"Furthermore, oc-ing is not what it used to be. " as per Rob Franks, and yes indeed one-clik overclocking will appear on new mobos.

There is actually a good reason as to why that is now true. The sale of computer components today is mostly driven by the gaming market - not withstanding specialty markets like NLE workstations - which is a very very small part of the marketplace.

In order to keep selling new mobos the manufacturers have seen the opportunity to graze in the gamer's field unlike they ever did before. One of the reasons that the O'clock setting is really not the default is that there is hope the manufacturer can still sell some kind stable systems for the bread and butter market like corporate desktops. But that segment has been steadily shrinking for years.

The way I see it, as you mentioning car. Well, I did couped up my Corvette Convertible to a 1000hp tubocharged 383ci stroker motor under the hood. You may ask, why need such insane hp? because I can and I love having the hp on demand. You don't understand until you sit on one. Same apply to computer. Having a super fast computer, I enoyed every times I use the mouse clicking on something and magically happen instantly.

I recall hearing similar yelps from the kids the other day, like why can't I have my supervisors job now?, or I should be able to buy a house without a downpayment?

Is this the millennial generation ?, ME NOW, or whatever, they call them? I've long since lost track.

Take it fast, take it now, there maybe no tommorrow.

hmmm I clicked your name to send and email.... I'll just post info here when I get home.

I did go into the bios and TRY 4.4ghz (was 3.50) using the built in presets. seems to run fine, and the temp in my liquid cool was not much higher... things actually FELT zippier is usage, adn Novabench and Cinebench both went up noticeable. No glitches, hiccups, lock ups but didn't use for very long.

4.60 was a bios option also, but the 4.80 option was in RED meaning don't do it obviously

wwjd,
You should really use torture test (prime95) to find stabilty. Cinebench only to find how fast cou/gpu.

" Just wonder if overclockers drive their vehicles with the tachometer pinned in the red as well?... Probably not!

That's actually a bad example. Manufacturers set their equipment up not necessarily for peak performance but rather for universality.... sort of a "one size fits all" setting.

The automatic transmissions on cars for example are set up to present a "soft" shift" strictly for reasons of comfort. Performance however suffers as a result. After market shift kits are available to minimize slip between shifting in order to regain that performance. Of course you feel the shifting pretty good thereafter, but then those who install shift kits aren't that concerned with comfort.

It all depends on who you are and what you're looking for.

"There is actually a good reason as to why that is now true. The sale of computer components today is mostly driven by the gaming market -"

And in fact that's exactly what I bought... a gaming mobo. Not that I play. Indeed I don't have any at all on my machine. The options/connections on these mobos however are insanely endless. These things come very close to washing your dishes and doing your laundry for you.

thanks Bruce I will give it a try

k, been running PRIME95 full bore on 8 CPUs 100% for 3 hours, no lockups hiccups etc... temps hit 80c MAX and hover around 75... is that fine for a i7 3770k?
the overclock is from 3.5 to 4.4 now

Look good to me. Were you runing Blend or small ff? I always run small ff.

i think I started the BLEND one that did everything.

how long do I have to bake this test for stability? overnight?

I don't like to run it more than 30 minute in small ff P95 on mine. After all I mean this is a torture test, it really put a beating on the cpu. If I were you I'll leave it, unless you are getting bsod in Windows.

"Manufacturer's have a reason for setting recommended speeds as they do. Serious business users can't afford the inevitable lost time (money) that
overclocking usually leads to."

That's not entirely true. If you are talking about PC manufacturers, then in general, they won't take the time to properly OC every PC they make, and they probably don't want to deal with the warranty returns when something breaks. PC makers, except for the gaming PC makers, just assemble a box of parts, plug in a test boot disk, burn it in for a day then slap their label on the front.

Chip manufacturers test chips while on the die with automated equipment. If too many chips on the die fail, the whole die is scrapped. (Melted down, actually). All of the working chips on the die are tested and the whole die is destined to meet the lowest performers' spec. This lowest spec testing is what overclockers depend on. Many of the chips on the die just happened to be on the die that was destined to be a lower-spec part. The individual chips are usually just as good as the higher performance part. Sometimes better.

On overclocking,

Hulk said: "[I]Only YOU can determine if it is worth your time.[/I]"

I've done it a lot over the decades, but now it's not worth my time.

I've overclocked for years. Currently my system is 12% over stock, not much, but still enough to make a difference. It's fun, like tweaking a car engine, and has NO bad side effects.

To a very large extent, limits are temperature-related. So the very first thing you do is install a temperature monitor. I use Motherboard Monitor. Then you go into the BIOS and up the frequency and/or the multiplier. Run Prime95 again and see hot hot the CPU gets. Keep incrementing until eventually the system hangs. Okay, no damage done, back off a bit and run Prime95 again. Eventually you find the thermal limit point. Below it, everything is great. Above it, the system hangs.

Now take off the stock cooler and put on a bigger aftermarket one ($30). A decent air cooler drops the temps by ten degrees. Now you can overclock even more, if you wish. You can go to water cooling if you want to get more extreme. Real fanatics use even more exotic cooling techniques.

But there is absolutely no risk in any of this stuff. The vendors bin the CPUs conservatively, assuming a wide variety of climates and dust on the cooler after a year.

I never fiddle with the dozen-or-so ram settings. Too much hassle for too little payback.

Prime95 can be use to stress your CPU and also ram. If it runs overnight on Prime95, it will chug away very happily on video encoding.


EDIT: I don't use the original Prime95, by the way, but a version that has a really nice GUI called "Orthos by Johnny Lee." And I use the blend setting which stresses the ram also.

And we are not "hurting" the CPU by running Prime95 (or rendering, for that matter). Below the thermal limit it will run practically forever. Long after you would want that particular chip.

I overclock, solved the cooling problem, my dog comes in and licks the processor incessantly.

A 12% overclock is not going to make much difference to VP12. Even if this did translate to a 12% increase in program speed which I doubt, that's about a 30 min saving on a six hour render. A 10% CPU/GPU OC is possible using the mobo and GPU developers turbo speed software. Again, is this worth it for non-commercial NLE users?

It's not true that it's perfectly safe to OC. Even if you know what you're doing it's possible to cripple or fry components. Maybe it won't happen that day but what about next week or three months time? The temptation is to push each OC that little bit further to > 25%.

Rendering animation can take days so even a 10% speed increase would appear beneficial but I don't know any other animators who OC. Even switching the GPU to turbo can result in coming home to find the render has frozen due to the GPU overheating - even using a high end case with excellent air flow and cooling. None of the NLE, 3D or renderer developers recommend OCing. Ask yourself why before taking the risk.

No big deal. Do it or don't do it. I do it because I love tweaking things.

As for parts failing down the road, I guess that's possible, but I've not seen it yet. The part that I'm pushing is the CPU, and it lets me know in pretty certain terms whether it will run at the new speed. Run it at 15%, and it hangs within a few minutes; back off to 12% and it can render or run Prime95 (even harder of the CPU than rendering) all day and night.

By the way I should have mentioned in my earlier post that what CPUs need to run faster is a slightly increased CPU voltage (also adjustable on most motherboards), and it's that elevated voltage that is mostly responsible for the temperature rise. But again, manufacturers put in a pretty generous margin to start with, and a cheap aftermarket air cooler can make a radical difference in die temperature.

I would NEVER mess with system innards unless I had an image backup first. In a worst-case scenario, I want to be able to get back to where I was yesterday. I'd also want to install a temperature monitor so I can get feedback about my tweaks. But given that, there is little danger in fooling around a bit, in my view.

I have always wondered about the long term chip reliability from long runs at high temps, say 75C and up.

Has anyone done a reasonable study of this, or is it all anecdotal. I assume Intel has an idea, and maybe they have a document on this.

I look forward to burning through my current chip and having to buy the latest, newer, faster, better one. :) It's an upgrade path. :D

Here is another 3930K @5.0Ghz 1.52v monster :) A cool read.

http://www.overclockers.com/forums/showthread.php?t=711191

I can understand the interest in tweaking to get more out of something. When I build my PC's I do a lot of research about the latest components trying to get a combination of stability and speed. The technical side is interesting in and of itself. Especially if money is no object and if anything goes wrong it can be replaced.

Switching to turbo mode to get a 10% boost probably isn't going to stress the components although it may affect rendering but the incentive to OC is more about pushing things to their limits. That temptation is where I think the problem lies - especially in relation to finishing projects and meeting deadlines whether it's commercial or non-commercial work.