1640 Throttling - post here

I do not have any problems with my 1640 either. The only problem I have is that it seems that Dell tries to hide some performance issues from the users.

Dell engineers are for sure aware of the problem since they wrote the Bios that is responsible for throttling. The way that they use to throttle covers the throttling from the user since the multiplier does not change.

The user just recognizes some situations in which the laptop seems to run slower and cannot really tell why unless having a deeper insight into CPU architecture.

That is something which really should not happen on a premium laptop. The next thing I will do is reading the Core 2 Duo data sheet. Maybe I will write a small program that uncovers throttling.

 

Me to Cin.

 

I just run RMClock and Throttlestop simultanuously.

Code:

  DATE     TIME    MULTI   C0%   CKMOD  CHIPM  DTS
*Prime Running*
02/14/10 12:54:11   9.00  100.0  100.0  100.0   48
02/14/10 12:54:12   9.00  100.0  100.0  100.0   48
02/14/10 12:54:13   9.00  100.0  100.0  100.0   48
02/14/10 12:54:14   9.00  100.0  100.0  100.0   48
02/14/10 12:54:15   9.00  100.0  100.0  100.0   48
02/14/10 12:54:16   9.00  100.0  100.0  100.0   48
02/14/10 12:54:17   9.00  100.0  100.0  100.0   51
02/14/10 12:54:18   9.00  100.0  100.0  100.0   48
*Started FurMark*
02/14/10 12:54:19   8.78   94.6  100.0   75.0   52
02/14/10 12:54:20   9.00   70.2  100.0   75.0   52
02/14/10 12:54:21   9.00   70.2  100.0   75.0   52
02/14/10 12:54:22   9.00   70.2  100.0   75.0   52
02/14/10 12:54:23   9.00   70.2  100.0   75.0   52
02/14/10 12:54:24   9.00   70.2  100.0   75.0   51
02/14/10 12:54:25   9.00   70.2  100.0   75.0   52
02/14/10 12:54:27   9.00   70.2  100.0   75.0   52
02/14/10 12:54:28   9.00   70.2  100.0   75.0   51
02/14/10 12:54:29   9.00   70.2  100.0   75.0   51
02/14/10 12:54:30   9.00   70.2  100.0   75.0   51
02/14/10 12:54:31   9.00   70.2  100.0   75.0   51
02/14/10 12:54:32   9.00   70.2  100.0   75.0   51
02/14/10 12:54:33   9.00   70.2  100.0   75.0   51
02/14/10 12:54:34   9.00   70.2  100.0   75.0   51
02/14/10 12:54:35   9.00   70.2  100.0   75.0   51
02/14/10 12:54:36   9.00   70.2  100.0   75.0   51
02/14/10 12:54:37   9.00   70.2  100.0   75.0   50
02/14/10 12:54:38   9.00   70.2  100.0   75.0   52
02/14/10 12:54:39   9.00   70.2  100.0   75.0   51
02/14/10 12:54:40   9.00   70.2  100.0   75.0   50
02/14/10 12:54:41   9.00   70.2  100.0   75.0   50
02/14/10 12:54:42   9.00   70.2  100.0   75.0   51
02/14/10 12:54:43   9.00   70.2  100.0   75.0   50
02/14/10 12:54:44   9.00   70.2  100.0   75.0   50
02/14/10 12:54:45   9.00   70.2  100.0   75.0   50
02/14/10 12:54:46   9.00   70.2  100.0   75.0   50
02/14/10 12:54:47   9.00   70.2  100.0   75.0   50
02/14/10 12:54:48   9.00   70.2  100.0   75.0   50
02/14/10 12:54:49   9.00   70.2  100.0   75.0   50
02/14/10 12:54:50   9.00   70.2  100.0   75.0   50
02/14/10 12:54:52   9.00   70.2  100.0   75.0   50
02/14/10 12:54:53   9.00   70.2  100.0   75.0   50
02/14/10 12:54:54   9.00   70.2  100.0   75.0   50
02/14/10 12:54:55   9.00   70.2  100.0   75.0   50
02/14/10 12:54:56   9.00   70.2  100.0   75.0   49
*Quit FurMark*
02/14/10 12:54:57   9.00   70.2  100.0   75.0   50
02/14/10 12:54:58   9.00   70.2  100.0   75.0   50
02/14/10 12:54:59   9.00   70.2  100.0   37.5   50
02/14/10 12:55:00   9.00   35.7  100.0   37.5   55
02/14/10 12:55:01   9.00   33.9  100.0   37.5   56
02/14/10 12:55:02   9.00   34.2  100.0   50.0   55
02/14/10 12:55:03   9.00   46.0  100.0   50.0   55
02/14/10 12:55:04   9.00   46.0  100.0   50.0   54
02/14/10 12:55:05   9.00   47.1  100.0   62.5   54
02/14/10 12:55:06   9.00   58.1  100.0   62.5   53
02/14/10 12:55:07   9.00   58.1  100.0   62.5   53
02/14/10 12:55:08   9.00   60.0  100.0   75.0   52
02/14/10 12:55:09   9.00   70.2  100.0   75.0   51
02/14/10 12:55:11   9.00   70.2  100.0   75.0   51
02/14/10 12:55:12   9.00   72.8  100.0   87.5   50
02/14/10 12:55:13   9.00   82.3  100.0   87.5   50
02/14/10 12:55:14   9.00   82.3  100.0  100.0   50
02/14/10 12:55:15   9.00   87.4  100.0  100.0   47
02/14/10 12:55:16   9.00  100.0  100.0  100.0   46
02/14/10 12:55:17   9.00  100.0  100.0  100.0   47

The results look quite similar.

By the way: Is it possible to get the source code of throttlestop. I would like to write something like RMClock. The Throttlestop source should include most of the necessary hardware methods.

 

Probably not.

ThrottleStop is not a very complex program so I'd be more than happy to tell you what registers need to be checked for temperatures, clock modulation, multipliers, etc. There is plenty of Intel documentation so I'll point you in the right direction if you're interested in learning more or writing your own program.

Intel® 64 and IA-32 Architectures Software Developer's Manual
Volume 3B: System Programming Guide
http://www.intel.com/Assets/PDF/manual/253669.pdf

Appendix B lists the registers. MSR 0x198 gives info about the current multiplier, MSR 0x199 is set to the requested multiplier, MSR 0x19A will show you if clock modulation is being used and MSR 0x19C is where you read temperature data from. To read and write to these registers in an operating system friendly way you need to use something like the WinRing0 library which is digitally signed so you can use your new program in Windows 7 x64.

If you ever have any questions, just send me a PM.

I'm still hoping to see an image of BOTH of those programs running side by side on your desktop when RMClock starts to show one multiplier on one core sagging down. I think this is a reporting error by RMClock and if I see what ThrottleStop is reporting then it will make a lot more sense. The method that RMClock uses to determine the multiplier is an easier method than ThrottleStop uses but in some situations, it is not as accurate. With a screen shot I can compare the two programs and explain that further.

Your P8600 supports IDA (Intel Dynamic Acceleration) mode which means it can use a 9.5 multiplier when only a single core is active if you have C3/C6 enabled in the bios. ThrottleStop can report stuff like this very accurately. Just run a single threaded benchmark like SuperPI and you should see average multipliers between 9.0 and 9.5.

 

I just played 3 hours of Star Trek Online at very high settings at max res with full LCD brightness and zero slow downs, infact multi tasking, etc was never slowed down either.

I have a feeling if your running a P series C2D and no RGBLED you won't be affected. The T series processors eat much more power than the P series as does the RBGLED display vs the regular LED one.

 

In other news, my psu may be the bottleneck, seen as locking the multiplier results immediately in a blue screen. Thanks again unclewebb, I appreciate the efforts you've put into creating this program!

 

Did you try turning off speed stepping? Though I'm not sure it would make a difference as I'm sure they are using BIOS level settings like CE1. I wonder if your using a piece of software to lock the multiplier is the cause of the blue screens...

 

My comment about the blue screen was because I used ThrottleStop, which means my psu may not provide sufficient wattage, rather than the bios disallowing it. I just needed to know why I down clock at times since it definitely isn't heat. If I had some other psu to try out I would. Maybe I can get my hands on my friends for his Asus G51J lol.

 

Lets keep this clean and leave any attitude at the door. The post in question was nothing more than a person stating they had not encountered an issue.

If an owner is experiencing an issue with throttling, then its a problem. On the same token, if another owner is not - well, that is fine also.

Please keep forum rules in mind when posting -

Thanks.

 

When adjusting the multiplier on Core 2 Duo computers always adjust the voltage first to its highest value in ThrottleStop if you're not sure how much voltage your CPU needs to be stable.

I might get around to adding a safety feature into ThrottleStop at some point so you you'll have less chance of forcing a high multiplier without also using more voltage. It's on the things to do list.

 

I havent been following, but how has the bios+130w solution worked out for the 1645 and 1647?

 

Bios A07 for the XPS 1645 does not allow access to the full 130 watts so there can still be some throttling. It is much better in many situations but it looks like it will never be perfect. I think Dell is still debating whether to ship 130 watt adapters for the XPS 1647 and release an updated bios for it.

 

Is there anyway I can just return my laptop at this point? It gets extremely hot whenever I do anything taxing. And I have a feeling when the fix comes if it ever does it will just make it run even hotter..?

 

tenknics, most laptops you get with a good GPU will be warm/hot(subjective) without additional cooling for high load applications. Most use a single fan to cool the system, which keeps it within specs but really need 2 fans or more to keep a CPU and strong GPU(s) cooled.

1640 has a little extra design flaw in that there is a chip, the southbridge, located directly under the touchpad. Southbridges normally have a small heatsink and are passively cooled (no fan). They can run warm/hot without issue though and on the 1640 that heat goes right up through the touchpad.

I run my 1640 mostly on a Zalman NC2000 cooling pad and the lowest fan speed cools the south-bridge with ease, so the touch-pad doesn't warm up.

1645/47 have addressed this issue by moving the south-bridge and extending the internal heat-pipe/fan cooling system over it.

 

Hi guys..

i am having throttle issues on my laptop

specs are CPU T9800 @ 2.93GHZ
15.6 1080p non RGB screen
4GB RAM DDR3
ATI 4670 1GHZ
500GIG HDD
Win 7 Home 32bit

I play games like Heroes or Newearth, Street fighter 4, counterstrike source and they all seem to throttle the graphics down to unplayable.

since then i have investigated that i too had throttle issues similar to the 1645's where multiplier will drop from 11x down to 4, 5 or 6x for some time before going back up to 11x again.

Since then i have been in contact with DELL AU tech support team which they are trying to replace parts which i am letting them but if nothing is fixed i will be fighting for a re fund.

Have been buying dell for a very long time but will not after this model. (inspiron 6400, inspiron 1520, 1210xps, 1330xps)

 

I hope to God that you are NOT running Windows 7 32bit...

 

unfortunatly i am...

 

Why not 32-bit?

 

Hello there.

Ever since I've upgraded from Vista 64-bit to Windows 7 64-bit my Dell Studio XPS 1640 has been throttling down. This has happened mainly while gaming. I'll be playing a game for around ten to fifteen minutes and then my CPU will throttle down to 266 MHz, which I find quite ridiculous. The only other time I've experienced it otherwise (which happens very rarely) is while I was watching HD video.

Before I proceed, I may as well mention that I have the RGBLED display (1920x1080), but I keep it at low brightness.

I have done several tests and I know it's not a heat issue (I even regularly use a cooling pad), but it's possible that a certain CPU temperature triggers the throttling itself. I've been lurking around several threads all over the net (including this site of course) and I notice this issue seems to be happening more on the 1645, with most people claiming it's a bios and power-related issue.

I don't know if it's the same for the 1640, although if it is, I'm curious as to why this would only happen after making the upgrade to Windows 7. I've tried reformatting; adjusting settings in Catalyst Control Center; pretty much everything I could think of, and this problem has still been persisting for months. This is really unprofessional of Dell, quite intolerable really; it needs to be fixed.

I love my XPS 16; it's an amazing machine (I was really happy with it before the upgrade), but this conflict has been really disappointing. I didn't pay over a grand for a laptop that locks it's CPU multiplier and runs my video games at the pathetic speed of 266 MHz.

I don't know what to think at this point. I just really want this all to be resolved, and quickly.

Any feedback would be appreciated. I hope everyone else who has this problem posts about it and supports a fix.

Thanks.

 

32Bit only addresses up to 3.75GB of memory. Hence Dell selling all PC's with 4GB or more with a 64Bit OS.

With Vista once SP 2 came out MS changed it so the control panel says 4GB of ram installed, instead of the actual amount of ram the OS can handle. With Windows 7 it does the same thing.

On top of that a 64Bit os is generally faster all the way around and if your running 4GB or more it's a must if you want to use all of that 4GB or more of memory.

 

OK 99% of this is way over my head. How do I even tell if my 1640 is "throttling"?

 

Get hwinfo32, furmark, prime95, and hwmonitor. Run them all at the same time and see if hwinfo32 reports cpu clocks lower than what they should be.

 

right, i knew the max addressable memory thing, just your post made it seem like there was some other major disadvantage to 32bit.
Either way, there was a an article that actually found that for everyday use 64bit OS was slower than 32bit. and regarding memory usage i find that its generally rare to see even 2gb used up on my machine, yet alone for it to be touching anywhere near 3gb. Even when running crysis I see no improvement in any benchmarks or games from upgrading from 2gb to 4gb, i am running 64 bit win 7 btw.

 

Dell doesn't sell all PC's with 4gb+ memory with 64 bit operating systems. I bought my M1530 which had 4gb of ram with Vista 32 bit (ouch!). It's not a huge deal for anything I've done to be missing 500mb of my ram, but 64 bit does feel faster in general.

 

Ok, back on topic...

As far as throttling on Windows 7 and not Vista. Maybe you just didn't notice it when you had Vista..Or something within Windows 7 power management scheme is causing our 1640s to throttle...Is that even possible Unclewebb?

 

I doubt it, it is being throttled by the bios itself. I experienced throttling in both XP and 7.

 

Hi there,
Im new to this forum and notebooks. I recently bought a 1640 for university thinking that it would be agood option for me as I could do all my work as well as play most games fairly comfortably for a decent price.
Actual specs:
P8700 @ 2.53
4GB DDR3
4670
W7 Pro x64 w/ BIOS A12

Ok so I didn't find out about the throttling issues until the unit was on it's way to my house *facepalm* and once recieving the notebook it seems that the problems are worse than I could have thought.

I play CSS regularily and particularily this game seems to affect my computer badly.
i ran throttle stop for logging purposes and as soon as i opened the program I noticed that my multiplier was 3?!
Anyway here are my results playing CSS with about 10 bots.

 

Power saving ftl doh.
Test redone and attached

 

Sorry to hear that the throttling bug has bitten you too Edmund. Thanks for the post though, your information will help add to the growing list, hopefully quickening the resolve.

 

I was wondering what my options with Dell were from Australia. Once I called Dell to help fix my eject button problems but got a clueless tech who took 40mins of fiddling around to tell me to update my bios. I was going to bring up the throttling issue but thought it was best to hold my tongue. The easiest fix seems to be to buy an aftermarket 130W adaptor and run throttlestop but i'm not sure if this is a good option. I am also having second thoughts about the 1640 as a whole and may want to return the 1640 and swap for 1645 plus cash diff with a 130w dell adaptor. Hmmmmm

 

This is a question only regarding temporarily fixing this. How many multipliers down would you have to lock the cpu at to eliminate the throttling? Having the cpu always down clocked slightly would be better than the huge drops from a high frequency in my opinion.

 

Sirhcz0r: Every CPU and laptop and game load is unique so it's hard to come up with an exact formula.

You might be able to use the clock modulation feature of ThrottleStop to limit power consumption and prevent your laptop from reaching the throttling point. A clock modulation setting of 87.5% or 75.0% would be a lot better than the multiplier dropping down to 3.0.

Another thing that would help is reducing your maximum core voltage if you can. If you are on the edge and trying to prevent this throttling from happening then I would disable C3/C6 in the bios which should prevent your CPU from using IDA (Intel Dynamic Acceleration) mode if your CPU supports that feature. By doing this you should be able to lower the CPU voltage and reduce heat and power consumption at the same time. When trying to do this with ThrottleStop, set VID to the maximum voltage to start with and then select the VID box to enable it before playing with the multipliers or anything else. Run a benchmark like Prime95 on both cores and then slowly reduce the voltage until you lose stability. Use just enough voltage to remain Prime stable and you should be able to play any game without any issues.

A 130 watt adapter and using ThrottleStop to force the multiplier to its maximum and forcing clock modulation to 100% is a much better solution. 130 watt PA-4E adapters are reasonably cheap on EBay.

 

Just some words of opinion.

I appreciate your contribution, unclewebb, but I'm personally not interested in ThrottleStop. I want a real, official fix from Dell. It appears that the program itself is a risk, and for that, I would not consider it wise to spread its usage around. I would not advise it.

I am no expert but, what I would advise is that we all stay patient in this matter even though it's very frustrating to deal with. If we rush into the idea of any fast solutions, we may only further our problems, and there is no telling what kind of damage could be done to our hardware. I will only accept an official fix from a profession. No offense sir.

I am hoping that this issue will be resolved before summer, and that is in the latest. It has been very intolerable from the beginning. I would like to think Dell is taking this problem seriously and has a solution on the way as we speak.

For now we wait.

 

I'll think more about getting a new adapter once I see how the sxps 16 replacement machine is. I desperately tried to get a huge psu working today but I didn't have the right adapter for going from Asus to Dell. I might be able to do that tomorrow though with a little adapter. I already have a great undervolt that I run with rmclock, but I only have guesses as to what the differences between clock modulation and multipliers are. Is the modulation keeping it at high multipliers and reducing the fsb speed? I'll check out my bios at some point to see if anything else can be disabled that couldn't be done through rmclock.

Thanks for everything unclewebb!

 

@unclewebb

I saw what you mean about decent prices on ebay. There are some of the 130 watt adapters that come with the precision M6500 available new for as little as $40. If there are problems with my sxps 16 when I get it I'll pick one of those up if Dell's not going to resolve this in a timely manner.

 

The XPS 1640 has had performance throttling issues for over a year so if you want to wait another 6 months for Dell to come to the rescue, that's fine. Just keep in mind that there's no guarantee that they will ever fix this problem.

Many users are completely fed up waiting for Dell to do something. This isn't some new issue for Dell. The Latitude E6x00 has been throttling for 18 months and Dell's December bios update shuffled the throttling scheme a little but didn't fix the problem and they have no intention of ever fixing this model.

ThrottleStop when combined with a 130 watt adapter is a solution that will immediately allow a user to run their laptop at very close to its full potential. It's also the most accurate tool to log and document throttling.

You may not be interested in this tool but a lot of users that are totally fed up waiting for Dell to do something are going to be very interested in finally being able to run their laptops at full speed.

 

I just heard from a happy user today that is using ThrottleStop with his XPS 1640 and a 90 watt adapter and the performance difference when gaming is much improved. No more stuttering.

These adapters have a safety feature built into them so if their output ever exceeds 90 watts DC, they are designed to automatically shut down. When this happens, you will need to power cycle the adapter on and off by unplugging it to reset it.

My original fear was that when using ThrottleStop with a 90 watt adapter, it might cause the adapter to try and put out more than its rated 90 watts DC and could cause it to permanently fail but this built in safety feature is designed to prevent this from ever happening.

I still highly recommend a 130 watt adapter since they run a lot cooler and provide additional headroom so you can recharge your battery while using your laptop.

 

Maybe we could all chip in and buy Dell some new spectacles because they have been "looking into it" for a long time without any success.....

 

How does this sound, I am going to issue the warning. Using a third party tool, is done at your own risk and any damage caused is probably not going to be covered. Bottom line, if you support webb's tool then great, if you don't want to use it - well that's fine also.

There is no need to slam someone for sharing his opinion and it will not be permitted. I previously posted a warning about this. Any more posts of this nature will be removed and infractions will be issued.

 

I just let it run for 30 minutes and I got no warning signs. I do not really know what I am supposed to be looking for. The hottest HWM ratting I got was 75c. On average it hovered around 70c. What was I supposed to look at again? Sorry I am a newb at this stuff.

 

While the other stuff is already going, look at what hwinfo32 says for the cpu multipliers and frequency. Pretty much if everything is okay both green bars are be full (indicating maximum cpu frequency and no throttling).

So step by step this would mean: open and minimize hwmonitor (just to check later (but your heat doesn't seem bad at all), open furmark and run it at a high resolution with 16x AA (but make sure it's windowed and not full screen so you can check hwinfo32), open prime 95 and start it, then open hwinfo32 and watch for any downclocking (greens bars not being full).

Also if you like you can try doing all the same things but using ThrottleStop in place of hwinfo32. For that the only thing that's different is you have to know what your maximum multiplier is.

 

I will run it again and then report back. How long should I let it run?

 

Just keep an eye on it for 2-3 minutes. If it does have the problem it should show much sooner than that.

 

You can run a ThrottleStop log and upload that to www.sendspace.com
Post a link here and I'll have a look at your data.

If you are using HWInfo32 then it should look like this at full load.



If the MHz is way down at full load then that would be a bad thing. The HWInfo graph responds to both multiplier throttling as well as clock modulation throttling.

The next picture is what you don't want to see.



That shows your CPU running at a fraction of its potential. You might see this at idle depending on your power saving options but you shouldn't see this at full load.

 

OK... so I did it and I definitely have throttling goin on (ie my screen looked like the second image). Soooo... at this point what can I do about it?

I realized by BIOS was only AO8 so I upgraded it to AO12... this did nothing. The green bars are still not full and still fluctuate.

 

Download RMclock and lock your p-state. Then hope and prey Dell cares enough to fix this.

Welcome to the wonderful world of throttlegate.



Speaking of which..Unclewebb, On full load. why is that when I have my t9800 at a locked p-state of 11x (and therefore a fixed voltage of 1.025) I don't get throttling..Then if I don't lock the p-state (no voltage lock in turn) the voltage climbs to 1.163 under full load then throttles.

full load at 11x 2.93ghz at 1.025v = no throttle (throttle occurs a little on a 2nd core)
full load at 11x 2.93ghz at 1.163v = heavy throttle

This is what lead me to believe in an earlier post that my throttling was due to heat. Why does the same CPU push to higher volts by itself when it doesn't need those volts to sustain full load? Im really confused..The only answer I can come up with is that under full load the CPU automatically will go to its max voltage, regardless of it needing it or not (unless a program like RMclock tells it not to).

I don't find that very efficient. IDA is disabled in bios, so it cant be trying to overclock itself. Why would the CPU increase it's voltage for no performance gain or jump in speed bin? It doesn't make any sense..

 

I just called Dell and they said that they will send me a 130watt charger with the technician that is supposed to fix my sub woofer rattle. So as far as I am concerned, problem solved.

 

Accept that the current bios you have in your machine isn't designed to take advantage of the 130watt adapter. Also he's going to replace your speakers with new ones and you will still hear the rattle. It's a design flaw of the system, not bad speakers. Not trying to be mean, but problems not solved.

 

I have zero speaker rattle. I've maxed them out never a rattle. And you are correct unless Dell puts out a BIOS update to allow the machine to use the power the 130watt adapter can put out, it's worthless.

 

Not worthless, at least he has a free 130W adaptor, I doubt Dell Aus will even give me a free adaptor.