Well I have a XPS M1730.The specs are listed in my sig.I 'think' I might have a CPU downclocking problem.The reason I say 'think' is because I have not faced any downclocking issue with any of my other machines in the past and therefore I am not sure whether I actually have a problem or whether it is just me misinterprating some of the tests I have done.From my sig you would see that my 1730 has the Core 2 Extreme X9000 processor.I have the Dell bios 'A06'.'Speedstep' is enabled in the bios(factory setting).I do realize that with 'speedstep' enabled, the CPU will automatically downclock itself when it is not under load or when it is idle.So I used 'Orthos' to keep my CPU load at 100%.
I ran 'Orthos' to keep my CPU loat at 100% and used 'RMClock' to monitor the core clocks and temperatures.Both Orthos and RMClock were run simultaneously.While running the tests, my XPS was running at 3.4 Ghz(overclocked from Dell Bios).My XPS was running in 'High Performance' mode and it was plugged it throughout the test.At first( after 4sec)RMClock shows that at 100% CPU load my core clock is 3391.46 MHz.Then after running Orthos for about 1 and a half minute my core clock drops to 2792 MHz.After 10 minutes it drops to 2394 MHz.Well obviously it does noy stay at 2394 MHz,jumps back to 2792MHz,then to 3391 MHz and so on and so forth.
Well If I understand it right, shouldn't my CPU clock stay at 3.4 GHz throughout the test as Orthos is making sure that the CPU load is 100%.But as you see the core clock drops to 2394 MHz(2.4 GHz).So guys do I HAVE a downclocking issue or am I just missing something?
I am providing some screenshots of my test with RMclock and Orthos.
![]()
![]()
![]()
![]()
Guys, any input on this matter would be appreciated.I wanted to get your opinion on this matter to see if there is actually something wrong before I contacted XPS support(don't wanna look stupid or ignorant when I speak to them).
THANK YOU!!
-
Commander Wolf can i haz broadwell?
If the spikes are far apart and very sharp, there probably isn't a problem. Orthos will drop the CPU load for a tiny bit as it shifts loads and stuff. Unless your processor is staying at those lower clock speeds I wouldn't worry about it. Can we see a CPU load graph or something? The instantaneous clock speed pics aren't terribly helpful.
-
If you have your power settings on "High Performance" you don't need to run any software to try to artificially keep your CPU running at its fastest speed. The default "High Performance" profile prevents SpeedStep from reducing the clock rate.
The problem you're describing sounds very much like the CPU forcing a slower clock due to a (probably very localised) overtemp situation. This has been a feature of Intel chips since the quite early P4's. It brings to mind the video of Quake 3 running whilst someone removed the heatsink from the CPU. The CPU slowed its clock rate almost to a crawl to prevent dangerous overheating.
Does the same problem occurr if you limit your maximum overclock to 3.2GHz? Not every X9000 in every M1730 can do 3.4GHz. -
I believe that the x9000 will downclock (throttle down) at 80 degrees.
So use orthos and check if that's the case by looking at the temps and cpu speed in RMClock.
I would suggest trying to lower the vcore at 3.4GHz to something less than 1.575v as this is too high.
Mine can work at 1.4125 but anything under 1.5v should lower temps and probably reduce (or get rid) of downclocking.
Give it a go and let me know if this works
p.s. Below is the best compromise I found that keeps my laptop very cool, almost silent (most times the fans do not work at all and when they do, even when running Prime and Crysis they are at low speed and barely noticeable) and very fast:
CPU: 3.0GHz at 1.25v (using RM clock)
GPU: 600 900 (using NVtune)
With this setup I get a 3DMark of 13850 (only 975 less then 3.4GHz) and Crysis 1900x1200 all high 31.5 fps (only 1.3 fps less than 3.4GHz).
The lower temps and noise are really worth the negligible loss of speed! (in your case you might gain some speed as you would not have heat problems and the subsequent downclocking). -
Keep a close eye on your temps.
I think it might be thermal downclocking after it hits 70c or so
You might wanna use CPU-z instead of RMclock to make sure RMclock isnt influencing the downclock -
Hey guys,thanks for your replies.It took me a while to get back to you guys because I was at work and had no access to my XPS(for further tests).
@Commander Wolf
I will provide a graph by today.
@micloi
How do you lower the vcore?(I have absolutely no idea,never done it before)Instructions please!Thanks.
@VinylPusher
I will be limiting the overclock to 3.2 GHz and will check if the same issue occurs.
@flipfire
I willl use CPU-z and post the results. -
Hello again guys,
I found that the CPU downclocks from 3.4 to 2.8 GHz in just a matter of a few seconds after running Orthos.Within a few seconds temp increases to around 70 degree centigrade and it is at this temp range that the CPU starts downclocking.My point is, if the XPS can't handle running at 3.4 GHz without overheating and downclocking,then what is the point of letting users to overclock to 3.4 GHz.
Another thing, after running Orthos for around 10 minutes,the temperature hardly goes below 80 degress(sometimes jumps to 70's,then back to 80's) and as a result the clock stays ay 2.8 GHz(machine running at 3.4 GHz).It hardly jumps back to 3.4 GHz(it goes back to running at 3.4 GHz only a couple of times for a few seconds). -
Here's a graph(used RMClock) after running Orthos for around 10-12 minutes.CPU at 3.4 GHz.
Graph after running Orthos for 3 minutes.CPU at 3.2 GHz.
P.S-I forgot to mention in my previous posts that I am already using the Zalman NC 2000 notebook cooler.
@flipfire
I tried using CPU-z and get the same results(downclocking occurs after temp reaches around 70 degress). -
So what do I conclude from the tests?Anyone?
-
Judging by the graphs its thermally downclocking and then upclocking again when the temps are fine
You might wanna consider a lower OC speed unless you have the best cooling.. -
Thanks flipfire.Any other opinions?
@micloi
How do you lower the vcore.Do you mind telling me the steps.Cheers! -
Then expand "profiles" and select "Maximal Performance".
Under "AC Power" select the checkbox "use p-power transitions" scroll to 17x and select that checkbox as well.
Then click on the "profiles" tab again and under AC Power and next to "current" select "maximal performance" and press apply.
VOILA!
Run Orthos to make sure it works ok. If it crashes upper the VID s bit and try again (mine works at 1.4125 but yours might need around 1.45, which is still much lower than the default 1.575 which causes overheating). -
Use micloi's instructions and you should get rid of the nasty throtling "feature".
My X9000 also works 100% stable with 1.4125V voltage at 3.4GHz, temps are very good. -
Offering an overclock to 3.4GHz is useful for those that can run stably at that speed. Not every X9000 is the same. Some will require higher voltages than others to attain a stable 3.4GHz. Some may never run at 3.4GHz stable.
Take heed of the other posters warnings about over-volting your CPU. Power consumption (and thus, heat generation) corresponds directly to the voltage used. Not linearly (i.e., an extra 0.03W per 0.03V) but as the square of the voltage! So, even small reductions in voltage can have big effects on heat generation. -
Thanks micloi.I followed your instructions and temps remained around 72 degrees.It downclocked MUCH less.I was wondering though, is it possible to start the laptop with these settings?I mean. do I need to set the voltage values everytime I start the laptop?Can you make RMClock to start during start-up.
Thanks nd4spdvn for your comment as well. -
What voltage did you use BTW?
Also if you get a notebook cooler you might get it under 70 and stop downclocking completely (or run at 3.2 which is still very fast!) -
I 'almost' got no downclocking with the processor running at 3.2 GHz and 1.400V.I ran Orthos for 10 minutes and found almost no downclocking.For 3.4 I set the voltage to 1.4125.
Micloi,I already use the Zalman NC 2000,and the tests were done with the notebook cooler.Now let me try starting the settings at 'start-up'.Thanks again. -
Also try a bit lower voltage for 3.2GHz (I use 1.25v at 3.0GHz so at 3.2GHz you should need less than 1.4v (1.35 maybe?) and keep downclocking to 0%)
One last question, what's the ambient temperature in your room? -
Good thing I brought the XPS with me to work today.Usually I never carry it around.
Do I have a CPU(X9000) downclocking problem?XPS M1730
Discussion in 'Dell XPS and Studio XPS' started by nahiyan13, May 15, 2008.