S-XPS 1645 AC Power Throttle Issue Investigation

Last thing I'm going to do is MORE beta testing for Dell, but I am enjoying all the people who are testing for throttling by playing a game like Crysis, which can be configured in a million different ways for graphics settings.

You can't play a game and test it. Theres too many variables. You may have your settings too low causing no throttle, or too high causing your GPU to struggle making you THINK its throttled.... Just run the tests that we've been running and that we know work please.

 

Well maybe but my E6500 is a business class machine bought by my business and it came with a 90W supply. Interestingly if you do a search for 130W power adapters on the Dell site there is a 130W that states specifically it is for the E6500 even though they ship a 90W with it.

 

I think going from silky smooth 30+ fps to crackling audio and less than 1fps while plugging in and unplugging the ac adapter on cysis warhead is a pretty good sign your throttling whatever settings you may be using...

Given there are reports of people bricking their systems by running extreme load tests I think the above is plenty enough evidence for me!

(I'm still on A01 by the way)

 

Zlog: I disagree. Crysis or any game is a great real world test and it's easy enough to configure ThrottleStop to monitor and log if there is any throttling going on while playing.

Everyone knows that an XPS 1645 can and does throttle in certain situations. The original Dell line was that users shouldn't see any throttling during normal use. Now we know that's not true.

The ThrottleStop log files posted yesterday while playing Crysis showed that the CPU performance on all 8 threads was being reduced by up to 90% at times due to Dell's throttling scheme. The GPU isn't being throttled due to one's choice of graphic settings. This is all Dell based throttling due to an inadequate 90 watt power adapter.

Furmark + Prime95 testing has clearly shown that these laptops will throttle when pushed hard. Now I think users need to test and find out if throttling is reducing the performance of their laptop during normal usage. Dell says no but the log file tells a different story, especially when gaming.

For users that never game, they might be quite happy with their XPS 1645 if it is able to run at its rated speed during their normal day to day tasks. Gamers on the other hand are going to feel like they were robbed if 90% of their computers performance is being sapped by Dell's poor choice to include a 90 watt power adapter.

 

Is it possible to do a fixed comparison test (with specific given graphic settings)? For example FPS-rate comparison on a CD2 (t9600 etc) versus the i7 just to prove the system will in some cases perform better with an inferior CPU?

 

The problem is that people are posting "YARP NO THROTTLE" threads without any monitoring. They're just eyeballing it. Then they go out to prove us wrong and they end up causing damage to their systems.

What I'm trying to get at is that unless people post some TS logs to back it up, then we have no way of knowing if its throttle or just the system not being fast enough based on the settings config.

On another note, got my return request escalated to customer relations

 

Bios A05 releases by a engineers of Dell France for resolve the throttling issue :

http://www.lesdelliens.com/upload/Vinky/Tuto/1645_A05.EXE

But the bios doesn't resolve the throttling. I think, it's the same bios of X29

 

I see what you're saying now Zlog. Anyone that claims that their laptop doesn't throttle needs to include what applications they have tested and include a link to a ThrottleStop log to back that up.

So far, with a 90 watt adapter, they all throttle if you load them hard enough. There's nothing left to argue about. That has been proven beyond any doubt.

I was contacted the other day by a user with a business class system and it throttles like crazy too, even though it comes with a much more powerful adapter. Throttling could be an issue with ALL Dell laptops that use a separate GPU and a Core i7 or even a Core 2 CPU.

Ichido: Yesterday's testing shows that a Core i7 can be so heavily throttled during a game that a Core 2 Duo, if it's not throttled, could easily outperform it as long as they were both using the same GPU.

It's hard to make fair comparisons because different combinations throttle different amounts. No matter what you buy I would suggest running a ThrottleStop log to make sure that you are getting the performance you paid for. As delivered from Dell with a 90 watt power adapter, the XPS 1645 with a Core i7 is being throttled so badly during some games that it is not anywhere near to living up to its full potential.

 

what is the release date of this bios?

 

11/01/2010.

I updated from the X29 to the A05 with no problem.

 

I checked it against the X29, and it is not binary the same. There are enough differences to say it's not the same BIOS. There are significantly fewer differences between X29 and A05 than between A05 and A03, so it appears the A05 is more related to X29.

 

Cool. Now can you give some numbers ?

 

I just got my 1645 two days ago and I was wondering whether anyone had a template message (explaining the whole shenanigans + a request for a 130w PSU) which i could use to send off to Dell.

Would be much appreciated.

 

And are there any release notes available? I just checked, and this BIOS is not (yet) available on the dell.fr website...

 

No available release note for the moment. The bios was send by an engineers of Dell France by email for fix the throttling issue.

So I think, it releases soon in the site of Dell.

But, I have the same result with bios X29 and A05. Same score 3DMark Vantage and always the throttling issue.

I saw no change in the A05 and X29.

 

Well Why would Dell France Release a BIOS [fix] ahead of everyone else? Seems like someone jumped the gun here and it doesn't seem to be a "fix"

 

That's too bad. Still, thanks for making the file available, and for testing it!

 

I think the BIOS is release by Dell US from the engineers and it release only in internal and not yet in Dell website.

 

Point was actually to show the engineer who is coming tomorrow to change my screen and m/board an example of an 'everyday task' throttling badly!

I think if you running a game on high settings smoothly on battery and and struggling on AC it does PROVE that the systems hardware can cope with those settings but is being held back when plugged in!

 

Is this tested with a 130W power adapter?

Regards the bios these are purposely being leaked, just like nVidia "leaks" drivers. They know there are people willing to be guinea pigs and test these bios. I can assure you there are Dell engineers watching these threads for feedback. This is the best way to test, as if you release to the general public you would either have bricked laptops, or useless data feedback.

On a side note, but you guys are seriously stressing the system when running Furmark AND Prime. These are laptops after all, sure they are meant to game but it's impossible to give desktop level cooling in such a small unit. I cannot see consistently benching can be doing any good to the machine. I have a high end desktop and certainly would not like to run these for hours, even though I overclock. Also I am not sure if still valid, but a while back when I had an ATI it would get crazy hot when running Furmark!

 

Just heard from my contact at Tom's. It looks like they are serious about learning more about the Dell throttling issues. Hopefully some media coverage will get this issue the attention it deserves. If anyone new shows up at this forum and asks a lot of questions; I don't want anyone here holding back what they really think.

 

I notice after even a few 1 minutes runs of furmark or prime95, that the "front wall" on my 1645, (the part just under the trackpad) gets pretty toasty. There is definitely a lot of heat getting trapped inside the case. If you have disassembled a 1645 you can see that the cooling system is made up of a heatsink+heatpipe+fan combo and the fan has an obvious intake and exhaust. Any heat that gets into other areas of the chasis is stuck there for a good long while. And it's heat above all else that kills computers, especially mechanical hard disks.

For this reason, I've always wondered why the heatsinks and heatpipes on these coolers are not insulated in laptops (except for the "radiator" part where the heat is supposed to leave). I noticed when I had my 1645 in pieces that those heatpipes get damned hot under load and as mentioned above the fan can't really get the resulting heat out of the chasis.

As a hobby experiment I plan to try to try to add some insulation to the heatsink+heatpipes. I figure if I can keep the heat from ending up in the dead zone inside the chasis, I can (maybe) increase the life of the computer (while probably voiding my warranty).

If it works really well, I'll share my story. If it doesn't I'll deny ever trying it.

 

Can someone with a 130/150W try it with and without ThrottleStop? Just to kill that little dream I have that Dell allowed us to use greater PSU. Thanks.

 

Just accidentally made myself a guinea pig. I'm used to the dell server bios updates where it will show the fixes before running (where you can cancel). Of course, run this and it just installs... At least it came back up. I had uninstalled Crysis but did try out Anno 1404. Still throttles using 90W. But one thing I did notice was that when I reduced the screen brightness, the system jumped back up pretty quick. Went from 75% throttle, multiplier 7 to 100% and multiplier ~9 (I'll save the log next time). And that was within seconds. Seems like in the past it took much longer to climb back up.

I know this isn't very scientific but if this is a prerelease or a beta bios, still doesn't fix the throttle issue with a 90W adapter.

 

I am an "intended" guinea pig: my expectations for this bios are far too high to wait. Although I don't have a 130W, only a 90W, i'd like to share a thought.

After installing the A05 bios I entered the bios configuration at startup. Under the "Advanced" menu I noticed, at the bottom of the page, two voices: Adapter Warnings and Charger Behavior. Both were enabled. Since I never entered the menu before, what I ask simply is: are those two voices new, or were they present in previous relases?

Also, I put again the kind request for people who has 130/150W, to post some results with and without ThrottleStop. Thanks.

 

just to let everyone know i have mirrored the two BIOS (A05) and (X29) - as well as latest ThrottleStop application. http://strags.com/1645 for a directory listing. Very fast mirror.

 

I don't know what the two 'voices' do but they are present in Bios A03 on my laptop.

 

Don't recommend applying some random BIOS update you found on the Internet, it's very easy to brick the machine.

 

Has anyone tested the x29 and A05 with 130W no throttlestop?

 

Yes, it runs wel.

 

Does it still throttle?

 

Runs well?

So does it still throttle?

 

What about the other manufacturers I7 models, such as the Sony F11, or the Lenovo W510 etc.? I suppose the only way to know for sure is to test during gaming with the tools available.

 

wbglider: The other manufacturers don't seem to be using 90 watt adapters so there isn't the need to resort to extreme throttling. Try heading to a thread for any laptop you're interested in and maybe your can find someone willing to do some testing for you. ThrottleStop works great for testing without changing anything on your computer. Just leave the top 3 boxes unchecked. When you download it and run it for the first time, that's how it will be setup so you don't have to worry about hurting anything.

Of course it does. An i7 + an ATI 4670 needs more than 90 watts when gaming. These laptops have to throttle when they are being used with a 90 watt adapter. No bios can create power out of thin air.

 

just glancing at envy 15 forums and they say that it throttles on battery not ac

 

I was asking if it still throttles on a 130W adapter , we all agree that 90W + new bios=no change.

 

Still doing some testing, got to find the video MHz monitoring program again.. err. Anyway.. So far this is what I have found.


With a 130W adaptor running both furmark and prime 95 I get the system to pull ~110W without TS enabled. So it looks like they upped the cap, on the downside, there was a change in idle wattage consumption:

32-38W consumption A03 BIOS
40-44W consumption A05 BIOS - never went under 40W

Obviously they upped the voltage on something...

The processor would not go passed 91W when maxed by itself. once the GPU was put into the test at the same time, the system pulled 110W nonstop but the processor then started to throttle, going as low as 9 multi with a clock modulation of 75% or higher, and it would pretty much stay around that.

TS enabled..... Was pulling around 139-140W but it was not constant; it would jump from 110W to 140W to 110W... I’m thinking its throttling the GPU.. GPU temps around 80-83C, i7 maxed at 79c

Going to further test and watch the graphics card...

 

This is good news. They are taking steps in the right direction I think it's capped around 110 maybe to avoid approaching the limit of a 130W adapter like they did with the 90W? Hopefully the final version would bring down the idle power consumption. thanks for posting these results!

 

The video card when stressed with the CPU is being throttled from 675mhz to 300 every 5 seconds or so... this isn’t a limitation of my 130W because even when lowering my brightness to min and maxing both, the system pulls 127W max... it looks like Dell is now actively throttling the GPU to manage wattage draw as well, this didn’t happen before.

I still need to test some games and see how dells new settings work..

 

Clearly, the 130W or even the 150W adapter won't be enough given the efficiency of the adapters. If you assume the laptop can pull 140W, as you see it can, then at max load a minimum ~155W adapter would be needed.

So basically on the initial tests you ran, the 150W adapter should be good enough for the majority of cases.

Maybe the GPU throttling you are seeing now is a result of a lack of energy available to all the components at high load, so the performance decreases sporadicallly (every 5s?) when now enough power is reaching the GPU.

 

Quick clarifying question: does this affect the laptop when the CPU is in high use or when the CPU + GPU is in high use?

The reason I'm asking is that I'm not a gamer, however the only reason I was going to purchase the XPS 16 was for video editing while traveling. If I'm rendering a video and it throttles down it's going to be virtually worthless.

Has anybody tried running a FPS comparison using a program like VirtualDub with/without ThrottleStop, 90w/130w/150w, the BIOS update, etc?

 

I've rendered video using Sony Vegas with no throttling but re-encoding HD video I saw some throttling using 130W w/ A03 BIOS

 

Interesting, thanks for the response. Was it actual throttling or possibly just a slow down due to the speed of the drive and the filesizes being dealt with?

 

First time poster, and I would like to thank everyone a ton for bringing all this to my attention as I bought a 1645 and would have known something was very wrong as soon as I tried to push it.

The 4670 could very well be throttling from furmark. AMD calls FurMark a "power virus"

Otho we can argue the same points we have against the 1645 throttling issues, unlike a CPU all corners of a GPU can be pushed because of the number of threads it can run. I don't think ATI ever considered such a condition would exist. Therefore when running furmark the GPU could very well be throttling.

In fact because of the software throttling on the pre 5xxx series cards from ATI you could potentially damage your card if the throttling doesn't kick in fast enough.

Just something to keep in mind. Not that I know of a better testing methodology, other then maybe finding the game that makes these GPUs run the hottest, setting Prime to run on 6 threads leaving 2 threads for gaming. This would find the worst case scenario for real world usage and still be a little unrealistic.

I'm concerned with the XPS 1645, what happens if a compute shader app comes out that can use the 4670 and 8 CPU threads, Isn't the point of buying something like this to be futurproof. Kinda defeats the point if its dropping to sub 500mhz performance plugged in.

I ordered a 1645 just before reading threw this, I've been keeping up on this thread with my emotions swinging both ways, I want to love my new computer. The RGB screen sounds brilliant. But I do want it to run at the specifications advertised. I could understand if it throttled a little bit in the situations provided ( given how unreal they can be) but the throttling is extreme and even bad under real world usage models even ones that don't include gaming ( like distributed computing apps).

A car company cannot sell you a V8 that pushes 300 HP, but have it rev limit way below the engine specs to save the under powered tranny they put in and maybe disable cylinders to boot. This is basically what dell is doing. It is simply not acceptable.

 

Does GPU-Z report core voltage for the GPU in these? I'm starting to think about a plan B. It might be time to look into adding some anti GPU throttling code to ThrottleStop.

It will be interesting to see during real world gaming if the GPU throttles or not with this new bios and if plan B needs to be pursued.

 

I know that probably no one will have, but is there anyone with a 210W adapter?

 

Hum, I runned Crysis yersteday and now for confirm, my kill-a-watt dosen't exed the 91W. Crysis in 1080p max brighness resume to CPU THROTTLING.

I tested with TrackMania Forever in 1080p, high and max brighness, when the game loads, I saw 70-75W and arrived in the game, the CPU THROTTLES and the kill-a-watt displays 65w. But with brightness set to the min, it works great.

130w PA-4E and I tested with the PA-3E (90w), same result.

EDIT: I tested with Prime95 + Furmark. When I start the stress test, my kill-a-watt displays 120W and 1 second after CPU THROTTLING (7x) with a variation between 80 and 91w

EDIT2: 2 others users on the LesDelliens forum has the same result of me. Very strange.

 

I have updated my BIOS package, and moved the info to its own thread.

This way we can keep track of all available BIOS versions.

 

According to Windows, without ThrottleStop and BIOS 05, the CPU throttled 34 times today in 4 hours.

Min throttle : 36 seconds
Max throttle 182 seconds

AC 130w

Dell will never solve the problem and it's for that Dell called me back for change my SXPS 1645 to the Alienware M15x for free.

 

guys,

I need a favor. I just received my studio 1645. Strangely, it came with the old A01 bios.

Now for the favor, all our laptops have a serial number which can be found below the battery.. you'll find it by removing the battery. My serial starts with CN-.. ends with A00. I was wondering if that is normal or does anyone have a serial ending with a greater hardware/bios revision, like A01 , A03 etc..?

I am afraid that I might have received a user's returned system lol :S

Please help! Thanks !

--------------------------------------------
my location: Malaysia.
specs: core i7 720qm, 4GB ram, 16" RGBLED, Intel Wifi 5300.