher clocking will come soon anyway :)
Thanks for the heads up on the temperature, got me a bit worried. Although this might sound as a very noobish question, will this overclock truly affect the mileage of my build? What experiences have you guys had with CPU's, memory or other components failing on you earlier than there 'best before' date?

Another question I made earlier in this thread, that never really got answered was, is there a way to get your BIOS info in Windows XP? So I can post it here. I remember seeing another thread here in OC where a guy posted all his info in a box within the thread...hm. I'll see if I can find it again.

Thanks again!
 
Top