T7200/Vista, Speedstep issues, Please help

I've searched all over for something like this, found one old post, but it wasn't really helpful.

My problem:
when I am under relatively light load and running on AC in 'high performance (set to 100%min/100%max) power profile,' my computer maintains 1900-2000mhz frequency. Which is basically 100%, and my CPU usage stays between 1-20%. This is checked by both cpu-z and vista reliability and perfomance monitor.

Now 'under load!' or significant load the cpu frequency drops to 50% or less. Even as low as to go to 331mhz cpu frequency and 111 mhz bus speed. This makes it rather hard to play LOTRO and listen to music. Hard enough just to play the game.

When I turn on the computer it runs at 100% cpu frequency (just about 2ghz). When I turn on IE with a few tabs. Same thing. Stays around 100%. When I turn on Windows Live Messenger, still no changes. When I throw up CPU-Z and Reliability and Performance Monitor, same thing. Almost no changes. I can even manage to throw media center at it on top of those my cpu frequency starts bouncing from 1980 mhz to 980 mhz and back, continuing this. But I still don't have much of a problem. I'm also running Avast! Virus protection and windows defender.

When I turn on LOTRO that is when more problems occur. Speedstep should be keeping it at 100% under such load, but it lowers it instead. It's acting in a backwards manner. I can't run the game and those other programs at 331mhz... the game struggles too much. My CPU just can't find a place to stick to, and I really want it to stick at 2ghz or close to it, except when I'm not under much of a load.

Any ideas? I need to give this problem fixed.

My Laptop is:
Dell XPS M1210
C2D T7200 (2.0ghz), 4mb cache
2x1GB G.Skill PC-5300 DDR2 RAM
Nvidia Geforce Go 7400
80GB 7200rpm HD
DVD-+RW
Windows Vista Home Basic

PS: This also happens after reformat and fresh install.

 

Update:

Since it seems no one has any ideas that might help me I decided to go ahead and send an email to dell tech support staff describing my problem and situation.

I'll post an update when I get the reply.

Does anyone have any other ideas that might be of help to me? Thank you

 

Not sure why it would be doing that but have you tried changing to balanced mode and observed what happens with the CPU speed?

But anyways I would give NHC a quick go to see if it can control your CPU speed better. http://www.pbus-167.com/nhc/nhc.htm#anchor_download

Basically if you want to to lock the speed juat tick teh box for only use one multiplier and make it your highest one or put it to maximum perfomance mode and see what happens.

 

Running at full speed may be caused by selection of the performance mode power profile.

Slowing down under load could be caused by throttling due to overheating.

An alternative to nhc is RMclock. You can use the monitoring page to see CPU speed, load and temperature. The CPU temperature is normally 50-60C at idle and shouldn't get above around 5C under load. What temperatures are you getting?

John

 

hmm
i say put your CPU through a torture test like ORTHEOS for 1 hour and see if it runs at 100% all the time the test is designed so that you use 100% of your cpus native power.. and use NPC to regulate your profiles btw what notebook are u using >.<

i am using a T7200 also in my V2je and it seems fine but i leave it at battery saving =) less heat same power twice the fun

 

As for the various power management settings. Yes I've tried them and none of them seem to do anything at any point in time. Except power saver when I am running only 1 thing. And it does what it is supposed to then.

Well I just downloaded RMclock.

I still have my game running cause I alt-tabbed out to check this. As soon as I had downloaded RMclock, CPU-Z showed my clock speed to jump up to 1997.46MHz. And it has been staying there. Of course I'm not doing anything heavy in the game right now. Occasionally it jumps down to 980MHz, but then jumps right back up.

RMclock is showing the same thing. This is what RMclock is reading

CPU 0
core clock 1997.32MHz
CPU load 48-60% (Jumped to 98% when clock drops to 980MHz)
Throttle sticks even with core clock.
OS load 28-60% Jumps to 80ish % when clock drops to 980MHz)
Core Temp 70 Degrees Celcius.... yikes. It was at 73.4 a minute ago.

Multiplier (FID)
current 12x (when at 1997mhz, 6x at 980mhz)
startup 6x
min 6x
max 12x

Req. Vcore (VID)
current 1.163V
start up .950V

CPU 1 is more or less exactly the same, though CPU 0 seems to handle more than CPU 1 at any given time...

The notebook is kept in a room that is 22 degrees celcius. And is kept on a hard service...

I'll play the game a little more and check when the game is pushing my computer some more. And report that as well

Any ideas on how to lower the cpu temp?

 

Right before i started playing again i opened up windows media center. temp immediately jumped to 75 degrees celcius...

Then I began playing more. after 5 minutes of play, clock speed was stuck at 998mhz, and throttle dropped to 660mhz putting OS load at 96% or so and cpu load to around 65-70% but temperature dropped to 68 celcius

EDIT
Since then i've checked often after a few minutes of play.
throttle got to as low as 450mhz, and put os load at 100%
temp dropped as low as 66 degrees celcius, and seems to drop the more active the laptop is or when i am more active on it... when i take a break to type this the temp rises. Throttles slowly goes back up too when i alt tab out to type this
core clock is stuck at 998Mhz w/a 6x multiplier

 

undervolting my T7200 shaved off 15 degrees for me, when i first got this notebook it was like 60 degrees when you on idle... that kinda freaked me out so when i undervolted my cpu and idle with max setting its only 47 idle and then it goes to 60 with gaming but this is reasonable because my graphics card is OCed also =)

if you want ill give you the voltages i used here

 

sure please

 

here is the link with the tutorial and instructions
http://forums.notebookreview.com/showthread.php?t=124420

the voltages i used for multiplyers 1-11 i used .95volts

and for the 12th multiplyers i used 1.06volts

 

Use the processes page of Task Manager to check what is using your CPU. When on light load the system idle process should be top of the list with ~90%. I get the impression that something is loading your CPU when it should be on idle.

You should be able to safely drop the maximum voltage of the T7200 down to 1.10V and then gradually work down from there.

John

 

hehe >.< *cough* i didnt do that lol no one responded in the fourm so i just started at .95 and went gradualy higher >.<

*ish tarted* but thanks John do you think 1.06 is reasonable..

 

Well after monitoring it under idle for a little while and under the stress i was giving it a little bit ago, speed step seems to be working. However the temperature is what is causing it to throttle down to extremely poor using conditions.

I'll try out that voltage thing chinmonkie mentioned. What else can I do?

I also try running the program that was mentioned to see what happens when that stresses my computer.

 

o that program is evil
when you run that test your computer is basicaly crippled >.< it uses the max proccessing power of your CPU so nothing else can run .... i kinda felt bad making my notebook run it for 7 hours ... 7 hours of complex math... that uber sucks but its a great benchmarker and it tests stability you will also notice because of the lower power comsuption you will have more battery life =) good luck on your trials and best of luck

 

Well with just running IE w/2 tabs, RMclock, Windows Task Manager, and avast anti virus, I seem to basically be idling... or at least running on light/minimal load.

I have 53 processes going according to task manager. Though the process page doesn't show me anything about idle. So I ran had vista generate a system health report because i remember seeing an idle % under processes there the last time I did one.

This is what I came up with:

Under Image statistics:
Idle is listed at the top of processes with 88.1 kernal cpu% and cpu%
svchost.exe is listed 2nd with 1.7 kernal cpu% and 2 cpu %
rundll32.exe is third and last on the list .7 kernal cpu% and 1.5 cpu%

My CPU is now running at 52 degrees celcius. I have done nothing but close my game and the other programs I was using.

 

52 degrees is good mines at 54... well i been running for like 3 hours so yea but you improved your heat problem and by alot =) you will notice if you play video games it will only go to 61 which is what it would be running normaly stock like idle congradts =) i am glad i could help

 

wait up chinmonkie, I haven't touched anything yet... that was entirely under a light load. I just threw up windows media center to watch something before i take a shower and it jumped to 63 degrees.

 

are you running at max load on WMC ?

 

no, i'm not running at max load when I'm running WMC. Takes more than that to fully stress it, it seems.

EDIT 1: when the cpu clock runs itself at 1997mhz the cpu load is about 48% with IE + 2 tabs, WMC, Avast, Windows Defender, RMClock.
it drops itself down to 980mhz which puts cpu load to 98%. It jumps back and forth between the two

Throttle maintains 1900-1998mhz and puts the OS load between 10-25%

This is for cpu 0, cpu 1 seems to get less of the action... and its percents are lower

EDIT 2:
I just threw up windows live messenger since I almost always run it, even while gaming because people need to contact me. made the temperature jump up to 66 degrees celcius.The other stuff didn't change much except that on cpu 0 when the clock drops to 980mhz the cpu load hits 100-115%. EDIT 3: The temp went back down to 63 degrees celcius.

 

1.06 is quite possible. I think I have my T7200 set to a maximum voltage of 1.05V. I suggested 1.10V as a safe place to start working down from.

The light load CPU temperatures look very reasonable - they depend on the cooling system design and settings for the fan operation. The big fluctuation in CPU speed has me wondering whether there are any speedstep or dual core patches which have been overlooked during the reinstallation? I would start with the latest chipset driver from Inel and then search Microsoft.

John

 

first off i like you to open notebook hardware control and go to CPU speed, thats what i go buy so yea jsut so we are ont he same book. Change the current power plane to Balanced NPC will adjust your speedstep clock frequences depending on your power. This should make up for the no functional speed step and if your not running that much it will drop down to a lower clock =)

edit john i tried going as low as 1.05 for the 12th.. the strangest thing on NPC tho it shows 2 12th multiplyiers so on the safe side i just set both of them at 1.06 and ran OTHEOS for 7 hours 36 mins and it was fine it got as hot as 65 degrees tho under max load i think thats still an accomplishment. But i am going to build my Peilter cooling dock to push the temperatures down even more.. In my calculations i can reduce my idle temperatures from 55 to 32 without causing condensation but the head on the back with the heatsinks will be 60+//

 

I've set my multipliers (1-11) all to .95 volts. I've stepped both 12 multipliers down to 1.0625volts without any problems, so I'm going to run OTHEOS over night since I am studying abroad in japan it is now 3:17am. I'll see how it works out. and post again in the morning.

As for the drivers for the chipsets... where might I find that for my computer on intels site... I looked around for a little bit but didn't see anything that particularly stood out.

 

thous settings are fine when you run ORTHEOS the program will run at full power so aware you should run things down to a min and as for chipsets i will take a look =)

 

Just woke up. Plan to go back to sleep for a little bit but.... here is what turned up or didn't...

Under the 'blend' choice the Torture Test ran 4 hours, 12 minutes 16 seconds - 0 errors, 0 warnings. My cores were running at 62 degrees celcius during this. ORTHEOS said my cpu speed was 1997MHz while it was running, but I think that is just a label of its top possible speed because NHC showed it only running at 999MHz during the test.

I decided to take the 12 multiplier and ramp it down 1 more to 1.05v's. The NHC's little test had no problem doing so. I am going to run the tougher test again to see what happens.

For the 2nd test I'm going to do the CPU stress rather than the CPU/RAM mix. Any ideas why my CPU was clocked at 999MHz during the test rather than its max of 2ghz? Could it be because I've lowered the voltage too much or something?

EDIT: Here is what Dell had to say after I told them speedstep seems to be working fine, but heat has become an issue forcing my CPU to throttle down to reduce it...

From Dell:
"So the machine is running a little too hot, huh? Well, it's good to hear that
SpeedStep is running correctly, but a little unfortunate that it is detecting
your M1210 as running too hot and slowing down your cores to compensate. Now
let's see here, if the machine was overheating, it would be giving you a thermal
event notice, which it's not, so it tells me the system is running within
acceptable limits. I was looking over the specs again as I couldn't recall what
the M1210 temperature safe zone was and discovered it is not uncommon for the
GPU to reach in excess of 90 degrees under heavy load, and for the CPU to
routinely hit over 80 degrees. I came across a forum that was talking about
cooling options and there were several good ideas, some of which you might
already be well aware of:

1) Raise the M1210 slightly higher off the desk then it already is
2) If you're notebook savvy, take it apart and use a can of compressed air to
clean the vents
3) Finally, and this is one that several people swear by on the forums, is to
use Arctic Silver 5 Thermal Grease on the CPU
( http://www.arcticsilver.com/as5.htm) <--- people were guaranteeing drops in
termperature

While I've never tried Arctic Silver 5, and this is by no means a Dell recommend
product, it does seem to have a fairly loyal following as indicated on several
forums I was browsing through. This is not to say you need to spend money to fix
this problem, but if you really want to lower the core temperature, that
wouldn't be a bad idea to think about.

If you've cleaned out the vents, and find the machine is still running a little
too hot, or SpeedStep is throttling your cores, let me know and I can try and
justify sending out a replacement heatsink fan and cooling unit, but it's going
to be a task to get it done since the machine is technically running within
acceptable limits. Nonetheless, let me know what happens, and how things turn
out, and we'll go from there. One way or another, we'll find a way to make sure
your machine doesn't die down when you need some juice!

Sincerely,

~~~~
XPS Email Technical Support
"

EDIT 2:
Under ORTHOS's Stress CPU with gromacs core
With my CPU's 12multiplier set at 1.05v

This is what I got when I chose to shut it down:
No problems detected in 100507 iterations, bye!
StressCPU ran 4 minutes 58 seconds.
Execution halted.

The temperature had hit 83 degrees celcius and was running at 1997-2000MHz the whole time... As soon as I stopped it the temp plummeted 10 degrees celcius and has since continued to reduce. I think I will shut down the laptop for a little while. Letting it completely cool off, then start it up and try again.I might try some gaming and multitasking first.

Btw, does anyone know anything about that arctic stuff the dell rep mentioned?

 

yes the arctic silver is a thermal compound that you place ontop a CPU or GPU before adding on a heatsink basicaly think of it a heat transfering glue if applied properly it can disipate your heat evenly on your heatsink and make it cooler.

as for NPC the profile you should be running at is max performance and that will set your CPU at the highest clock then run OTHEOS if it goes up to 70 degrees dont worry that is well in respected zones as both cores are proccessing at max power

the temperature guide is as followS

49-59 excellent
60-70 good *excellent under heavy load*
80-90 not bad * additional cooling is recomended*

 

My first thoughts are at this time with what you are describing I would not be undervolting. I would be trying to get the CPU speeds and temps operating correctly, If I could not I would if under warranty have them figure out, fix and or explain what is going on. I would undervolt later to further reduce temps but never to achieve normal operation. Solve the problem first, tweak later.

Edit: How old is the computer, if it is a couple months old and used in a low dust non smoking environment I can not see “blowing out doing any good”. Also while I have read good things about “Artic Silver” I will not use it on mine until warranty period is over so as to not risk voiding. If you are under warranty and Dell guy suggest great way for Dell to get rid of you. Refuse coverage, also I don't like the way he recommended with no personal experience (he works for Dell show some professionalism).

 

hmmm i agree on you there i must have gotten the question mixed up >.< if the cpu speeds arnt working correctly then i think there is something up with the chip or the mobo board... If its under warranty i think you should send it back for alittle look see >.< or take Johns advice 2nd page >.<

 

I agree about not putting anything on the computer while its still under factory warranty...since I did not buy the laptop myself I had no choice what sort of extra warranties/coverage that came with it, so I only have the most basic normal warranty. The laptop was bought in February of this year. However, from what I can tell, Dell isn't willing to do anything about this problem because they feel the computer is operating within the 'acceptable' parameters. Even if in those parameters the CPU throttles down making games performance extremely poor.

 

I understand and accept what Dell says about the temp if it is going to 83C and not higher. I would not like it if it was mine but would not worry too much as they are rated up to 100C (and you have a 12”? what GPU?). The throttling down is unacceptable, especially when it effects performance. Now when you talk to Dell don't confuse the issue anymore because they are easily confused. Many people in the past have posted regarding the stupid things they sometimes say like they read from a script where the final sentence is always “doesn't sound like you have a problem”. The other thing is do not get discouraged if one or two or five don't get what you are saying, you only need one to understand you. The Dell guy told you if it throttled down to save itself it would report and it does not so you have a problem. If it is a driver issue as John said at a minimum they should be able to inform you and help you, but if not you might have a serious problem and don't take no for an answer. If they say no and you go away from their twisted view that is good business (profits).

Have things changed since you received this or have you just become more aware of what is happening.

 

Alright been gone a few days...

So In response to powerpack... when I got the laptop I don't remember having any issues like this. Only did it occasionally cause problems, but rarely did I ever stress it like I do while I am in Japan studying abroad. When I was at home I had my desktop and stereo and other things, so I could let my laptop do just one thing... which might explain why I never noticed anything.

Since I installed NHC the throttling down has seemed to stop... to an extent. If I set the current power plane to high performance, and the cpu speed to max performance.... it doesn't work. If it does it is just for a few seconds... then it drops back down to 999MHz, if I am running RMclock simultaniously to check for cpu clock and throttling, they stay the same, both at 999MHz give or take 20MHz.

If I keep the power plane at high performance and set the CPU speed to dynamic switching... my clock goes from 980MHz-2GHz, and the throttle follows but it is constantly changing... going up and down under dynamic switching.

When I am running my game, WMC, Windows live, and an internet window, and the cpu stays at 1ghz (under max performance (even though this is clearly half max...)) my cpu temps stick around 58-65 degrees celcius... however, this is after I undervolted.

I seem to be able to run fine and stable with multipliers 1-11x at .95v and both 12x multipliers at 1.05v. Now when I have it set to dynamic switching, or when it manages to stay at 2ghz clock and throttle the temp bumps into the 70's. However, I haven't seen any major down throttling since I put NHC on and undervolted. I've also lifted the laptop about 3/4 an inch off the via its corners.

Yes its a 12.1" Laptop. The GPU is a Nvidia Geforce Go 7400. Most powerful GPU that would fit in this thing back in february. Hard drive temps stick around 38-42 degrees celcius, so I hardly doubt that they are contributing much heat. I have no idea what the GPU temps are running at under heavy game load.

But I agree about the horrible dell comments... Saying that throttling down is acceptable because its the computers way of protecting itself when it gets hot... that is like saying its acceptable for a car's engine to shut down when someone takes it close to the red line... or that its okay for the breaks not to work when you need them most. Probably not great analogies, but still vital components shouldn't operate this way, and if they are, its a problem.

EDIT: Sent dell another email telling them what I've done, how things have worked and what not. Also telling them that I want this problem fixed without me having to undervolt. Though it seems undervolting hasn't completely worked because only under dynamic switching do I ever get the comp to run at 2ghz...

 

It might be to late now but I hope you did not tell Dell about "undervolting" as they might grab onto that and put the problem back on you.

I would uninstall (or at least stop using RMClock) and uninstall NHC, then reinstall NHC and only monitor temps and see where you are at. Then contact Dell from there. And yes after that do what you have to do be it undervolting or whatever but at least you will have your baseline.

 

Too late about mentioning the undervolting to dell. I said it was an option. I tried it and it seeemed to work to a point, but that I did not want to do a quick fix that way. and that I intended to revert back to the original voltage and search for a fix that doesn't require that sort of tweaking. We'll see how it goes with dell.

I'll do what you suggest and see where it goes from there. I might not have the internet for the next two weeks though, so I do not know if I will be able to report anything here or not.

 

well the manufactuer has to give you the warrarnty regardless because you got it like that plus the sevice guy told you that was fine... wow bad service people..