Desktop Utils (QuickBoost / TurboV) vs. BIOS

[enright]

Hi All. Long time PC user / tweaker here who has never really delved into overclocking - until now.

Quick question: for the "casual" overclocker - someone that wants to get an additional 20% - 25% out of a Core i7 920 - are the desktop utils offered by Asus and Gigabyte (e.g. TurboV and QuickBoost) any good?

Here's the thing - I've seen a few OC guides and they all seem to have a slightly different approach in terms of which BIOS params to tweak, and in what order. I find it all a bit intimidating, but would still like to take advantage of the extra headroom in my CPU if there's an easier way to do it.

I realize you probably don't get the MAX amount of boost with the desktop utils vs. manually tweaking settings in the BIOS, but do they work at all? Is one any better than the other (e.g. ASUS vs. Gigabyte)?

Thanks!

John

 

[EarthwormJim]

I use TurboV, but not for my main overclocking settings. Just to let me bump my voltages up maybe a tad if I failed a round of Linx or Prime. Saves me a reboot.

There are much less settings available, so you probably will not get the same maximum level of overclocking using those programs. Plus you're at the mercy of those programs, should you update them or reinstall Windows; your overclocking settings are gone.

If you can overclock using those programs, you can do it in the BIOS. None of the settings have different names.

 

[enright]

If you can overclock using those programs, you can do it in the BIOS. None of the settings have different names.

I guess I just like the idea of dumbing the process down to "low, medium, and high" at least in the case of QuickBoost:

http://www.hardwarecanucks.com/foru...te-ga-ep45-ud3p-p45-motherboard-review-8.html

Since I'm just going for a medium level of boost, maybe this util is enough? I guess what you loose is the fun of mastering the settings, and getting every last mhz out of the CPU.

John

 

[thefreeaccount]

I can't seem to find it, but wasn't there just a thread where some guy's i5 was getting 1.4v at speeds barely above stock because he'd pressed some kind of "overclock button" on his Asus motherboard by accident? I would sure not trust any of those automated programs not to do anything stupid or dangerous.

Edit: n/m, it was actually an i7 but I found it: http://www.hardforum.com/showthread.php?t=1487748

 

[EarthwormJim]

I guess I just like the idea of dumbing the process down to "low, medium, and high" at least in the case of QuickBoost:

http://www.hardwarecanucks.com/foru...te-ga-ep45-ud3p-p45-motherboard-review-8.html

Since I'm just going for a medium level of boost, maybe this util is enough? I guess what you loose is the fun of mastering the settings, and getting every last mhz out of the CPU.

John

I would not suggest using any auto settings (low, medium, high). They often WAY overshoot voltages.

There isn't much to overclocking really if you're looking for modest overclocks.

Increase FSB/QPI by 10mhz or so till Windows blue screens upon bootup. Once you get a blue screen, back off by 10mhz or so. Run Linx for 100 passes or so. If it passes, run Prime95 blend overnight, if not back off by a couple mhz till it passes. If P95 blend passes, run memtest for a pass or two, if not lower your FSB/QPI by 5mhz or so till it passes. If memtest fails raise your memory timings a bit.

You won't get too far probably without a voltage bump (or tweaking other settings), but you don't seem to want to overclock that much anyway. This should give you a 10-20% overclock.

Quite honestly if you can't be bothered to overclock manually, I would not do it at all. I don't mean that in an insulting way at all, just you might be setting yourself up for more headaches.

 

[enright]

I would not suggest using any auto settings (low, medium, high). They often WAY overshoot voltages.

Quite honestly if you can't be bothered to overclock manually, I would not do it at all. I don't mean that in an insulting way at all, just you might be setting yourself up for more headaches.

No offense taken. I don't mind tinkering with the BIOS - but I am the type of person who obsesses about whether or not something is optimal. What is the best ratio of performance to stability?

Here's what I don't get: figuring out what optimal setting is for multiple variables that will get you to that "sweet spot" seems like *exactly the kind of thing that is best done by a software algorithm* and not manual trial and error.

To put it another way - what is the advantage of manually trying different setting combos? Even if the process requires trial and error due to random factors unique to each CPU - why not automate the trial and error process and have an application solve the equation for you?

So I guess that's my real question: DO the apps bundled with ASUS and Gigabye mobos do that? Do they automatically determine the optimal settings for your particular CPU / MB / RAM based on trial and error or some intelligent formula? Or do they just assign hard-coded "small, medium and large" values?

John

 

[fffblackmage]

What is the best ratio of performance to stability?

I don't believe the relationship between system stability and performance is a linear one. At some point after increasing frequency, you'll start failing one of those tests, but increases frequency even by +1MHz might just cause your computer to not even boot anymore. Don't take that as a fact. That's just how I see it from experience.

If I use my computer as an example:
My rig is pretty stable at 3.05GHz. At least I havent had it crash on me yet.
At 3.08GHz, it's actually crash once every couple of days.
Anything above that and my computer won't post.

30MHz isn't a very large increase (well, I guess that actually depends..). However, stability changes significantly. Not all cpus are the same either. So, there is no real set point of "best performance to stability," it's just something you have to find yourself through experimentation.

To put it another way - what is the advantage of manually trying different setting combos? Even if the process requires trial and error due to random factors unique to each CPU - why not automate the trial and error process and have an application solve the equation for you?

One problem with the autooverclocking thing I had with using ntune (years ago...) was that the computer would freeze and the program was unable to reboot the computer... well because it's frozen....

Another problem could be that the instability of using such a program could potentially corrupt the OS. Seeing how the people who would use such a program isn't knowledgeable enough to do it themselves... I don't think it's reasonable to expect that they're able to fix the corruption themselves very easily.

Hopefully, that made sense and was useful. >_>