Hello. How do I verify my CPU is actually downclocking to multiplier of 6 when idle?
I thought TMonitor was showing correct clock but it doesn't actually change. CPU load of 1 to 3% yet multiplier well above 6 - that can't be right.
And what does the Minimum Processor State at 0 cause? Mine is 5.
Thanks a lot!
P.S.: Iron Man just noticed TMonitor does not scale...
-
-
Something has to be wrong. ThrottleStop shows like this below and never changes either monitoring or on. Changed multiplier, C1E, EIST, Power Saver and both modulation randomly and to no avail. Perhaps I am doing something wrong. Would you please guide me through the steps to make my CPU works at the lowest multiplier? Shouldn't it be 6 by the way? Lowest shown is 7.
Notebook is a Dell Inspiron 1564. Thanks.
-
Make sure your power paln is on power saver or balanced, if its on high performance all these programs will give funny readings
-
@OP: Where did you read that the lowest possible multi was 6? That would be 798Mhz... petty low. I don't know, but TS reports everything properly for me...
Why do you want to run the CPU at the lowest freq? Battery I guess? I'm actually not sure whether you can fix the multi at a certain value with your CPU... (Try set multi and you'll see whether it works or not, EIST must be enabled)
When you wan't to safe battery I recommend adjusting the power plan, the CPU will handle the rest pretty well imo. -
Download HWiNFO32 and press the "Summary" button.
-
Maybe you heard of Inspiron 1564 fan issues? Initially I worried that mainboard could not regulate CPU clock especially downclocking and I'm still not sure it does well. I am confident it cools down effectively but maybe unnecessarily as clock is still too high.
I've got a Latitude D500 originally with a Banias 1.3GHz then upgraded to a Dothan 1.8GHz - which was very nice. These are the only computers using SpeedStep I have experience with. CPU-Z reports very steady state at the lowest multiplier when idle and few load.
Inspiron is constantly modulating under load of 0 to 3 percent.
High Performance and Balanced power profiles with Minimum Processor State of 0 to 5 already tested. There is no apparent conflict like between mainboard BIOS and SpeedStep settings and Windows power profiles.
Just one thing noticed though. ThrottleStop shows C1E unchecked while RealTemp has it enabled. C1E is not present in the mainboard BIOS.
Someone was very convincing about 6 being the lowest multiplier on intel processors and I researched that a lot, still don't know what it is for the i3.
Open Hardware Monitor always reports 7 the lowest used multiplier and has to be very very brief because I don't see it happening which is very different from the D500... -
No, didn't heard about this issue.
Imo the best thing is to let the CPU do the work, monitor it with TS when you don't trust it... but keep in mind that there are always processes running in the background.
Assuming you could lock the multi at the lowest possible value I think the experience with the performance of the system won't be really satisfying.
I'm pretty sure TS reports the multi value correct, so it would be 7. -
Ok. I have no problem with 7 or 6 but 10, 11... That doesn't seem right to me, especially compared to the D500. And also not concerned about CPU strictly speaking. My belief is that mainboard is not letting CPU idle at 7.
1564 issue is the fan often accelerating to medium through high speed and still at temperatures of around 40ºC - this fan is very powerful and (obviously) loud. People have commented here. My take on that is different though as is my only concern. Most of the time my notebook is idle or under loads of 1 to 3 percent and I am away.
If you or anyone could help, I just wanna know if my CPU should idle at multiplier of 7 steadily, as I see happening with my D500 on Windows XP, instead of averaging at 10.5. -
Well, did you try it with Throttlestop? It should work... Check EIST and Set Multiplier and set it to the lowest value. Then click on "Turn On" and your CPU should be running at the lowest possible multi. (You can verify this in the monitoring section, FID)
As I said, I don't think your CPU will go to such a low state by itself... there are always processes running in the background. My multis are at about 12 at the moment (writing this), every core a bit different of course. This is for the performance power plan.
When I set balanced, the multis are between 9 and 10 immediately (9 is lowest possible for me) so the first thing you should do is to set the power plan to balanced, let me know whether this brings your multi down (use Throttlestop to monitor this) -
tilleroftheearth Wisdom listens quietly...
On Power4Gear Perfomance mode on my Asus U30Jc, my i3 350M also jumps between 10/11 in ThrottleStop.
If I enable TS and set the multiplier to 7, turn on all the check boxes that I can click (and have change) Except for 'More Data' and also set the CPU Clock and Chipset modulation buttons to 12.5%... then I can see the cpu hitting the 7 multiplier quite often, but it is still mostly 10,11...
However, if I simply set the mulitiplier to 7 and enable the 'More Data' checkbox, I can see it hitting 7 'a lot' - but for very, very short time slices.
I'm guessing that even the 'more data' checkbox is still not able to display the switching the i3's are capable of in 'real time'.
Simply put... o770, you're asking for too much (with the current tools we have).
Good luck. -
Well there is something certainly very unusual and probably wrong happening - that I really hope is known to someone here. I've got more screenshots to help us.
As I said earlier, cpu idle state is always the same no matter how the Minimum Processor State is set in the power profile. It's the average multiplier of 10, even with ThrottleStop on, like this:
However under load, just as soon as any task needs process, ThrottleStop and/or the Maximum Processor State in the power profile becomes totally functional - behind ThrottleStop here, Handrake's very demanding task of x264 encoding:
It seems to me that the processor idle state has been associated with the higher multiplier and is enforced somehow.
P.S.: Dear tilleroftheearth
Thanks a lot for your reply! Would you please try to duplicate this I just described - multiplier 7 and EIST checked, does the actual multiplier decrease just as any processing starts?
It's not asking too much from the i3 if a Dothan Pentium M can do it, or is it? -
tilleroftheearth Wisdom listens quietly...
o770,
My i3 350M works like that too!
With the settings you've shown in ThrottleStop, (and not playing with any Maximum Processor Settings in the power options), when I load a heavy duty program like PS CS5 or LR3 the cpu down clocks to the 7 multiplier (CPUz shows ~931MHz).
I think you're on to something (I don't know what yet, but this is very interesting, at the very least).
Why would the cpu down clock when a load is presented? -
I just tested it on my system again, and throttlestop works perfectly for me. The multi gets down to the value I set it, even if the power profile is set to performance and minimum processor state is set to 100%...
Maybe it's an issue with throttlestop and your CPU, you could ask unclewebb in his throttlestop guide thread... It's really a wierd phenomenon. -
tilleroftheearth Wisdom listens quietly...
svl7,
I realize the multi was set to 7, but before a 'heavy' load is introduced, the multi is jumping around 10/11. -
-
Yeah it's definitely Windows, thanks - not my hardware.
Linux live sessions of openSUSE 11.3, Ubuntu 10.04.1 and Fedora 13 can rest my CPU at the lowest multiplier 7 very steadily, just as XP does for my Dothan.
ThrottleStop has worked for me although only when CPU is under load. When nothing needs it - system is idle, multiplier increases to about 10 regardless of ThrottleStop.
Which version of Windows Seven are you running?
Thanks again!
I beleive your setup would only match Intel hardware with mine, right? Chipset and storage controllers, perhaps graphics. Could device drivers cause the issue?
Which version of Windows are you running?
Thanks a lot! -
Yes, that was without load... So I just double checked it with 100% load across all cores with a wPrime test... the multi stays at the lowest level.
I run Windows 7 Professional.
It's really weird that Throttlestop works for you as soon as there's load... once again, I recommend asking unclewebb, the creator of Throttlestop about this issue, over here. -
-
I don't know about HWinfo32, this must be inaccurate, either the multi is wrong or the frequency...
-
tilleroftheearth Wisdom listens quietly...
I'm running Win7x64 Ultimate.
ASUS U30Jc A1. -
Hopefully my post in the ThrottleStop Guide thread will answer your questions.
http://forum.notebookreview.com/7095950-post197.html
ThrottleStop tells it like it is. A lot of other monitoring software does not. When ThrottleStop shows you that your multiplier is jumping up and down like a yo-yo at idle it is because it is doing that. This happens when Windows and the CPU are fighting over the multiplier. One wants it high at idle while the other wants it low. The average is somewhere in between and it will be constantly varying from low to high. That's exactly what ThrottleStop reports. It will completely settle down once your Minimum processor state and C States are in alignment and agree with each other.
T-Monitor is a load based tool that draws a nice graph but the data it reports, last time I checked, has nothing to do with the actual multiplier the CPU is using, especially at idle.
Here is the discussion we had with the developer of T-Monitor and some examples of why I don't recommend using it.
http://www.xtremesystems.org/forums/showpost.php?p=4571661&postcount=4247 -
As slv7 has recommended, I've reported this in ThrottleStop thread for your help and consideration. I beleive my most recent post there demonstrates there is no conflict between Windows power profile and C States.
Also important is the other two accounts of the same issue, one in this thread by member tilleroftheearth, second one here.
For others to follow. Thanks!
CPU Clock
Discussion in 'Hardware Components and Aftermarket Upgrades' started by o770, Jan 23, 2011.