Personal computing discussed

Moderators: renee, Starfalcon

 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

A quick question about overclocking.

Thu Mar 27, 2014 4:00 am

I'm a noob at overclocking so bare with me =P Why is it that benchmarks such as RE5/6, Unigine Valley Benchmark, PSO2, etc etc show that my CPU (4770K) is only clocked at 3.5ghz?

In my bios/uefi settings it's OC to 4.6ghz.

Also when I look at my system settings (in windows) it also has my CPU listed at 3.5ghz.

Helpful comments will be appreciated ^^
 
f0d
Gerbil XP
Posts: 422
Joined: Sun Jan 23, 2011 3:07 pm
Location: austrALIEN

Re: A quick question about overclocking.

Thu Mar 27, 2014 4:30 am

because most programs just go by what model your processor is and the speed of the standard one instead of actually reading the speed its running at

mine does the same in the same programs
Image
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Thu Mar 27, 2014 7:03 am

you can use CPU-Z to monitor the clock speed in real time: http://www.cpuid.com/softwares/cpu-z/ve ... story.html
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Thu Mar 27, 2014 11:23 am

Thanks for the response :)

Does this look ok?

http://imgur.com/91JQbaa
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Thu Mar 27, 2014 11:55 am

Well it says 4598MHz doesn't it? That sure sounds like an 4.6GHz overclock to me :)

BTW, you really should stress test / benchmark your overclock as well, using utilities like Prime95, Intel Burn Test and OCCT. There are probably a few more that are relevant but it's been a while since I've messed with my OC settings. Anyway, key is that you see a performance increase that corresponds with your overclock, and of course that the benchmarks/stress tests don't produce any errors. Why the second part is important should be obvious, but the first part is because you might be able to overclock your CPU to 4.6GHz even if your system won't actually be able to run it at that speed for an extended time. I can overclock my sandy bridge to 4.5GHz and beyond easily, but my motherboard doesn't agree with that (should've sprung for a better one..) and will start throttling my CPU, probably to save the VRMs from overheating. This throttling will be very obvious in decreased performance in Intel Burn Test for example, and of course CPU-Z will also show that the clock is not what you would expect. Games and other graphics benchmarks typically don't stress the CPU that hard, so throttling or errors might not show up until the worst of times.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 5:55 am

After an hour of prime 95 my pc blue screened :(

What should I do. Obviously it's not a stable OC. The highest temp was 80c on one core.

Should I go for a lower clock like 4.4ghz and a lower voltage? So for example 4.4ghz with 1.2volt?
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 8:06 am

I suggest you read up on the basics of overclocking. I don't mean to come off as condescending, but from your question it seems that this is the first time you're trying to overclock anything, so I recommend you google for overclocking guides for beginners and try and soak up as much knowledge as possible!

To more directly answer your question though, I would reset the clocks and voltage to stock values for now. Then, increase the multiplier one step at a time (you can probably start off at 4.0GHz) and test with Prime95 that it's sort of stable. If it is, go ahead and try another step! If it crashes, well good, you found a limit, back off one step and test that speed with everything you can throw at it. Just by doing that and nothing else, you can probably run your 4770K at something around 4.2 to 4.4 GHz. After you found that wall, if you want to go further you will have to increase the core voltage and maybe tweak some other settings, but I advise you to not even bother until you've read up on what that means, what kind of effects you can expect, what the risks are and that you understand the risks.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: A quick question about overclocking.

Sat Mar 29, 2014 12:05 pm

OCing is a crapshoot. It is not a matter of "this CPU should be capable of OCing to x MHz on y voltage, set it and forget it". You will need to try different combinations of clock, voltage, and probably a better coolong solution to unlock the full potential of your CPU, if that is your intent.
Nostalgia isn't what it used to be.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 1:52 pm

No worries :) It is the first time I've overclocked a CPU so that's why I'm asking all these noob questions.

I used a couple of videos on youtube that I used to do this lol.

https://www.youtube.com/watch?v=0CHs5_TdpXE

https://www.youtube.com/watch?v=Z7zPu9255ZI

I basically followed their rules. They say that I should do a quick and "dirty" OC of 4.6ghz at 1.2volt and if the system boots up then you can go from there. This is my brand new lil over a month old 4770K so of course I don't want to mess it up.

That said I would still like atleast a lil OC! =P I already did what you said and reset my BIO's settings back to stock.
 
LASR
Gerbil First Class
Posts: 147
Joined: Fri Jan 10, 2014 9:35 pm

Re: A quick question about overclocking.

Sat Mar 29, 2014 2:29 pm

There are no rules in overclocking. Every CPU is different. it is very much a game of chance.
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 3:07 pm

diasflacog wrote:
reset my BIO's settings back to stock.

Great, now work your way up again! As the others already said, it's a crapshoot, so you won't know beforehand how far your chip will go. The only way to find out is to methodically work your way to its limits, by changing one thing then testing, then changing one thing again. Don't try and change more than one thing at a time because if it proves unstable, you won't know which change was the culprit! Once you've found settings that you're happy with and that seem stable, I suggest you lower the multiplier by one step. That way, you can be pretty sure that it will be stable for every day usage.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 3:27 pm

Thank you for all the input. You want me to take it one step at a time by going 4.0ghz but what voltage should I be using?

From my understanding once you OC you have to raise the voltage aswell correct?

lastly, what would you consider 'stable" I looked around forums and such and many people have different opinions of stable. Some say that 24 hours of prime without errors and bsod is stable others say 8 hours is stable some even say not to use prime...lol
 
Captain Ned
Global Moderator
Posts: 28704
Joined: Wed Jan 16, 2002 7:00 pm
Location: Vermont, USA

Re: A quick question about overclocking.

Sat Mar 29, 2014 3:53 pm

diasflacog wrote:
Thank you for all the input. You want me to take it one step at a time by going 4.0ghz but what voltage should I be using?

From my understanding once you OC you have to raise the voltage aswell correct?

lastly, what would you consider 'stable" I looked around forums and such and many people have different opinions of stable. Some say that 24 hours of prime without errors and bsod is stable others say 8 hours is stable some even say not to use prime...lol

Keep increasing speed at stock voltage until you get issues. There's usually plenty of headroom at stock voltage and there's no point in playing with voltage until you find the limits at stock. Then bump voltage by a bit and try to increase speed some more. Lather, rinse, and repeat until you find the right combination of speed & voltage. Tuning an OC is not a 5-minute process. To do it right could easily take a week or more of patience followed by a few months of careful monitoring and grepping Event Logs.

Personally, I'll stop when I max out at stock voltage, but that's just me. I tend to run CPUs for 5-7 years so I'll avoid the chances of electro-migration from increased voltage. Those who swap every 2-3 years can take on a bit more risk.

Stable is 24 hours of Prime and a full MemTest86 with no errors at all.
What we have today is way too much pluribus and not enough unum.
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 4:18 pm

diasflacog wrote:
From my understanding once you OC you have to raise the voltage aswell correct?

No, not necessarily. You can just increase the clocks of GPUs and CPUs without touching the voltage at all, and most of the time you will be able to get at least 10% without changing anything other than the clocks or multipliers. Once you've found how high the CPU or GPU will go at stock voltage, you could increase the core voltage (or any number of other settings) and that might increase the maximum stable speed at which the CPU/GPU will run. There is no guarantee though, except that a higher voltage will exponentially increase power usage and heat, and it can lead to premature death of the chip if you push your luck. What I mean with exponentially increased power and heat usage is that a 10% overclock at stock voltage will lead to approximately 10% more power usage (often a bit more), but a 10% increase in core voltage will lead to 21% increased power usage (1.1 * 1.1 = 1.21). So if you increase your clocks by 10% AND increase the core voltage by 10%, you will have roughly 33% more power usage! And all that extra power is being turned into heat on a very small surface that your cooler is trying to keep cool.
To give you an idea, I have my i5-2500K overclocked to 4.3GHz, at stock voltage. The chip can do 4.6GHz or more I think, but to get there I have to increase the voltage, at which point my motherboard starts throttling the CPU to keep the VRMs from blowing up.

diasflacog wrote:
lastly, what would you consider 'stable" I looked around forums and such and many people have different opinions of stable. Some say that 24 hours of prime without errors and bsod is stable others say 8 hours is stable some even say not to use prime...lol

I would consider a configuration stable if it reliably does what I need it to do. That means that a gaming PC that might crash once in a blue moon is stable for me, but a mission critical server that does the same would be unstable, because it cannot be relied on. For the purposes of overclocking, there is no 100% way to be sure that it is absolutely stable. The reason is that Intel and AMD leave a good bit of headroom because they can't be sure either. All they know is that if you keep well away from a chip's limits, it will probably be pretty stable. Overclocking is the art of getting closer to those limits and thereby extracting extra performance, with the caveat that you give up the big safety margin that the manufacturer left.

So, how would I test for stability? For the purposes of roughly finding the chip's limits, 10 minutes or so of Prime95 will give you a good idea. Then if I think I found a stable setup, I would test it with Prime95 for a couple of hours, and then with Intel Burn Test as well. Those two programs do a good job of putting a lot of stress on the CPU whilst verifying that they're spitting out the expected results. I would even let one of those running over night. Then, if that has worked out alright, I would either accept that as the stable configuration for now, or I would back off 1 step. Backing off 1 step would give you back a little of that big safety margin that the manufacturers left, so you can be (again) pretty sure that it will really actually be stable. If you don't back off at all, don't be surprised if you find that a particular game or application crashes and burns in a most unpredictable way.

edit: oh yeah, as Captain Ned said, if Prime95 or similar programs give any errors at all, you consider that as a failed test. Your overclocked PC should behave exactly as stock, only a bit faster. Any deviation from that is a sign of pushing it too far.
 
Melvar
Gerbil XP
Posts: 381
Joined: Tue May 14, 2013 11:18 pm
Location: Portland, OR

Re: A quick question about overclocking.

Sat Mar 29, 2014 5:50 pm

Pardon me for butting in, but I always see people saying to test with Prime95, but they never say which one of the torture tests they run. Which test do you folks use?
 
Voldenuit
Minister of Gerbil Affairs
Posts: 2888
Joined: Sat Sep 03, 2005 11:10 pm

Re: A quick question about overclocking.

Sat Mar 29, 2014 8:17 pm

Melvar wrote:
Pardon me for butting in, but I always see people saying to test with Prime95, but they never say which one of the torture tests they run. Which test do you folks use?


Disclaimer: I am not an authority in overclocking, but here's my $0.02.

If I'm testing a CPU overclock (or undervolt), I usually run small FFTs for ~1h to test CPU stability. I then run in-place FFTs to see what kind of temperature delta I'm getting with the last stable set of clockspeed/voltage settings, and decide if I'm happy with the numbers or want to dial things back to reduce heat output/power consumption.

If I'm overclocking RAM or playing with timings, I use blend or a custom FFT size of min: 2048 and max: 6144 (if I have 8 GB of RAM) for RAM stability. Be careful not to exceed your physical memory size, or a lot of the FFTs will get paged out, then you won't be testing the physical RAM, and you'll be doing nasty things to your hard disk or ssd page file. (Memtest86 is a better test for RAM than Prime95, but it's also a lot less convenient, so I use p95 for quick and dirty tests)

To the OP: 4.6 GHz on air is a pretty high target for haswell. Most people are maxing out at 4.5 GHz. Sandy (and to a much lesser extent, Ivy) are still better at high overclocks on air; perhaps the upcoming batch of overclocking-tweaked desktop chips might fix this, but they're not here yet.
Wind, Sand and Stars.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Sat Mar 29, 2014 9:58 pm

Thanks for all the input guys :)

My PC has a Hyper Evo 212 with a very thin layer of artic silver 5. The only reason I put 4.6 was because of the videos I posted. I never overclocked before but even I knew that 4.6 was a bit too much.

My ideal OC would be 4.3-4.4 no more than that.

I haven't messed with it at all today, but I will now that I have time. I will start by going to 4.0 without adding any volts.

I'm curious do multiple blue screens affect my cpu in any negative ways? Meaning does this kill my cpu little by little? I ask because if I go for 4.4 (or 4.3 if I can get 4.4 stable) im going to have to do many test which will cause blue screens if unstable.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: A quick question about overclocking.

Sat Mar 29, 2014 10:09 pm

The blue screens do not directly harm the CPU, though they may be a sign that the CPU is in some distress. As long as your voltages and temperatures are not seriously out of whack the CPU should still outlive its usefulness.

Your OS and files, on the other hand, may not survive if you're blue screening frequently. Hard drive corruption is a distinct possibility.
Nostalgia isn't what it used to be.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Re: A quick question about overclocking.

Sun Mar 30, 2014 11:49 am

"Stability" here is more about your tolerance level and degree of paranoia. There is always a next test to run, and/or another few passes to test for. And then there is the next different testing to try, and then all the tests that follow. Rinse, and repeat. For me, I would at least test all the components once if they are at stock, and further testing the overclocked components some more, and with a mix of stress tests. You can use the stress test tools thread as a reference.

You can be as anal as the next guy. Look at what this guy does. It will take weeks to test all of that thoroughly. It is up to you at some point and say "this is enough" and declare victory. And then you have to always keep an eye on things. This is the price to pay for overclocking. In this day and age of "mostly good enough" computing, a lot of the older folks have given up on this pursuit due to the time and effort involved and the relatively minor gain. Sometimes it may still be worth it. Only you can decide.

[Mod: Moving to OC forum.]
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: A quick question about overclocking.

Sun Mar 30, 2014 11:55 am

I'm quite anal about stability. If the system *ever* crashes unexpectedly, and it can't plausibly be blamed on a known hardware, OS, or driver issue, then to me the system is not stable and needs to be fixed. The *only* times you should ever be forced to reboot are to install OS updates or add/remove hardware.

I realize that this is probably an extreme viewpoint, especially among avid overclockers...
Nostalgia isn't what it used to be.
 
f0d
Gerbil XP
Posts: 422
Joined: Sun Jan 23, 2011 3:07 pm
Location: austrALIEN

Re: A quick question about overclocking.

Sun Mar 30, 2014 12:22 pm

just brew it! wrote:
I'm quite anal about stability. If the system *ever* crashes unexpectedly, and it can't plausibly be blamed on a known hardware, OS, or driver issue, then to me the system is not stable and needs to be fixed. The *only* times you should ever be forced to reboot are to install OS updates or add/remove hardware.

I realize that this is probably an extreme viewpoint, especially among avid overclockers...


im a pretty "avid" overclocker (been doing it since i swapped crystals on a 286 from 10 >12mhz and i overclock everything - htpc / main pc / gpu / cpu and even router and phone) and i 100% agree if the system is not able to pass some VERY extreme stress tests then to me its a useless overclock and needs adjusting

when i stress test i do many different things like run linx (a varient of intel burntest) at 4 gb 8 gb and 12gb memory sizes for 4-6 hours each as well as prime95 with various setting for hours and as well as running combinations of both for hours with other programs

sometimes i run linx and prime 95 as well as unigine valley and 3d mark and encode a test video using handbrake all at once for hours at the same time and if i see a crash or bluescreen during any of the tests then i adjust my settings and retest

when i find a stable overclock i add (usually) .015 more volts to ensure its stability so if i need 1.2v to pass a range of stress tests i usually will finalize it by making the voltage 1.215

i tested my 5ghz overclock on my 3930k for 3 days straight doing nothing but various stress tests and running the cpu at 100% for the whole time


JBI - your viewpoint isnt an extreme viewpoint, its the RIGHT one
Image
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: A quick question about overclocking.

Sun Mar 30, 2014 12:33 pm

f0d wrote:
JBI - your viewpoint isnt an extreme viewpoint, its the RIGHT one

Just because it is "right" doesn't mean a majority of people share it. :wink:
Nostalgia isn't what it used to be.
 
Captain Ned
Global Moderator
Posts: 28704
Joined: Wed Jan 16, 2002 7:00 pm
Location: Vermont, USA

Re: A quick question about overclocking.

Sun Mar 30, 2014 12:36 pm

just brew it! wrote:
f0d wrote:
JBI - your viewpoint isnt an extreme viewpoint, its the RIGHT one

Just because it is "right" doesn't mean a majority of people share it. :wink:

Being "right" is usually associated with being lonely, especially when your new "right" is orthogonal to accepted reality.
What we have today is way too much pluribus and not enough unum.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Sun Mar 30, 2014 1:26 pm

I'm starting to see what you guys mean by all of this. It's very tedious work. I don't think I will be able to stress test my machine as long as you guys. Going 24 hours without using it is impossible for me lol.

I'm going to settle for somewhere in between. As long as I can run prime95 for like 8 hours straight I'm calling it a victory. After that I will test game benchmarks like the Valley one and other normal game benchmarks and lastly I will play games for hours. If I don't get any blue screens while playing Tomb raider at ultra settings then I'm calling it a win.

I just don't have the time or patience to stress test like you guys.

All that aside, I have one more question.

Right now my XMP values are set to "disabled" my Ai Overclock Tuner is set to manual mode. It was recommended that I go back to defaults and just tweak the Core Ratio Limit and then up the volts as I needed.

Once I find that 'sweet" spot for the overclock should I leave the xmp the way it is? or do I change it to profile 1? The bios already reads my RAM's proper speed of 1600 so I don't know if I should turn this on or leave it as is.
 
Firestarter
Gerbil Elite
Posts: 773
Joined: Sun Apr 25, 2004 11:12 am

Re: A quick question about overclocking.

Sun Mar 30, 2014 2:04 pm

Yeah diasflacog, I don't have the patience for that anymore either, that's why I try and find an overclock that works for a good couple of hours of testing, then just back off one step for good measure.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Mon Mar 31, 2014 3:50 am

I managed some decent settings now. 4.4ghz with 1.18v seems to be "stable" for now. I've ran prime95 for 4 hours and I also ran Real bench 2.1 for 30 mins none had any errors or blue screens. Now I know that for you guys who like to run these things for 24 hours this doesn't prove anything, but ima rock these settings in a normal gaming environment.

I also used Real Temp to monitor my temps and the hottest core was at 81c in a somewhat hot room. I live in CA so during the day it's kinda hot here.

My next question is, should I switch to adaptive voltage or leave it at manual? People say that I should leave it at manual for full stability. What do you think?
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: A quick question about overclocking.

Mon Mar 31, 2014 4:15 am

Not sure what your motherboard vendor means by "adaptive voltage". If it is a form of Load Line Calibration (LLC), it may actually be of some help. What are the choices for that option?
Nostalgia isn't what it used to be.
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Mon Mar 31, 2014 4:52 am

Auto, Offset, Manual, and Adaptive Mode.

For now I'm keeping it at Manual since people say that's the best for stability. However I just wanted to post it here aswell.

Oh btw Asus makes my motherboard. It's a Asus Z87-A.
Last edited by diasflacog on Mon Mar 31, 2014 5:26 am, edited 1 time in total.
 
Jon1984
Gerbil XP
Posts: 336
Joined: Wed Mar 23, 2011 3:07 pm
Location: Portugal

Re: A quick question about overclocking.

Mon Mar 31, 2014 4:54 am

You should try to use offset voltage so your CPU isn't always running in full voltage even at idle.
i5 3570K - Nepton 280L - P8Z77-V - 4x4Gb - 290X Gaming 4Gb - 2x F90 - Spinpoint F3 1Tb - S340 - CS750M
MX239H AH-IPS - Ryos MK - Kone Pure Military - HyperX Cloud - Win 10
Xiaomi Mi A2
 
diasflacog
Gerbil Team Leader
Topic Author
Posts: 239
Joined: Tue Dec 18, 2007 3:58 am

Re: A quick question about overclocking.

Mon Mar 31, 2014 3:01 pm

What happens if I leave it at manual? I know I get the same voltage 24/7 with this option. What I'm asking is are there any harmful side effects of this? My guess is heat, but as long as I keep my temps down I should be ok?

I haven't gone over 65c while gaming.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On