S-XPS 1645 AC Power Throttle Issue Investigation

Just so everyone knows, the main problem was originally due to the fact that because the video card was unable to ramp up and receive enough wattage when playing games (or doing other things for that matter) due to the 90W intake limit set by the BIOS's i7 Clock Modulation, it would throttle the processor to allow the video card to obtain more power (from the 90W BIOS limit).

Since this wonderful little program disabled Clock Modulation, the BIOS has no other means to modulate the wattage intake and you can push it to its limit. Please be warned, if you do not have a 130W adaptor, you will more than likely blow your 90W adaptor up. You need to make sure your running an adaptor that can handle the higher intake or watts.

That being said, when playing games such as Dragon Age, Borderlands, COD Modern Warfare 2, the system only pulls around 100-110 watts with an external mouse/keyboard, max RGBLED brightness, and Wi-Fi/Bluetooth on and I never see the processor temps go above 67c I do however see the video card top out at around 81c.

All Dell needs to do is code a Clock Modulation table into its BIOS to work with a 130W adaptor when detected so we do not have to disable it with this tool, you should not need a system board change or anything, just a simple BIOS update.

 

I believe this issue with the keyboard is something dell will also need to fix, for some reason when using it, it kills the system, I’m very curious if this whole speed issue will not be present if you just play with an external keyboard. I will need to do more testing later today to see if it makes a difference or not.

In regards to the previous posts asking if an external keyboard is needed, I would assume yes, since the games run like crap on the laptops keyboard, this leads me to believe the keyboard was always an issue, and the i7's clock modulation was just masking it, now with it disabled its showing its true face.

 

Its a power issue, the keyboard doesnt lock down to 10fps for me under normal use/battery. Battery plays fine with/without external keyboard. Even when not pressing keys fps are far better with throttlestop enabled. Its when i enable throttlestop is when issues happen. But when my computer comes out of sleep sometimes the actuall keyboard works fine with throttlestop, wierd i know

As for technical doc, Seeing as we have undeniable proof this laptop requires more then the 90w to perform, as we can disable their active modulation and physically measure, i dont think a tech writeup is needed. We just need to make sure dell realizes we wont put up with crap performance for their aesthetic choice of a 90w adapter.

 

Out of curiosity, why do some of you think that Dell will address the throttling issue specifically tomorrow?

 

TurboBoost also works with four cores stressed. So an i7-720QM can run all four cores at 1.73GHz (13x) if they are fully stressed. At one time, I could hit 13x constant by turning EIST off in the i7 Turbo tool with Prime95. But now I'm back to 12x again. Not sure why it worked that time, but I can't replicate it again.

@Zlog

I'll forward the PM to you.

@all

I have only noticed some flickering on the media control thing, but I have to really look for it and it's pretty intermittent.

@n3verm0re

Someone said a executive Dell dude or someone told them that they would address the issue on D2D tomorrow.

~Ibrahim~

 

How about having throttlestop enabled while on battery? Does the keyboard still have issues, I did some testing and found that even on battery, if you fully stress the system it will not max both the video and processor as much as you can max it on the ac adaptor.

I believe I watched the processor lock at a specific multiplier or something when on battery with the video maxed.

What I want to know is, do games still have performance issues when not using throtlestop but using an external keyboard while on battery? it might not feel or look like it while testing, you will have to watch your FPS in a game and see if it goes down any while pressing your keys, Fraps or something, this is what I’m going to try later today.

If it was a power issue, it would do it on the external keyboard as well, I’m trying to isolate the issue with the internal keyboard, if a bios updated can fix it or if it’s just crap.

 

Oh i think i mis understood your previous post as you where talking strictly the keyboard issue. I havnt noticed any slowdown while playing battery at all. but your right i dont think its using the computer to its fullest abilities either when under battery, at least not 100%. As gpu load and cpu load dont seem to be as high compared to ac+throttlestop. Valve games, like tf2, it is very easy to monitor fps while playing. Ill try to investigate more later tonight.

The wierdest thing is with all the monitor utilities going, it shows nothing. The mult is still high etc. I wonder maybe if its halting processes somehow like an interrupt when the key is pressed. anyone smarter then me know if theres a way to monitor devices like the keyboard better?

 

Keyboards have the lowest reserved I/O for a system, meaning if your computer is doing a lot of crap and locked up, you’re still at times able to do a ctr-alt-del or whatever and terminate processes. For whatever reason, the internal dell keyboard is taking the laptop to its knees. It could be the way this keyboard is designed or a bios update may fix the issue, it could also be the way the system board operates the port the KB is connected to separately than a USB port.

It could be that the internal keyboard is using a PS/2 interface and a USB keyboard works better based on how the systemboard interprets signals from its PS/2 interface.

 

Hopefully we see something promising tomorrow =]

 

Just don't hold your breath

 

I was going to say the same thing, lets hope Dell proves us to be wrong this time.

 

*crosses fingers*

They have a huge chance to save face, with what seems a rather easy fix. Seeing our experiences and as that is just common sense, makes me a bit nervous.

 

Except it's going to cost Dell a pretty penny.

There will have to be a PSU exchange program setup and this time of year is gonna hurt them BIG time.

 

If unclewebb could come up with a good fix/workaround in such a short amount of time I don't wee why DELL cannot.... sigh....

Dell..I am Disappoint

 

Maybe unclewebb should work for Dell and make big money

 

maybe he already makes big money

 

I also experience sporadic flickering, noticable every 10-20 seconds usually. Had my RGB lcd replaced once, but it's still present, so I'm hoping it's driver related. Ofcourse the replacement lcd had to have two dead pixels, three short of a free replacement, meh...

 

Go to:
http://www.jscreenfix.com/

Try launching the webapp and see it revives your pixels. Maybe they are just stuck this could help *shrugs* . I can't vouch for it...but give it a shot

 

What makes you think Dell would schedule anything tomorrow? 1 user mentioned something without providing any details. The 15th has been mentioned in passing by Dell that they would have a BIOS update hopefully by then. They released one recently that I though covered that statement.

Are they supposed to issue a statement, software patch, etc? Don't get your hope up, I doubt tomorrow will be any different that today.

 

Unclewebb, Thanks for the hard work and the temporary fix. I really like my 1645 and didn't want to return it. This may just allow me to play with it until Dell fixes the issue. I cannot PM yet but will as sound as I get to 15 posts.

 

I've been talking with a Dell tech rep since last week. He actually called me up to tell me that the engineers had solved the problem and to update my BIOS. He failed to tell me what version. The site lists the A03. I updated to it 5 days ago, and the power supply still gets hot while gaming. Does anyone know the details of the A03 update? I'm going to quiz him on what version he's speaking about because that was out before I ever talked to him.

 

A03 doesn't fix anything related to throttling.

 

I wish. Unfortunately writing free software doesn't pay the bills. After half a million plus downloads of the last version of RealTemp, I had made less than $200. I'm coming to the end of my free software days. I can't afford it. I am glad that I was able to help you guys out and make your laptops usable.

The way some people talk it's like they think Dell doesn't understand this problem. They are not that dumb. They know exactly what's going on with these laptops and they know exactly what's going on with the Latitude E6x00 series. They designed these things to throttle like crazy to stay within the power envelope of the pathetic 90 watt adapters they shipped with these laptops. A bios update and a minimum 130 watt adapter is needed. A 150 watt adapter would be more appropriate so you'd have some more head room and then your battery could recharge while you're playing a game.

Will Dell do a massive re-call to keep a very small percentage of their customers happy? I don't know, maybe. It's sad but most users buy something like this and are happy just seeing the i7 sticker. That's fast enough and as long as it doesn't burn through their lap, they're happy. A bios update that is going to cost them a huge amount of money and is going to create more heat in their laptops isn't going to make everyone happy. It will make most everyone on this forum happy but lots of users want fast, cool and quiet with cool and quiet being more important than fast.

Anyway, ikjadoon is correct. A Core i7-720 is physically capable of using turbo boost even when all 8 threads are fully loaded. The maximum multiplier when fully loaded is the base multiplier of 12 + 1 for turbo boost for a total of 13.

The Intel docs say that turbo boost will be limited by core temperature and current or power consumption. When you try to fully load all 8 threads, that is the wall you'll run into. It will likely be impossible to get any turbo boost when you run a high stress program like LinX or IBT with all 8 threads fully loaded. There are other programs that can fully load 8 threads but the type of load is much more gentle so in that case, then the CPU will be able to use turbo boost when fully loaded. I wrote a program called Load Tester that lets you gently load your cores so you can test for stuff like this. It's included in the RealTemp download.

http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

It's a variable load tester so you can gradually increase the load when testing. It is only capable of fully loading one thread so you'll have to run 8 instances of it if you want to fully load your CPU. It won't create the heat or power consumption that a typical load testing program will create. This program is more designed for multiplier / EIST testing purposes.

The i7 Turbo program will give you a very accurate look at what each thread is up to. The multiplier on these new CPUs is completely dynamic and can be constantly changing based on load.

max420 sent me a good example of this.



He is doing some encoding and the overall CPU Usage is only at 58% but his i7-820 is not able to maintain full turbo boost. There's nothing ThrottleStop can do because he is running into the physical limitations of this CPU. An i7-820 should be able to run with a maximum 15X multiplier when all 8 threads are fully loaded but the Intel Turbo White Paper says that turbo boost is ultimately limited by 3 factors:

1.2 Dependencies / Algorithm

Intel® Turbo Boost technology core frequency upside availability is ultimately constrained by power delivery limits, but within those constraints, it is limited by the following factors:

• The estimated current consumption of the processor
• The estimated power consumption of the processor
• The temperature of the processor
---------------------------------------------------------

When you exceed the limits you run into what I call turbo throttling. That is when the multiplier will constantly bounce back and forth from 13 to 15 on an i7-820 as turbo boost is engaged and disengaged rapidly hundreds of times a second. He's averaging a multiplier of about 14.7 which isn't too bad. i7 Turbo uses the method outlined in Intel's paper and is able to very accurately determine the average multiplier when this is happening using high performance timers within the CPU that were specifically designed for this monitoring purpose. Most software will sample your multiplier once per second and not be able to give an accurate look at what's really going on.

If you are doing some encoding like this and you see the average multiplier bouncing around at less than the default maximum, you can open up a program like RealTemp and set clock modulation to 87.5%. This will reduce the internal load on the CPU, will reduce the temperature and power consumption of the CPU and you should see the multiplier instantly go up to 15X. It's now operating within the Intel thermal / power limits so the full turbo boost is once again available. Doing this will likely give you less overall performance but it will at least show you what is holding you back. Remember to disable the ThrottleStop program before doing any testing like this.

Ibuprofen: If anyone needs ThrottleStop and can't PM me then send an email to the address in the RealTemp about box. I don't like to post that address in forums to help cut down on the spam.

 

LionelatDell has info on Twitter- there will be info on the Direct2Dell blog "by Tuesday" . Maybe we'll get something today.

 

-[Deleted]-

 

Thanks for the update.

 

Thanks. Will do! I was hoping not to make unnecessary posts

 

I read the tweet from Lionel.

Sorry to say but he said Tuesday next week, not tomorrow.

 

While it's somewhat pedantic to argue such a detail, I interpret Tuesday the 15th as being next Tuesday when posted on Saturday the 12th.

That said, I don't expect Dell to do anything tomorrow.

 

Yeah, I am just setting myself up so I am not disappointed

lol

 

Thanks. That was the most coherent and evidence-based post on this issue yet.

 

jbartlettjr: Thanks. A large part of the problem is users don't fully understand how these CPUs are designed to work. These are not Core i7 Desktop CPUs and work quite differently than them.

No. I was just trying to scare some people into making a few more donations to the cause. I don't mind writing free software but I need to eat too.

I have provided ThrottleStop to anyone that has asked. I might wait for the big Dell announcement before officially releasing it.

 

Is there a specific number of replies I have to post to get the opportunity to send a PM? I am really anxious to test ThrottleStop, but there is no sendPM-button.

 

Send an email to the address in the RealTemp About... box and I will send you a copy of it as soon as possible.

 

I feel the turbo boost never turn on with 8 thread even on battery.
I use intel desktop widget to watch turbo boost.
It is only occasionally on when overall cpu usage below 50%. I have a 720 cpu.

 

I think you need 15 posts to do this. I would not advise making non-sensical posts to reach this post count.

 

Try i7 Turbo. It's not quite as beautiful as an Intel widget but it's a little more precise.

As I mentioned above, you might not get any turbo boost on these CPUs when all 8 threads are fully loaded. It depends on the type of load you're trying to run. That's just the way these things are designed by Intel. They have to operate within a maximum power limit and no software can push them beyond that.

 

Thanks. I will not post until I`ve got something more or less important to say.
To be honest, my fellow 1645 owners here in germany don`t seem to be aware of this problem. Didn`t find a single thread about the throttleing issue; at least none taking it serious. Only a review of the new Studio 17 with the i7 mentioning throttleing while using CPU+GPU at once.

So I`m really thankfull for the work you people do.

BTW, "non-sensical" is a pretty good description for my chat with the technical support I had last week, which at least got me a 130-W adapter and the advice to send my 1645 back for good, after the funny guy verified the throttle issue with his private 1645, but refused to call it a problem.

 

I think a lot of users have no idea about this yet. I'm a computer enthusiast and I didn't know squat about this until a couple of days ago. Thumbs up to tinkerdude for really getting the ball rolling on this subject. When I saw how much effort he put in, I was motivated to do the same to help out users.

Hopefully the latest version of RealTemp that clearly shows Clock Modulation might bring this problem to the attention of more people.

http://forum.notebookreview.com/showpost.php?p=5620844&postcount=1149

No one wants to spend good money on a laptop that in actual fact can't deliver the goods because it spends too much time throttling.

 

In your opinion, are the Core i7 Mobile CPU's more powerful than the most powerful last gen Core2Duo Mobile CPU's?

I.e. Do you get better real world performance out of the i7 720 (lowest end right now) than you would out of the highest clocked C2D?

 

It's significantly faster than my Q6600 at encoding video.

 

How many thread do you use? I am wondering will it be better to use less thread so turbo boost can be turned on.

 

8 threads, I don't think turbo boost is any substitute for software that takes good advantage of mulitple processors.

 

Thank you for the link, didn't help though The pixels are dead, not just stuck unfortunately.

 

Ok, i feel okish. ok so, ok umm... right ... ok so umm i had a day to play with this laptop. the keys feel off and the keyboard uses more data than a normal keyboard should be acustomed to be useing... maybe from the data that shows when its active to decide when to turn the backlit keyboard on. When the A/C is hot or scorching it will not charge anything to save its own life. that being said it is going in the blender. ok not really but still. After playing L4D on SUPER DUPER HIGHEST like x8 AA and like x16 tri lsogaorga some next thing that supposedly makes things more nicer. I can say that this laptop gives me blue balls. Apart from the AMAZING SUPER COOL RGB screen and almost giving me and several other people sezures from the light show i put up this laptop is bad in my opinion. when playing l4d noticing lag as soon as i jumped into the game. LAGG!!! then i dimmed the computer ever so slightly until it was playable on minimum barely. then i umpluged it and it ran like a beast on minimum. throttling occurs on max even running battery. i mean this laptop is able to handle left 4 dead easily with a 4670 and the 820qm should only be helping. not helping and my cpu is saying screw you to my gpu and my gpu is taking it out on my fps which is eating my eyes and slapping the sound and the sound i making my ears go plop. the trackpad does stupid things in firefox that i dont like, like the tap on the scroll bar once then tap + drag and it sticks. why on earth would i want something to stick to the thing. so far the only thing keeping this laptop from the blender is the screen. amazing screen. btw i cant even watch youtube while the charger is hottish and charging my 98% battery. rediculous. so it is a power problem that the millionair figured out how to make better.i expected this thing to be good out of the box for general use. i cant watch youtube while charging? i put the power brick in a ziplock then in the freezer for 2 hours and i take it out and guess what! its still warm to the touch. . 10 more days of playing with it then its off to dell. i dont see dell fixing this because of the -rep and the -cash and stuff but they will say. FORGET ABOUT IT THIS TIME, NEXT YEAR WE WILL MAKE A LAPTOP THAT WORKS PROPERLY SO DONT WORRY CEO something something, (said 92 times in the previous week). i honestly think they will try to convince them that this will be fixed when they release a new laptop sporting the new i7 in q3 of 2010. til then i will hunt my but down for a RGBLED. did you know they have rgbled tvs? lol it cant even play a 2008 game lol lol lol lol lol

 

you have a 130W charger? and you used thottlestop?

 

AND..... That is why we don't smoke crack, people. lol

Strange but amusing post, razrkillr.

 

Razrkillr, you should call up Dell and tell them exactly what you just told us, and please record the conversation for the thread, lol.

 

Just an FYI from my side. My contact at dell sent me an email today that said the following regarding the issue:

"130W is not designed for this system. Using this will cause thermal issues. The group has been able to duplicate the issue and are going to keep me up-to-date as to the resolution."

John

 

Yet many people have been using 130W adapters + throttle stop and haven't reported any major thermal issues.