Solved CPU frequency fluctuation.

[Requimatic]

For about the past two weeks now, my CPU has started behaving strangely. I noticed it doing this while gaming, when what should have been perfect performance was kind of choppy. I alt-tabbed out of the game and peeked at SpeedFan, and noticed my temperature not rising above 50c while playing a game that normally shoots it up to around 75c.

Looking at CPU-Z, I see that for whatever reason the Core Voltage keeps randomly changing between 1.0 and 1.3v; this in turn is causing the multiplier to variate between 16-22. While typing this, it was staying at 1.235v, but dropped down to 1.0 when I ran a game just then to see if perhaps it had fixed itself like it seems to randomly do also. Also on that note, CPU-Z used to show my multiplier as "x23 (x23)", but now shows it as: "N (12-25)", N being whatever it's at currently based on the voltage, I assume. I don't know why it's suddenly listing a range of multipliers instead of just x23 as it used to. Also not sure why the range listed extends beyond what I thought was my CPU's maximum, x23. (I've never seen the multiplier go higher than 22 or 23 during all of this, or ever before for that matter.)

As far as I know, no Windows or BIOS settings are causing this, as nothing had been changed when it started happening (other than Windows updates). The only thing I've changed today was turning Turbo Boost off in the BIOS, but that had no effect. I was contemplating turning SpeedStep off also, as it kind of seems like the culprit right now. Almost as if it's underclocking when not under load, like it should, but then not clocking back up when under load.

So I'm not sure if it's the CPU itself starting to show signs of its age (and use), or if something else is causing this. What's everyone's opinions? I know my hardware is fairly old now, so I could understand if it was the CPU itself; I just hope it isn't. hah

 

[Layback Bear]

Your bios setting, things like speed step will make a computer do things like this.
It' designed to work that way.
When your computer has less demand the voltages will go down and the multiplier will change. When you computer is ask to do more work it will work in the opposite fashion.

Their are settings in the bios to change all this but it varies from motherboard to motherboard.
It's done for a reason. When you computer doesn't need the extra voltage and therefore the voltage is lowered the computer runs cooler. Under load the higher voltage will raise the temps so you might hear the fans speed up. That will depend on the fans and the temp.

If you don't have special cooling I would leave it set as it is. Changing such things without extra cooling will cause a computer to over heat.

I have this computer to run 4.6 all the time but I got lots of extra cooling.
2 pumps, 2 radiators and 14 fans.

To learn more it will take some time. My post is just touching what can and can not be done in bios.
Open you motherboard manual and go to the bios section and read. Then on ever setting Google it and you will find a explanation of what these thing in the bios do.

 

[Requimatic]

Yes, I understand how all of this works.

What I don't understand is why it isn't clocking up when the demand for increased performance is met. SpeedStep dictates that it should increase or decrease everything accordingly, but it isn't for whatever reason.

That's the mystery here.

 

[Layback Bear]

When you work you system hard like gaming and it does not step up it could be your temps are getting to high and the system is backing off to cool itself. It's called throttling. It's designed to do that to protect itself.

You do not show in your system specs what type of cooling you are using.

 

[Requimatic]

I monitor the temperatures of my CPU & GPU every day, and I know it's not them getting too hot. It's only sitting at 50c while idle, and will not get any hotter with any game running. This is according to both SpeedFan and the LED on the motherboard which constantly broadcasts the temperature.

Normally, yes, my temperatures rise. But while my processor is in this.. state.. it will not exceed 50c, which seems weirdly specific.

As for cooling, I use the stock fan/heatsink that it came with, and some Ceramique thermal paste. Air cooled case, with 5-6 fans all pumping air away from parts and towards the nearest exhaust. There are no obstructions or failing fans.

 

[Layback Bear]

Well I can't tell you what your temps should be because I have never run a stock cooler on anything.

The computer I'm on now has a i-960 and like a i-950 they do run hot. Using my system with just a couple forum open I'm at 38C.
Idle temps on these cpu really don't mean much. You need to know what the temps are under load using something like Prime95. I don't recommend using Prime 95 with stock cooling because the temps will clime very fast.

Intel cpu temp chips don't really read all that accurate at low temps. The temp chip reads accurate at high temp. Intel doesn't really care how cool the cpu gets but they do get concerned about how hot it gets.

I really can't think of anything except throttling at high temps causing your problem.

We have a lot of members so hang in there and see if someone drops by.

 

[Digerati]

And I never run aftermarket coolers on anything because we don't need them - except when extreme overclocking, or when "silent running" is desired - such as in a HTPC. Plus using an after market cooler on an Intel or AMD CPU that comes packaged with a OEM supplied cooler, voids the CPU warranty. This is NOT a concern for many - but everyone should understand this before deciding to use an aftermarket cooler.

I am currently at 36°C on this PC with an OEM cooler on my i7-3770 3.4GHz pushed to 4.10GHz. Ambient (room) temp at 70°F.

So as you can see, OEM coolers on today's CPUs are perfectly capable of providing sufficient cooling - even with mild to moderate overclocking - with proper case cooling. It is critical to remember it is the case's responsibility to provide a sufficient flow of cool air through the case. The CPU fan need only toss the CPU's heat into that flow.

That said, it does "appear" your system is toggling down to minimize heat. If your CPU (and system) temps are fine (and I don't start to worry unless they sit above 60°C for any length of time) then something else is telling your system to slow down.

I would look again in the BIOS Setup Menu to see if threshold settings have changed, for some reason. If your system is getting along in years, you might just replace the CMOS battery just for Ss and Gs. They are inexpensive so little is lost if no results. Just remember to unplug from the wall and touch bare metal of the case interior BEFORE reaching in to remove the battery. Do NOT touch the new battery with your base fingers as skin oils promote corrosion and attract dust. I put a clean sock over my hand. And while in there, ensure the innards are clean of heat trapping dust, and all your fans spin freely.

Once new battery is in, connect power and boot directly into the BIOS Setup Menu to reset date and time, verify drives are properly identified, then Save and Exit to boot normally. This may just fix your problem.

I am assuming a PC but sadly, you told us absolutely nothing about this computer. We don't even know if it is a PC or notebook! And that can be important because notebooks typically support some form of Eco-mode and/or battery conservation mode that toggles down speeds to extend battery run times. So look for a setting there too.

I know my hardware is fairly old now, so I could understand if it was the CPU itself

CPUs don't slow down with age. Generally, they work, or they don't work. If it is slowing down, it is being told to.

 

[Layback Bear]

Itaregid I see we don't agree on everything and that is okay.
We do agree that something is probable throttling the system.

I also like your idea of replacing the battery. It's cheep and can't hurt. It just might help. Setting the bios to default and giving the cpu cooler a very good cleaning and seeing how the system works at that point.

Like I said I have never used a stock cooler so I really only have what others have posted about a stock cooler.

How good a stock cooler on your newer system compared to the older i7-950 which run very hot even when they run proper could be the difference.

Intel has drastically lowered the heat output from your i7-3770 compared to the older I7-950 and improved the cooling.
Requimatic i7-950 is like my i7-960, little toasters.
I still highly recommend better cooling.
Case air flow like you mentioned is a must. All cooling starts from there and ambient temperatures.

Intel also has it's own step up from stock cooling on its 1155 socket for those that want a bit more.

http://www.newegg.com/Product/Product.aspx?Item=N82E16835203017&nm_mc=KNC-GoogleAdwords-PC&cm_mmc=KNC-GoogleAdwords-PC-_-pla-_-Water+%2f+Liquid+Cooling-_-N82E16835203017&gclid=COXH1ub798ECFZSHaQoduWIAeg

 

[Digerati]

Like I said I have never used a stock cooler so I really only have what others have posted about a stock cooler.

This is actually the problem. Most of what "others have posted" is just repeating what they have heard. You for example, appear to automatically replace your stock coolers because of what you heard - without even seeing if true. And then pass along the same "rumors" when the fact is, the vast majority of users don't need aftermarket cooling. So they end up wasting money on an aftermarket cooler when it could have been better used on more RAM, faster CPU, better graphics, or nicer monitor - without violating their CPU warranty!

If OEM coolers were so bad, there would be 100s of millions of overheating computers out there every day. Not happening!

The facts are, years ago, neither AMD or Intel expected users to stress (overclock) their CPUs so the supplied OEM coolers were inadequate for those users - who tend to be very vocal users. Also years ago, fan noise was not really considered a problem, or it was just considered a fact of life we must live with. Companies like Zalman capitalized on those facts and came out with monster and quiet coolers and overclockers went wild with enthusiasm for the Zalmans, complaining how horrible the OEMs were - when in fact they were not horrible, just not as good or quiet.

The problem is, those users are still saying the same thing, and the same complaints are being re-posted and re-posted over and over again when both Intel and AMD have greatly improved their coolers.

Another problem is many users assume 30°C is better than 40°C, or even 50°C. It is not. It is not better for performance, and it is not better for CPU longevity. CPUs are designed to operate within a safe temperature "range". As long as they remain safely within that range, their performance and life expectancy is the same. All 30°C gets you is bragging rights. But sadly, some think otherwise - even though they have no formal electronics training.

I still highly recommend better cooling.

And are you a better electronics and/or thermal engineer than the folks at Intel or AMD? I am all for good cooling but again, it is the case that is responsible for it. If the CPU is overheating with the stock OEM fan at default clock speeds, the USER has failed to setup case cooling properly! Or something else is wrong - like the cooler mounting mechanism is not fastened securely, the fan is weighed down with dust, or the computer was bounced off the floor breaking the TIM bond, or the HSF assembly was twisted too hard to see if still tight, breaking the bond.

Remember ONLY Intel and AMD warranty their coolers with the CPUs. And only Intel and AMD will replace your CPU should a failed cooler destroy it (regardless how remote that is). No aftermarket cooler maker will cover damage to the CPU should their cooler fail.

Intel has drastically lowered the heat output from your i7-3770 compared to the older I7-950

Not true - but irrelevant anyway! You again are assuming the folks at Intel don't know how to cool their hotter CPUs. That's not true. And not all Intel CPUs use the same cooler either - so apples and oranges.

As for not true, note the Tcase temps for the i7-950 is just a mere .5° higher than the i7-3770 with 67.9°C compared to 67.4°C.

 

[Layback Bear]

It is not a false statement when one suggest to run your system cooler is better. Now is one willing to pay for running cooler. I prefer to run cooler stock or over clocked. One has to make those choices themselves.

I never stated that is wrong to run a stock cooler. You are putting word in my post along with many others post by others.

We have agreed that the system is probable throttling it's self. The number on reason that happens is high temp. I don't care what cooling is being used their is a good chance it's not doing its job.

Whether it's the fan, fan mounting, thermal past, case cooling or a combination of them I don't know because I'm not there.
I'm like everybody else but one the OP; is behind their own keyboard.
I fully agree their are millions of people running stock coolers. At this point we have one in a million member that is probable running hot.

So why don't we just stick with helping this member and discuss the plus and minus of stock and updated cooling for another day.

I really don't care to argue the point.
I understand your opinion and I think you understand my opinion.

 

[Digerati]

It is not a false statement when one suggest to run your system cooler is better.

It depends on where you are starting from.

80°C to 50°C is better. But 50 to 40 is not better! Neither is 40 to 30.

It does not improve performance, stability, or life expectancy of the hardware.

I never stated that is wrong to run a stock cooler. You are putting word in my post along with many others post by others.

What? I never said you did! I said most users don't need aftermarket coolers. So who is putting words in others mouths?

I am not here to argue either. I am just here to ensure posters have the true "facts" - not rumors, or unsubstantiated opinions that don't apply to today's technologies.

Have a good day.

 

[Requimatic]

I didn't think this would turn in to some kind of debate, but..

Disabling the Intel SpeedStep in the BIOS solved this issue. For whatever reason it wasn't initializing like it should have, and I haven't had any of the symptoms since doing so.

Sorry for the necro, but I've been busy with other things.. like trying to figure out why I'm suffering from BSODs now (almost positive it's hardware related at this point), and why my thread about it here (in the proper section, no less) was deleted.

Going to mark this one as solved.

 

[Layback Bear]

Requimatic I do apologize for your thread getting side tracked.
I'm don't know how turning off Intel SpeedStep solved your problem; but it is very interesting.

Not knowing your expertise I was concerned about proper cooling for your system because Bloomfield's do run hot normally.
You are happy with the cooling I'm happy with your cooling.

Happy computing.