SOLVED Proper motherboard voltages/timings

BIOS settings

Can I do a screenshot of the BIOS, and paste it in windows, or do I need to use another medium... lol. Also, there is under the timings section, auto, dct1, dct2, and both, when I turn it to both then all the timings for both slots are displayed. When I hit enter on the other 4 timings there's all the 'clocks' in a list and auto, however when I hit enter on the cycle time(tRAS) there is only auto. I've also noticed now after going to the previous overclock that even with the other timings it's at 8-8-8-20-29.

mixin12s said:

I tried to do 230MHz on the FSB again for a processor speed of 3.450, and again it would not boot into windows. It did give me a bsod this time though, which it hadn't when I tried with the 7-7-7 ram timings. The bsod it gave me said USB Driver BUGCODE or something like that. I then tried 230MHz again with the 685MHz ram timings as you indicated I may have to, and it booted successfully. Also, I tried adjusting the voltages but it still wouldn't let me as far as I know. Specifically it read- Left+Shift: step +10 Right+Shift: step +20 for the processor. For the cpu-nb it said Left-Shift: +10 Right-Shift: -10 .... none of which any combination of pressing did anything. However in the H/W monitor the cpu-nb is at 1.208 currently.

Click to expand...

If the 9-9-9 RAM timings, which you need for that speed, and faster clock worked I'm not sure why you went back. Was it unstable?

Looking back over the other posts you might want to increase vDIMM (RAM voltage) to 1.55v.


Also, I may know what the problem is with those voltage settings. I forgot you had an MSI board (not in sig). MSI's have TWO voltage settings for the CPU and cpuNB - "VDD" and "voltage" for both. From trying to help others with MSI boards I've discovered you have to set both for a component for everything to work right. Try 1.45 CPU VDD and 1.45 CPU voltage to see if that works. You may also have to set one of them (can't remember which) slightly lower/higher than the other. A guy in Hawaii had that issue. We kept having to set one to 1.46 and the other to 1.44.

stability

I only went back to the other oc after seeing the cycle time at 20 clocks when it should be at 25 with the 9-9-9 settings. The cycle time (tRAS) only allows for auto setting in the BIOS. However even with the 8-8-8 timings it sets the cycle time to 20. As far as the voltages, all of them are greyed out on auto... although according to the site I found reviewing my BIOS they should be able to be adjusted. I found a site that shows screenshots of just about all the settings in my BIOS here.
http://www.legitreviews.com/article/1004/3/

Currently, I just got done flashing the BIOS to the most current version to see if that helps any at getting the cycle time to the right settings with the overclock.

Those voltage settings should be farther down on the Cell page - is that where you're looking? Or are you just looking at the top of the Cell page where the "CPU Specifications" are?

Voltages

I figured it out. They're auto for the voltages unless you adjust them, and to adjust them I have to use the + and - keys on the number pad. As far as the RAM timings go... even after the BIOS flash it's still only giving me an auto setting for the cycle time, and still setting it at 20 clocks for all of 7-7-7-auto-26, 8-8-8-auto-30, and 9-9-9-auto-33 timings with the overclock applied. Here is a ss I found on google of my memory timings section, only in the ss where it says 24 clocks on his, on mine it says auto, and if I hit enter the red box that pops up with the list of possible timings only has AUTO. If I remember from past pc's I owned, isn't there freeware software out there that can change the RAM timings from in windows, like Systool... could I use one of them to manually adjust the cycle time to the correct setting for the MHz I want and use cpu-id to verify the change? Is it possible it's auto setting it to 20 clocks because I haven't adjusted the voltages yet?
I was also wondering what exactly I should do with the voltages, just set them manually to what they're currently at? I don't know if it's wise to do too much with them as I read this on Legitreviews.com referring to this board: "We followed our typical overclocking rituals, and after about 15 minutes or so, we raised the vcore up to 1.45v. When we did that, we got a "pop" from the board, and then the all-too familiar smell of electrical death was in the air... we killed the board before we got to do too much with it in overclocking at all."

You're already out of step with the way I overclock. Before I even start overclocking I manually setting everything I might change to their stock settings and work my way up from there in small steps. I hate Auto for the important settings because you never know what the BIOS will do to it when you change something else. My way may be slower but I've never had any rude surprises. If there's a potential problem I get plenty of warning.


I have no idea what that site was doing so I can't comment on their results. If that board is one that adds vCore to the manual setting, and some do, then maybe they pushed it too far jumping to 1.45v - or maybe it was just a bad board to begin with. Theirs isn't the first MSI I've heard of going bad long before it should. Manually setting 1.45v shouldn't kill anything since that's still in the "AMD safe" range for many CPUs.


PS
If you want to stay in the "AMD safe" range then you're topped out, now.

OC/Timings/Voltages

Well, the system seems stable so far with the current settings of 230MHz FSB, processor @ 3.45GHz, RAM 9-9-9-20-33. Currently the voltages on auto are CPU 1.4v and Cpu-nb 1.184v. The main thing I'm still wondering about is that cycle time auto setting to 20 clocks when JEDEC for those timings says it should be at 25. I mainly need to know if it makes it through stress testing fine, if having the cycle time set that low has a negative effect in any way. For the voltages however, you said cpu-nb 1.2v and memory timings 1.525v, and what about trents recommendation to turn the vcore up to 1.45v? Also, from the looks of it their overclock in the review that's quoted from, didn't damage the cpu, they blew a voltage regulator on the motherboard.

If the system is Prime95 Blend stable for an hour then don't worry about the RAM timing.

I have no problem with the vCore at 1.45, plenty of people have gone much higher than that without problems, but you seemed reluctant. I don't push people to overclock. If they don't feel comfortable doing it than they shouldn't.

It usually is a board that goes out, though the CPU can fry when people pump too much voltage for too long or at too high a temp. In the case of that board, it's not the first time I've seen Dr.MOS lose it. I've said it before, MSI needs to kill Dr.MOS, or at the very least send it back to the labs, and get back to basics.

Timings-YAY!

Well, I got the cycle time set correctly. Trents was right, I did have to use the +/- keys on the numberpad, if I hit enter, it only shows auto in the list, even though all the other timings show all the clocks, but then hitting +/- allowed me to cycle through the available clocks for that setting. I was playing around with the voltages, and noticed I can set the cpu-nb vdd to 1.2v, but when I try to set the cpu-nb itself to 1.2v the timings go from 1.18 to auto to 1.22 when I hit the +/- keys, so I'm assuming it sets that to 1.2v automatically on auto. Also, for the memory voltages it only allows me to set to the hundredth of a volt, allowing me to set it to 1.52 or 1.53, not 1.525... and was wondering if mfr spec is 1.5 if I should still try to up it or not. Below are the current settings/specs and temperatures while undergoing prime95 stress testing. EDIT: While undergoing the said stress testing 20 seconds after posting that it gave me a bsod for 'page fault in unpaged area'. Backed overclock down to previous setting of FSB 230MHz, processor 3.45GHz. I was however doing a custom torture test using 6Gb of the 8Gb of RAM, so could it just be that that was too much RAM for it to use at once for testing in addition to what my pc was using otherwise? Also, according to the SPD tab in Cpu-id it says 9-9-9-25-34 for the 685MHz timings, but below it's showing it as 626.7MHz, is that normal? Sorry about all the questions... lol.

For the vDIMM I always use spec + 0.05v, which would be 1.55v in your case, so 1.53v will be OK. Better to have a hair too much than a hair too little.

I edited the above post QI, please re-read. I adjusted the RAM voltage to 1.53, however there was another setting below that that said it needed to be half of the RAM voltage, and = to another voltage... I left that one on auto since there was no voltage setting anywhere I seen for the value it had to be = to. Could the RAM voltage have had anything to do with the 'page-fault' bsod I got?
Update: After adjusting the voltage on the memory to 1.53, and re-setting the FSB to 235, I am currently once again going through stress testing with the overclock of 3.525Ghz on the processor. I also adjusted the memory timings slightly to 9-9-9-25-34, which is what the timings are in cpu-id under JEDEC#4. Originally, I had them set to 9-9-9-25-33, mimicking the one clock less for the last timing as my BIOS had for the default settings just in case. I decided to try to put that up to the recommended setting, along with the increase in voltage to see if I can now get through stress testing without the bsod I experienced earlier. As of right now during torture testing the cores are hovering around 41C, cooler than before even, just turned case fans up a notch. Also, I didn't turn the processor/cpu-nb voltages up at all as the system is stable so far with the current cpu core voltage of 1.4v, and for the cpu-nb because it is automatically set at 1.2v, and as mentioned before when cycling through the settings with +/- it goes 1.18v - auto -1.22v, and so auto in this case I believe is 1.2v. Also, I've read some other overclockers prefer to see how far they can OC on stock voltages before they up them, and since I don't see myself going too much higher, if it's not broken at 1.4v why try to fix it?
Anyway, Thank You QI and Trents for helping me with all this and answering all my questions.

If you need to increase the vcore a little your cool temps certainly give you room to do that. You're doing a good job.

voltage/temps

I realize the temps give me the room to do that, as well as the processor specs which is up to 1.5v. I'm more worried about having a similar incident like the one review website that I posted quote from above where they blew a voltage regulator on the motherboard, and then also QuietIce mentioning he's seen msi boards go before they should too. My question basically on that is, if the system is running stable with the processor on the default 1.4v, is there any drawback to not raising it to 1.45v until/unless I find a need to?

OC

Well, I tried to go up to 240MHz on the FSB for a processor speed of 3.6GHz, and again it did what it did before when I tried to go above 225 with the RAM timings set lower than 9-9-9-25-34... gave me a bsod during loading windows with USB DRIVER BUGCODE for the error message. Could that be because I haven't raised the voltages on the processor yet? Also, QuietIce, I think I found what you mean by the CPU VDD and CPU core voltages, and I think (and you may want to tell your friend) that they are automatically set the same unless you set the cpu core voltage higher or lower than the CPU VDD... as whenever I try to set them to the same value, when I try to set the CPU core voltage to the same as the CPU VDD voltage it says [Auto], and then if I hit +/- it'll go to the next step higher above or below what the CPU VDD voltage is at, and I tried it with the CPU VDD voltage at different settings, no matter what, when the values are the same for both it changes the CPU vCore voltage to [Auto]. Anyway, please get back to me on whether low voltage to the processor could be the cause of that bsod I mentioned, or if I've probably reached my max. Currently still at FSB 235MHz, CPU 3.525GHz, RAM 9-9-9-25-34.

mixin12s said:

Could the RAM voltage have had anything to do with the 'page-fault' bsod I got?

Click to expand...

That's a difficult question to answer because RAM voltage, speed, and timings are all inter-related. For the question I think you're asking, no it's probably not the RAM voltage - but keep in mind that's a very simplified answer and bit wrong on some levels.

mixin12s said:

Update: After adjusting the voltage on the memory to 1.53, and re-setting the FSB to 235, I am currently once again going through stress testing with the overclock of 3.525Ghz on the processor. I also adjusted the memory timings slightly to 9-9-9-25-34, which is what the timings are in cpu-id under JEDEC#4. Originally, I had them set to 9-9-9-25-33, mimicking the one clock less for the last timing as my BIOS had for the default settings just in case. I decided to try to put that up to the recommended setting, along with the increase in voltage to see if I can now get through stress testing without the bsod I experienced earlier. As of right now during torture testing the cores are hovering around 41C, cooler than before even, just turned case fans up a notch. Also, I didn't turn the processor/cpu-nb voltages up at all as the system is stable so far with the current cpu core voltage of 1.4v, and for the cpu-nb because it is automatically set at 1.2v, and as mentioned before when cycling through the settings with +/- it goes 1.18v - auto -1.22v, and so auto in this case I believe is 1.2v. Also, I've read some other overclockers prefer to see how far they can OC on stock voltages before they up them, and since I don't see myself going too much higher, if it's not broken at 1.4v why try to fix it?
Anyway, Thank You QI and Trents for helping me with all this and answering all my questions.

Click to expand...

With the exception of adding that 0.05v to the stock vDIMM I always start at stock voltages (but set manually) and only increase them as needed. To increase the voltage while a system is stable is just crazy. As long as it's stable you keep increasing the speed. When the system becomes unstable you have to make a choice - increase voltage or live with what you've got. If you don't want to push the vCore past 1.40 that's OK we all have our limits - but we don't stop pushing the system until we're sure we've actually found the best clock that voltage will give us.

mixin12s said:

Well, I tried to go up to 240MHz on the FSB for a processor speed of 3.6GHz, and again it did what it did before when I tried to go above 225 with the RAM timings set lower than 9-9-9-25-34... gave me a bsod during loading windows with USB DRIVER BUGCODE for the error message. Could that be because I haven't raised the voltages on the processor yet?

Click to expand...

Just to check, at a 240 MHz clock the HT Link is still below 2000 MHz and the RAM isn't faster than 667 MHz (should be right at that speed, if my math is right). How fast is the cpuNB running? You may want to try turning it down a notch just to see what will happen.

Yes, it could be that you're almost at the end of the 1.40v setting. If the cpuNB speed change doesn't work you might try increasing just 2-3 MHz instead of the full 5 MHz. If you're at the top and aren't going to increase the vCore you may as well tweak every tiny bit you can.

Voltages

Well, upon further examination of the BIOS, there's the settings you mentioned, CPU VDD voltage and CPU vCore voltage... the CPU vCore would allow me to turn up the voltage past 1.4v, however the max setting for the CPU VDD whatever is 1.4v, then auto, then lowest possible setting when cycling through them with +/-... so I need to know, can the CPU vCore voltage be higher than the CPU VDD voltage? Also, set the CPU-nb VDD voltage to 1.2 with the CPU-nb still set on Auto as when I try to set it to the same value as CPU-nb VDD it just says Auto anyway. I attached a ss to show my BIOS voltages section, with the CPU VDD highlighted for the description on the right side. EDIT: Upon reading another post on overclockers, I have an understanding that CPU VDD, and CPU-NB VDD, are extra controllers or something like that... I read this on another forum thread on your guys site:
CPU VDD is the voltage regulator of the CPU-V
CPU V is the voltage of the CPU itslef
CPU-NB VDD same idea as the CPU VDD, but for the CPU-NB
CPU-NBv is the voltage for the CPU-NB

So I guess I need to know what to set these too, if I just adjust the CPU-NB Voltage, and just the CPU Voltage, or if anything has to be done to the other 2 settings as well, so far from what I'm reading online I don't think I should have to adjust the 'VDD' ones, only the actual voltage ones (CPU Voltage and CPU-NB Voltage). Is this correct?

Voltages update

Well, I adjusted the CPU voltage to 1.42v and it booted with the 240MHz FSB and 3.6GHz on the processor. I guess that was the cause of the bsod after all. At this point I guess I'm wondering if I should try taking it any further on a stock cooler(provided it makes it through stress testing)? I don't know if I need it I guess. This processor was more than a step up, as my old processor was a 2.8GHz dual core, and I don't even know if I ever really used all of that... lol. The only thing I use this pc for atm is gaming... and come fall for school running programming, database, and design applications on. I think the only reason I even took it this far was out of it being a new machine, and wanting to see what it could do. 3.6GHz was farther than I thought it would go, although I have read reports online of people getting them to 4.0, although I doubt on a stock cooler, and the stock thermal compound that comes on it as well.

What does CPU-Z show as the vCore at load?

Voltages and Current OC

Well, I think I realized that the max I can probably go with this hardware combination is 250MHz on the FSB. At that setting my RAM will be at 666.7/8MHz, and the HT Link will be at 2000MHz with the x8 multiplier, so short of going down another RAM divider, and going down further on the multiplier that will be the max I can achieve with this combination of hardware. I tried to go up to that, and it bsod'd immediately upon stress testing. When I put it back to 240MHz it still wouldn't boot without bsod'ing on me, so I had to load optimal defaults and then try again from there. While in the BIOS again however, I noticed I can adjust the CPU-NB voltage, just not to 1.2v, it gives me the option for 1.195v, and then the next step up is 1.205v, so I chose 1.205v. If it makes it through the stress testing at this OC stable with that voltage now being adjusted I may then attempt 245MHz, and 250MHz. I have attached a ss for your reference as to the current settings. With the new voltages for the CPU-NB, and the CPU Voltage at 1.42v temps are hovering at 50C so far during blend testing. Also, this is at load... in answer to your question QuietIce, with the voltage on the CPU set to 1.42, under load it is at 1.424v.