I76700k OC - ASUS 5 ways optimization or Old School BIOS digging?

[Gary A 8]

and Asus know better.

 

Hi Martin,  I do not agree with that broad statement.  Not wanting to bash or discredit ASUS or anyone else for that matter either.  However ASUS are in the business of sales, period. In this case the sales of a MB. If ASUS can hype that their board can do more than the other guys board then that = sales. ASUS would have to have no material gain and otherwise be impartial for me to subscribe to that they know best.

 

My comparison of 4.6 to 4.6 is apples to apples no variables.  RealBench to RealBench same settings in RealBench (lets not forget RealBench is designed by ROG, an entity sanctioned/supported and authorized by ASUS), same room which is temperature controlled. Fan Expert was set to "bench."The only variances were the adjustments that I made in the Bios vs. whatever adjustments 5WO made.  The net result was manual settings allowed the same OC for 10c less temp.

 

I do not believe that 5WO is doing anything more special in the manner that it overclocks because ASUS know best.  My hypothesis is that the 5WO/AI 3 sets balls to the walls high voltages in attempt to quickly achieve a high OC.  Further that these high voltages are not necessary to sustain the OC result given as 5WO final.  Actually it (5WO) simply follows the CPU VID instructions for voltage, this was stated by Raja in one of his posts.  To be fair and to further paraphrase Raja, he said that these voltages were necessary as determined by intel via the VID instructions in order to remain stable at frequency. However my argument to that is we have been overclocking CPU's for many years in the absence of ASUS's AI 3/5WO program with seemingly good success.  We did this by utilizing benchmarking and temperature software to determine stability, and we didn't need or follow VID instructions to obtain our results.

 

I also ran Intel's own "Extreme Tuning" stress test against my manual OC and passed.  You would think this would dispel further notion that perhaps ASUS AWO test is somehow more stringent and detecting errors that other tests are missing. Result  Then there is, what about all the other MB manufacturers software OCing is it less than ASUS.

 

I am just affording my personal opinions/arguments against your very good questioning.

[martin-w 3,235]

Hi Martin,  I do not agree with that broad statement.  Not wanting to bash or discredit ASUS or anyone else for that matter either.

 

 

 

You shouldn't assume I mean in such a broad sense Gary, from just three words.  I merely meant in terms of experience overclocking the platform. When a new platform is released, like Sklyake, Intel supply the chips to companies like Asus well in advance. They actually get the opportunity to learn about the platform many months before we do. They literally test hundreds of CPU's to determine safe voltages and overclocking headroom. Asus, or any of the top companies engineers are academically well qualified in this regard, we aren't. Their engineers go to college and university to be educated in the field, we don't. So I don't mean Asus are all powerful genius's that always get it right, I simply meant that they have more experience and academic qualifications pertaining to this field and more experience in the new platform than we do.

 

Given the above, I see it as plausible that you and I might feel our overclocks are stable, after testing in a limited way with our synthetic stress tests, and simply hoping we are right... whereas Asus are technically qualified to determine if  more voltage or less frequency would be appropriate, as they have greater experience in the platform than we do, not to mention close ties with Intel. They won't always get it right of course, thanks to the variance in a CPU's capabilities referred to as the "silicone lottery".

 

 

 

However ASUS are in the business of sales, period. In this case the sales of a MB. If ASUS can hype that their board can do more than the other guys board then that = sales. ASUS would have to have no material gain and otherwise be impartial for me to subscribe to that they know best.

 

 

Yep, agree with you entirely. All companies will over hype their own creations, 5WO included .Again, my comment was merely in regard to Asus having received the CPU's way before we have and their "qualified" engineers having tested a multitude of chips with sophisticated equipment we don't possess.

 

 

My comparison of 4.6 to 4.6 is apples to apples no variables.  RealBench to RealBench same settings in RealBench (lets not forget RealBench is designed by ROG, an entity sanctioned/supported and authorized by ASUS), same room which is temperature controlled. Fan Expert was set to "bench."The only variances were the adjustments that I made in the Bios vs. whatever adjustments 5WO made.  The net result was manual settings allowed the same OC for 10c less temp.

 

 

For more heat to be generated as a result of using 5WO, there MUST have been a variable. What was it?

 

I suggest voltage. I suggest you didn't use the same voltage with your manual overclock as 5WO used. I posed that question in my last post and you didn't answer, so I'm not sure if you agree on that.

 

 5WO can't magically generate more heat unless it has used higher voltage or another variable has been modified. I'm just speculating as to the cause, as you didn't take notes, so we are just relying on your memory.

 

 

The only variances were the adjustments that I made in the Bios

 

 

And was that lower voltage than 5WO used. Or another variable that you entered? For an extra 10 degrees to be generated there MUST be a variable. Was it adaptive voltage for your manual overclock and offset voltage for 5WO for example? 5WO only uses offset currently due to Intel microcode issues.

 

Don't forget, true stability is long term stability. 5WO may provide that wheres as your [i assume] lower voltage might not. We will know in hindsight.

 

 

 

I do not believe that 5WO is doing anything more special in the manner that it overclocks because ASUS know best.

 

 

But neither of us can say that for sure. Neither of us have a clue what routines are running during the automatic 5WO stress tests! 5WO may well be analyzing the CPU in a way that reveals instability that you and I can't detect with our own choice of stress tests. Even RealBench may not be stressing the CPU in the same way. We simply don't know, so neither of us can be definitive. Asus may be aware that in order to achieve "long term" stability a tad more voltage is advisable.

 

 

 

 My hypothesis is that the 5WO/AI 3 sets balls to the walls high voltages in attempt to quickly achieve a high OC.  Further that these high voltages are not necessary to sustain the OC result given as 5WO final.

 

 

 

Oh right, so you do agree that the variable at play is "voltage".Namely that you used a lower voltage than 5WO, hence your lower manual overclock temperature? So as I said previously... how do you know whether,  in 6 months time, your overclock will show sighs of instability and the slightly higher 5WO voltage won't?

 

 

Yes, 5WO does seem to add a little extra voltage. But it's certainly not anything like "balls to the wall". I've seen plenty of examples of 5WO overclocks at quite reasonable voltages. And as I say, true stability is determined after a protracted period of time, running real world applications. 5WO's slightly higher voltage may well be superior in that sense.

 

 

 

 

 To be fair and to further paraphrase Raja, he said that these voltages were necessary as determined by intel via the VID instructions in order to remain stable at frequency. However my argument to that is we have been overclocking CPU's for many years in the absence of ASUS's AI 3/5WO program with seemingly good success.  We did this by utilizing benchmarking and temperature software to determine stability, and we didn't need or follow VID instructions to obtain our results.

 

 

That might be a bit arrogant. It's assuming we know better than someone who could be regarded as a qualified expert. I presume Raja has been to college or university and gained qualifications in this field that we don't have? I presume Raja has tested countless CPU's in his day job? I presume Raja and Asus have contact with Intel that we don't have? I assume Raja has overclocked more CPU's than we have had hot dinners? 

 

It also goes without saying, that Skylake is a new platform. 

 

 

 

I also ran Intel's own "Extreme Tuning" stress test against my manual OC and passed.  You would think this would dispel further notion that perhaps ASUS AWO test is somehow more stringent and detecting errors that other tests are missing.

 

 

 

Well no, it doesn't dispel anything. To truly test for stability a wide range of stress tests are required. I've had the experience many times, of passing the most commonly used stress tests and then failing in the software I run day to day within seconds. We also need to remember that most of these stress tests are synthetic and bare little resemblance to the software we run 24/7. 

 

I've also had the experience of assuming my rig is stable, after "extensive" stress testing, only to find, some time later, months in some cases...  instability rears it's ugly head. A tad more voltage is then required. I'd be surprised if you haven't had a similar experience over the eyars... so perhaps why 5WO gives you a tad more voltage.

 

At the end of the day Gary... you may find that in a few months, what you assume to be a stable overclock isn't. Or you may find yourself running software you only run on occasions... and it fails.

 

Then there is, what about all the other MB manufacturers software OCing is it less than ASUS.

 

 

 

Nope, absolutely not. Some of the other manufactures have pretty good software. In my previous experience Asus auto rules are better though. Ironically lower voltage.

 

It's also worth mentioning that even in this very thread, their have been comments from those very happy with the results from 5WO, and not complaining about "balls to the wall" voltage.

 

Hopefully in a few weeks I'll be able to test this and let you know. Who knows, I may be agreeing with you entirely and cursing Asus.  :smile:

[Gary A 8]

 

 

I may be agreeing with you entirely and cursing Asus

 

I am not bashing or cursing ASUS in any manner.  I gave a reasonable account of my actual (non hypothetical) experiences to the OP's question of 5WO or manual oc.  What anyone chooses to use is their business.  5WO is not necessarily terrible, I personally currently prefer manual oc.

[martin-w 3,235]

Yep, fair enough Gary. And I value your feedback. I didn't mean to interrogate you or anything like that. It's just that if it's true that 5WO is inconsistent and thus not very polished, and there's evidence of that replicated by others, I wouldn't even bother with it, I would then stick to my previous preference of overclocking manually.

 

The above is why I've been countering your claims with an opposing view.