S-XPS 1645 Throttling Info. and Updates

I know this is no shock to anyone, but I just want to say that my brand new XPS 1645 also throttles. I had somehow convinced myself that my system (which arrived yesterday) would ship with a bigger ac adapter (it didn't) or that Dell would have fixed the mistake on all new machines going forward (it didn't).

Loaded up Fallout 3 and it was a slideshow on ac and the sound got all garbled, but when I unplugged it, it was smooth as silk. I'm well within my 30 days, so I'm going to buy an Envy 15 today and return the XPS 1645 when the HP system arrives.

 

hahaha, I love suggestion number 5

 

It's 21 days.

 

Try using mobility modder and install the latest ATI desktop version.

 

Goto http://strags.com/ for 8.683.2 - latest from Dell that install without modification. They're high performance and stable.

 

I have so much time on my hands you guys give me a good paragraph or 2 on what you would like me to say and I will spread that like wild fire!!!!!Just give all the info and we can do this.Lets burn dell to the floor!

 

I have started this in one of the biggest laptop related forums here in Bulgaria. Do it in your countries as well.

Nikolay

 

I'm past 21 days is there a chance i could get a refund?

 

I was just thinking you have a 1 year warranty tell them its broken and tell them what happens about the throttling see what they say.In reality its broken so by their warranty they need to fix it or replace it,So if they replace it and it does the same thing call back and say hey this one is broken too

 

Yes, I have had my 1645 for over 2 months and dell has approved me for a full refund.

 

Does the throttling issue only affect applications (i.e., games) that take full advantage of the 3D capabilities of the GPU? If I were to get one of these and use it mostly for development, photo editing (photoshop -- mostly 2D), and perhaps some light gaming (e.g., flight sim), would I experience the throttling in these uses as well?

I get a decent Dell discount through my employer, but don't want to waste my money obviously.

Thanks for the help.

 

Did you file a complaint before the 21 days? If not what department did you speak with and what did you say to them? Can you provide the name of the person you spoke with?

Thanks

 

My 820QM with RGBLED throttles running the Prime95 blended tests and that doesn't exercise the GPU. Not ridiculous mind you, it goes from 13x (where it should be) down to 11-12x multiplier. C0% stays at 100.

More worrisome (for Dell owners) is the trend towards using the GPU for general computing tasks. What happens if your photo editing software wants to offload some of the repetitive tasks to the gpu while using the cpu at the same time? Its possible in that scenario that you would get very significant throttling like a gamer would. That's the risk of holding onto this thing in the long term IMO (unless you run TS).

 

Good points. That sucks. Guess I'll be buying something other than a Dell, unless this gets fixed by Dell in the next week.

 

Intel® 64 and IA-32 Architectures Software Developer’s Manual
Volume 3B: System Programming Guide, Part 2

http://www.intel.com/Assets/PDF/manual/253669.pdf

Appendix B has a listing of the Model Specific Registers. MSR 0x198, MSR 0x199 and MSR 0x19A are the 3 important ones for controlling clock modulation and multiplier throttling.

TomK2: You spend extra money upgrading to the i7-820 and it performs slower than an i7-720 is supposed to run at. Time to ask for a refund of the difference or tell them to shove the whole damn thing.

 

Yes I had several emails into dell starting at about 2 weeks after I purchased the machine and the email address I use is:

[email protected]

Good Luck

 

So I'm talking to the rep who called me (from the michael_dell email) and here's what they're saying:

"I have spoken with three other customers about this issue and each one said that the problem was resolved with A06."

When I told him that it was not resolved with mine, he said that was strange and put me back on hold.

More of that sweet sweet sax!
http://www.youtube.com/watch?v=NX_eHKqDwZ0

 

OK, sax is over.

The call concluded with the rep telling me that he was sorry that the problem was not resolved with my machine, and "per our agreement" (my email?), a service representative will be calling me in 1-2 business days to arrange for the laptop to be returned.


Well, that's saddening. I really really want to keep this laptop, but I also don't want to keep a machine that isn't doing what I paid for. I guess it's the only way I'll be able to get anything through to Dell

Time to scour the forums for a good replacement.

 

OK, so my little Linux program now seems to be setting IA32_PERF_CTL and IA32_CLOCK_MODULATION msrs correctly, to fix the multipliers (to 13) and disable the software clock modulation. I've got this on a loop with a small delay, and the values I read back suggest that it is working.

However, my meter tells me that the power draw is still capped at 90W, and when I pull the plug the increased temps and fan speed indicate that it's drawing more from the battery.

So I'm wondering if there is some other throttling mechanism at play here. Does anyone know if bios A01 did things differently to A03? Is it possible that "Chipset Clock Modulation" is being used? I'm not even sure how I check for that, let alone disable it.

I've thought about updating to A03, but that is not so easy for me, since Dell's updater is Windows-only, and I haven't managed to get Windows installed on a secondary disk yet. Also, I'm not sure whether the throttling is worse on A03 than A01.

I'd like to PM Uncleweb with my questions, but I can't seem to find the button/link to do this . Could someone point me in the right direction?

 

Its another hello from our friendly insider Todd over at dell forums.

http://en.community.dell.com/forums/p/19306277/19648030.aspx#19648030

Re: DELL Studio XPS 16 1645 i7 power problem
03 Feb 2010 12:46PM

I'm here. I'm confirming what the A06 bios fixes, but from what I understand it is mostly directed at issues with iPhone sync'ing. There really hasn't been any news to come along since the update we did at the end of last week. At the time I said it was going to take a bit of time and unfortunately that still holds true.

Brad and I have a conference call tomorrow night with some of the engineers working on this issue to try get more details on where exactly the process is, and hopefully more information that we can share with everyone.

Don't bother Todd.

 

You need to have 10 posts before you can use the PM system.

If you start forcing the multiplier but don't take care of Clock Modulation then the CPU will start using clock modulation to keep you under the 90 watt barrier. You need to handle both of these.

Monitor MSR 0x19A. The simple thing to do is write a zero to that register over and over again to keep clock modulation disabled. The "by the book" way is that if you read that register and it shows a number like 0x18, the 1 indicates that clock modulation is being used so you just change that single bit back to a zero and write it back to that register. Writing a zero is simpler and seems to work just as well. With this method you don't even have to read the MSR first which is more efficient. Just write 0 all the time and there will be a lot less clock modulation.

When testing, be careful when switching to battery power when at full load. More than one battery has taken a beating when doing that.

Edit: Chipset clock modulation is not being used on the XPS 1645 series. It's being used on the Alienware M15x as well as quite a few Acer laptops. You shouldn't have to worry about this.

 

Thanks, but I'm already correcting the clock modulation via 0x19A. In fact, that is the first thing I managed to do. Before I did that, I did see clock modulation kicking-in under load.

And the multipliers are all at 13 (according to 0x198). Still, it won't draw more than 90W.

Did you ever try Throttlestop on a 1645 with bios A01, or just with A03? I'm just wondering whether that version uses a slightly different scheme.

Incidentally, how do you go about detecting chipset clock modulation?

 

Well unfortuantly I ordered a 1645 on Jan 5 before reading this with a EDD of Feb 1. The day came and went and after an email they informed me that it MABEY will be Feb 26. So I emailed them back to ask if the delay is for time to fix the throttling issue and if it was I would wait otherwise cancell the order. They replied saying the delay is because of parts shortages and I CANT cancell the order. Well, if and when it shows up, I'll dig as far as the PSU and if it's still a 90 watt I'll tape the box back up and return it. I've been a customer of there products for over a decade, but I think it's time for a change.

flynte

 

Guys, for messages like this, or if you ever write to any Media regarding this issue, you will have to do it on the Dell Community Thread.

I wrote some angry threads today and got people going, then Dell_Todd shows up and confirmed that he will have a conference call with Engineers tomorrow night.

Seriously, we need to make this bigger, and get more media pressure on this. Any input will help all of us, poor xps 1645 owners.

 

What sort of load are you using? You have to be working the CPU and GPU such as when gaming to draw more than 90 watts. I don't think Prime95 Small FFTs is enough of a load if the GPU is not also being used.

Unfortunately I don't have, never had and never will have an XPS 1645 let alone any new Dell laptop, ever again. You'll have to find another user who hasn't returned his laptop yet to do a comparison between A01 and A03.

Chipset clock modulation is a little more hidden and I don't have any documentation for that. My program has to look in the Windows registry for the correct port location to read and write data to. It works exactly the same as regular clock modulation once you know the correct port. I believe for the XPS 1645 it is port 0x410. All of the Intel chipsets I've seen so far are either 0x410 or 0x810 since the Pentium II era. I use the ReadIoPortByte() and WriteIoPortByte() functions in the WinRing0 library to get access to this port. This is the first time I've used those two functions so my understanding of them isn't that great. I just know it works.

With the XPS 1645, you shouldn't have to worry about that. All laptops I've seen so far use one clock modulation method or the other but not both.

 

Actually web, at full brightness with the 1645 wled and prime95 blended test the system would throttle to 10-11x for me. Lower the brightness and it no longer throttled and remained locked at 12x multiplier. Nothing like the 12.9x a proper laptop should run at. This was on the A03 bios.

Wattage from the wall from just prime95 is about 90-94 watts. Same with the envy i have as well.

 

have a look at this people is this correct or am i having throttle issues?!?!?!

 

I got another call back from the rep to get some additional info about the return.

I asked him if there were other models similar to this one that did not exhibit the same throttling issue. He replied with, "No, all of our similar models currently have a similar problem."

Then they offered me a Studio 15.

 

If they cant deliver the laptop within 30 days, you can and have a right to cancel the order. so, if you dont get it by Feb.5th you can cancel. Call them on it. (the reason he wouldn't let you cancel earlier was not because of parts shortage, but you are not at 30 days yet).

 

Mark this day on your calendar. Finally a Dell rep that was honest. All of our laptops throttle and run slower than they should. I knew that but I didn't think they'd ever admit it.

sunny1985: Your CPU is running as it should. Background tasks always wake up cores continuously so you will never see the full 21 times multiplier that these CPUs are capable of. It only runs at 21X for a few milliseconds here and there so the average will always be less unless you can disable 3 cores in the bios or maybe disable 3 cores using msconfig.

RealTemp GT is designed for the upcoming 6 core CPUs. Try running RealTemp.exe instead to see all 4 cores. If you have any problems use the RealTemp.ini option:

NoATIGPU=1

For some reason the ATI code is flaky on some computers. It works fine on my computer so I haven't been motivated enough to fix ATI's code.

 

OK, well I'm using 2 instances of the mprime torture test to load it up. I've also tried running OpenQuake arena on max settings in addition to those, and on max screen brightness with RGBLED screen, and it just hovers around 90W. So either I've programmed something wrong, or there's some other throttling going on. I'm 99.9% certain that it is using 0x19A clock modulation, and multiplier throttling, but I think my program is fighting these, and yet it still throttles.

Thanks again for the info. I'm going to look into this 0x410 port thing and see where I get to. Presumably, if I read non-zero from that then there is some throttling taking place. I'm guessing Linux will expose these ports as a special file in the /dev fs, like it does for the MSRs. We shall see.

Just a thought: could the bios be issuing HLT instructions and forcing the CPU into C1? Have you seen that before?

Failing that, I guess I will have to find a way to install Windows without nuking my Linux install, and see what Throttlestop tells me.

 

If we could get that quote from "the highest levels", maybe that would finally be enough to get some media attention.

 

No, the "highest levels" say that A06 fixes the throttling issue, despite the Dell forum guy telling us otherwise.

Then again, who am I to say that anyone one of these people are correct?

 

It's almost so bad it's funny!!!

On a side note I'm on A01 (which was supplied with the laptop) at the moment. Would people recommend I update to A03? I'm using TS with 130W adapter no problem at the moment. I'm gonna steer clear for A06 given the recent reports on this thread....

 

I wouldn't recommend A06. I think A03 with TS works just fine.

 

I can vouch for that, they work better than the mobility modded ones

 

More inspired talk from Todd at Dell


DELL-Todd S
Dell Community Blog Admin
Joined on 04/10/2008
Posts: 411
Points 3,940
Re: DELL Studio XPS 16 1645 i7 power problem
03 Feb 2010 02:57PM

Chichiman:



It would be really great if you could get the answers to some of the questions below in your conference call tonight:

1. When will this new BIOS be released! -- if you give a date you need to keep it!

2. Would we get a 130W AC Adapter with this new BIOS to truly fixed the problem.

3. If this estimated date is more than 2 weeks, we need writen confirmation on this thread saying all of us with this Thortlle issue can return our computers after 21 days due to the lack of response form the Dell Engineer Team.

I hope that we will hear some great news from you tomorrow. Thank you very much for your follow up and understanding.​

First, the meeting isn't until tomorrow night so there won't be any updates tomorrow. At the earliest it would be Friday some time and quite possibly not until early next week.

We haven't been giving deadlines on any of these posts for the simple reason that you stated. We don't know when any BIOS revisions will be released nor do we know for sure what the final decisions will be. I've said it before, but it probably bears repeating since its buried in the thread, determining the correct way to rewrite the BIOS (if that is the path decided upon) isn't the quickest process. Writing code, especially such base code that has to interact with various hardware components, takes time and patience. And testing. Rushing out a BIOS revision that ultimately creates other problems isn't a satisfactory answer for anyone.

 

what I felt from my experience last night is that the A06 makes th machine much hotter. The fans are activated much much more often.

I tested two games with 90W AC adapter plugged in.

Residential Evil 5 at highest setting:
A03 version: Multi 7.0; FPS 30.5; system ranking by the game: B
A06 version: Multi 8.8 (Avg); FPS 27; system ranking by the game: C
therefore, I guess now the GPU is taking less power and the CPU is getting more.

Mass Effect 2 at highest setting:
A03 version: Multi 7.0; FPS (i don't know how to see it in the game)
A06 version: Multi 9.3; FPS (i don't know again, but the games seems much faster, however, i only played for 3 minutes.)

But Dell_Todd on the Dell forum confirms that the A06 only fixed some problem with Iphone.


finally, I have a question for all of you...I asked couple times, but nobody answered...how long does your battery last? Mine only lasts about 1h50m with 50% brightness while browsing the web and watch movies. When I play games, it would last less than 1h25m...Is this even normal? My configuration is i7-720, 6g, RBG screen

 

Your battery is normal.

 

Sxps 16 is weakest laptop i have ever had.

It lags lags lags in COD6 multi players map, and it also takes about 1-2 min load in game , when my other laptop only has t6400 3G ram, it taks 20s load in


I can't believe why people rank this laptop as 5 stars in dell.com

Mostly , it's....defetive or junk . A laptop carries core i 7 but 10 times slower than a 1 and half year old laptop

 

Hi guys off on a side not here but I just wanted to say thank you to everyone who gave me advice over the last couple of weeks, I have requested someone come ot pick up my XPS next week and hofully money will be back in my account by same time nxt week. I am moving towards the sager now as it doesnt have any issues with throtteling battery or otherwise.

I will try to keep my 2 cents in were ever possible with the threads, good look to all concerned! if anyone need to return theirs after the 21 dyas it IS possible I have written proof if you nee it.

Thanks again, Josh

 

Most casual users are delusional or else Dell sales reps must spend their spare time writing the reviews. In most forums you see a wide variety of writing abilities but the Dell reviews show a very consistent style. Coincidence I'm sure.

Thanks for telling it like it is. I don't know why anyone would accept an unfinished product with a known defect. On the 1647 forums there is severe throttling but the game is still playable so it's all good. It's all good until you get a new game next month or next year that is completely unplayable due to throttling.

 

Hi Fluffyuk,

I will take that written proof. can you please PM me with it?
Thanks,

 

Hi guys, I'm getting rid of my XPS16 (I7-720, 4GB, 500GB/7200, 1080p RGBLED, HD4670) tomorrow.

I fought some days with dell about an official power supply-replacement with 130W (I have one and it works very well, but I wanted them (dell) to move their a**), but in the end the technical guy won over the sales guy.

sales guy was that far to send me one for free, but technical guy said no, it will destroy the device (he didn't say, it *can* or maybe, he said, it *will* ... huh?). I don't think it would kill the device and sales guy said, doesn't matter what the technical guy says, he'll send me one and I asked about my guarantee. He said, he would also give me an email, for my guarantee won't be affected, but he has to ask the boss before … and the boss said no.

Well, today was my last day for free return and now it will be fetched by UPS tomorrow.

Next month here in Germany is the CEBIT-exhibition in Hannover and I'll be there on the first day, 02. March, and I will look for Dell, catching a technician for a serious talk. Perhaps they'll have some new devices there?

 

Hi Fluffyuk, could you send me the written proof as well^

Thank you so much.

 

Sure thing mate I just need to edit it and blank my name, Ill get it to you tomorrow or tonight if its as easy as I think.

Will they not let you treturn it ?

 

Thanks Josh.

I have not tried returning yet and it I might have more luck with the written agreement they gave you.

 

@fluffyuk: I'd also appreciate the written proof. Thanks in advance!

 

Actually the A06 Bios release was for the Iphone syncing issues with the P55 motherboards. Something regarding how the USB ports communicate with the system.

Guess we will have to wait for the next release and the one after that n the one after that.....

 

I'll take a copy of your written proof if thats ok. PM me.

Thanks