• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Incorrect cpu freq

A

abax2000

New Member
Joined
Mar 11, 2012
Messages
5 (0.00/day)
When not on default multiplier, the multiplier is reported wrongly, and concequently freq is wrong also.
That is for a Q6700.
 
OK, how are you checking cpu frequency? What is reporting?
 
Give us more information please, your post is rather vague. Maybe what you're trying to do, what you're using for monitoring, and what you're build specs are (might want to add that to your profile.)
 
abax2000 said:
CPU-Z reports 6x320=1921
RealTemp reports 9.3x320=2976
Click to expand...

Do you have SpeedStep enabled? If you go into Power Options in Windows and set it to "Performance" do you notice it cranks up to 2.976ghz and when you set it back to balanced it goes to 1.921ghz? It is very possible that your computer is throttling your CPU clock because that is what Intel does to save power when the CPU is idling. A good example is how my 3820 is at 1.5ghz when it is idling and when I put it under load it goes up to 4.75ghz.
 
Yep, speedstep. Put some load on the CPU, use Prime95 or something, and the speed in CPU-z should be correct.
 
newtekie1 said:
Yep, speedstep. Put some load on the CPU, use Prime95 or something, and the speed in CPU-z should be correct.
Click to expand...

Do I hear an echo? :p
 
EchoEchoEchoEchoEchoEchoEcho ;)
 
abax2000 said:
CPU-Z reports 6x320=1921
RealTemp reports 9.3x320=2976
Click to expand...
Maybe it's about the quote below from this post?
CPU-Z said:
Of course I admit that CPU-Z is not accurate anymore at idle on latest Intel generations, that is why TMonitor was developped.
Click to expand...
 
1. SpeedStep is in use, so multi fluctuates 6-10 (quite normal)
2. CPU-Z is reporting just fine.
3. I was expecting that RealTemp could also report correctly.
 
All I can say is that RealTemp is reporting correctly. :)
 
You are correct. Your multiplier is being reported incorrectly when your CPU is lightly loaded but the problem isn't with RealTemp. CPU-Z has been designed for MHz validation purposes and the programmer has chosen consistency over accuracy when lightly loaded. The multiplier of a CPU can be changing 100 times a second when lightly loaded which most users are not interested in seeing. It looks a lot nicer to report a steady 6 multiplier even if the real multiplier is bouncing up and down all over the place.

RealTemp does things differently. It uses high performance timers within the CPU and follows the monitoring method recommended by the manufacturer, Intel. RealTemp tells it exactly like it is. The multiplier it reports is an extremely accurate average of the multipliers your CPU has used during the last second. If RealTemp is not reporting a nice and steady 6.0 multiplier then I can guarantee you that your multiplier has not been sitting nice and steady at 6.0. When a CPU is lightly loaded, CPU-Z can be completely misleading.

You didn't bother mentioning what operating system you are using. If you are using Windows Vista or Windows 7, the Minimum processor state value controls what multiplier your CPU will idle at. Whether you have C1E checked off in RealTemp also controls what multiplier your CPU will be using when idle.

poweroptionsy.png


Try playing around with the Minimum processor state. If you want your CPU to idle down more when lightly loaded then try using a value of 5% instead of 100%.

If you are still using XP then this works a little differently. If you want your CPU to truly idle down then you will have to go into the Power Options - Power Schemes tab and select Portable/Laptop. If you don't want it to idle down then you need to use the Home/Office Desk setting.

2. CPU-Z is reporting just fine.
Click to expand...

If you want to take charge of your CPU and learn a little more how these things really work then check out my other program called,

ThrottleStop 4.00
http://www.techpowerup.com/downloads/2090/ThrottleStop_4.00.html

If you start playing around with ThrottleStop you will soon realize that CPU-Z may be the most used monitoring application in the world but being popular has nothing to do with being accurate, especially when a CPU is lightly loaded.

When a Core 2 Quad is truly using the 6 multiplier, RealTemp has no problem at all reporting that.

6multiplier.png


And when the CPU is not idle, CPU-Z continues to report the same thing.

multiplierlie.png


One of these programs isn't being 100% honest. :)
 
Last edited:
Same thing when you put 100% load on your CPU?
 
It's Windows7 (32bit).
Now, that's an enlightening answer. Thank you.

Interesting question by Aquinus.
 
abax2000 said:
It's Windows7 (32bit).
Now, that's an enlightening answer. Thank you.

Interesting question by Aquinus.
Click to expand...

SpeedStep could just still be active, regardless of power options. I've never found CPU-Z to be inaccurate on any of my machines I've owned in the past, including a C2D system. I just noticed how both sets of pictures both have <10% load on the CPU which wouldn't normally kick off a higher power state.
 
CPU-Z over simplifies things. It seems to be designed so that it reports the default multiplier or the lowest multiplier and not much in between. At full load both programs agree but when the CPU is not fully loaded, depending on how you have your C States and Minimum processor state set up, the multiplier can be bouncing up and down very rapidly. CPU-Z chooses to ignore this while RealTemp reports a very accurate average of what the multiplier is really doing. ThrottleStop can report the multiplier of each core or thread in the case of hyper threaded Core i processors and does a better job of telling the whole story.

I've never found CPU-Z to be inaccurate on any of my machines I've owned in the past, including a C2D system.
Click to expand...

Everyone assumes that CPU-Z is always correct because they have nothing to compare it too.
 
unclewebb said:
CPU-Z over simplifies things. It seems to be designed so that it reports the default multiplier or the lowest multiplier and not much in between. At full load both programs agree but when the CPU is not fully loaded, depending on how you have your C States and Minimum processor state set up, the multiplier can be bouncing up and down very rapidly. CPU-Z chooses to ignore this while RealTemp reports a very accurate average of what the multiplier is really doing. ThrottleStop can report the multiplier of each core or thread in the case of hyper threaded Core i processors and does a better job of telling the whole story.



Everyone assumes that CPU-Z is always correct because they have nothing to compare it too.
Click to expand...

I trust CPU-Z when i compare it to my BIOS, and I don't call a Core 2 Duo an i-series Intel chip. What you're saying is not correct and I found that even with my old Core 2 Duo and Phenom II 940, it properly reports clock speeds, bus speeds, and multipliers, in both lower power and performance conditions.

I'm not saying that it isn't working properly, it very well might not be. I'm just saying that I personally haven't noticed inconsistances with custom built machines I've used it on. Now checking it under load is a perfectly valid question and doesn't take that long to test. So instead of barking at me for asking a question and throwing in my 6-sense, maybe should should check it, verify that it doesn't change and then say that. I'm just trying to help, and honestly, wouldn't your opinion be sort of biased since you're the "Author of RealTemp"?

Don't bash CPU-Z and say it doesn't do the same things just because it doesn't work with one platform.
 
Originally Posted by cpuz
Of course I admit that CPU-Z is not accurate anymore at idle on latest Intel generations, that is why TMonitor was developped.
Click to expand...

That is a direct quote from the programmer of CPU-Z. No one is barking or bashing anyone. CPU-Z is a great utility but users need to understand that it has been designed as a MHz validation tool so what it shows you when lightly loaded may not be an accurate representation of what multiplier your CPU is using.

At full load, CPU-Z rounds things off to the nearest whole number so it can also be one step behind RealTemp such as in this example posted by rge at XtremeSystems.

rtthrottles238pst.jpg


abax2000 had some valid questions and I wanted to try and convince him that he can trust what RealTemp is telling him, whether he has a Core 2 or Core i processor. Hopefully he can post some of his own results after he plays with the Minimum processor state setting. I will try to post another Core 2 comparison later today when I have more time.
 
Yeah, CPU-z always shows that my Multiplier is 18, because I set the multiplier at 18x in BIOS while having turned EIST and C1E on. Realtemp shows my real multiplier.

6DG9S.png
 
Just for comparison, here is what it does on my 2600K. CPUz shows it go to idle multi but realtemp fluctuates from 40-43 multi randomly.

Realtemperror.png
 
newtekie1 said:
Yep, speedstep. Put some load on the CPU, use Prime95 or something, and the speed in CPU-z should be correct.
Click to expand...

I wouldn't recommend, using any burn-in test... Unless, you want to wear out the capacitors on your motherboard & increase the amperage, that the entire system uses...

The CPU, GC & MB, will use more Wattage etc.. Unreversible & Shortens the life, of all your components...

SpeedStep, will down clock your voltages and multiplier whenever your computer is idle (low load) or not in use.
 
Last edited:
To set up ThrottleStop when testing you need to check off the Set Multiplier box and set that to the appropriate multiplier and you need to click on the Turn On button. It might help solve some of the multiplier confusion.

NanoTechSoldier: If a computer can not run Prime95 for 30 seconds then there is something seriously wrong with it. If it shortens the life during a 30 second stress test then you need to upgrade your computer anyway.
 
unclewebb said:
To set up ThrottleStop when testing you need to check off the Set Multiplier box and set that to the appropriate multiplier and you need to click on the Turn On button. It might help solve some of the multiplier confusion.

NanoTechSoldier: If a computer can not run Prime95 for 30 seconds then there is something seriously wrong with it. If it shortens the life during a 30 second stress test then you need to upgrade your computer anyway.
Click to expand...

Well, Capacitors, shorten in life, within a split second of full load & Can't control the Power on the MB As well...

So much you know about electrotechnology..

It Clutters your registry too.. Try & clean that up.. I Bet You'll Say; "I'll Use A RegClean" LOL..

The BIOS Chip, Also Gets A Beating Too within 30 seconds.. Using SpeedStep etc..

The minimum Burn-In test, normally is for 15mins-3hours etc..

30 Seconds Tests Nothing Of The CPU & Is Actually bad for It.. If The Instruction Sets, Aren't Finished Processing & You Quit Program & Full Load etc..
 
You must log in or register to reply here.
Back
Top