S-XPS 1645 AC Power Throttle Issue Investigation

Ah ok,
so let's hope they find the bios update too for the 1747 and ship with a 130w adapter.

 

Since you're jere giving out some news, which we all appreciate, could you elaborate on that a bit:

- Which thresholds are you talking about. The ones for 90W or the ones for 130W / 150W? Personally I would not use a bios that throttles even later on a 90W PSU. I would have to bring marshmellows before a gaming session.

- Can you confirm or deny that the total power requirements for the 1645 is greater than 90W? I'm really curious to know what Dells official stand is on that.

- Is there a timeframe for this fix?

If it helps I can post a photo of my 1645 running only FurMark and pulling more than 90W from the wall. A few moments after that my system slows down all the way back to the previous century. There's no need to run other tests at the same time. FurMark (which only loads 1 thread of the i7) on its own can bring my system to its knees (and beyond).

I know some of is are still hoping for a proper fix. Please let us know where you really stand.

 

OK, in response to my own post above, I redid the simplest test with a Watt meter and took photo's. It takes under five minutes and pretty much proves the 1645 needs more power!

Download ThrottleStop / i7Turbo, RealTemp and FurMark (which only uses a single thread out of the eight threads available).

Start 'm all up, and don't let ThrotleStop change your system. Only use it for monitoring purposes.

In a little over a minute and a half, my system pulls more than 90W:



This is where it starts throttling, down to 25% or even 12.5%.



Even when fully throttled down, my system still manages to pull 95W from the wall. This is in under three minutes from starting FurMark! If you leave this running, your PSU will shut itself down due to overheating, I've had this happen in under thirty minutes, using the stock PSU and no modifications by ThrottleStop.



These tests were done with a brand-new PA-3E adapter, a 1645 and no other software running but FurMark, i7Turbo, RealTemp and ThrottleStop. The 3 utilities were used for monitoring only.



After shutting down FurMark (I don't like to break my PSU) the system still pulls 45W at idle, with the battery full, so no charging needed.



The lowest I've ever seen it go is 32W, with the CPU/GPU idle, screen on its lowest brightness, keyboard backlight off, WiFi off, BlueTooth off. But why buy a machine with all these features if you cannot use them.

The base power usage leaves very little headroom for GPU and CPU, although other tests I've done indicate that using the CPU only (so Prime95, Office work, Photoshop, etc) doesn't tend to lead to throttling most of the time. It is mostly GPU usage that leads to CPU throttling.

So to summarize once again for Dell, this time with visual proof: the 1645 needs more than 90W to fully utilise the onboard hardware, specifically the GPU. For a machine with a powerfull dedicated GPU this makes no sense. Why incorporate a ATI 4670 if it cannot be used.

 

Another proof for DELL. The above should be added to the first post.

 

Now I'm wondering: are 130W enough, or should we request higher wattage (150, or even 210)? Could someone with a watt meter and Throttlestop make some test please?

 

Can anybody with knowledge of power supply and motherboards please explain how the extra wattage can have consequences for the 1645s hardware, long term?

Is it really a genuine risk of frying the on board hardware by using a higher wattage power supply than the 90w dell includes, or is this just another shot in the dark excuse from dell to avoid having to send out 130w adaptors?

@Mitchell

Thanks for doing the tests! I guess it's beyond any doubt now that our machines really do need alot more power than what the 90w can ever deliver, with or without a BIOS update.

 

No problem!

I don't have a 130W or 150W brick yet, so I cannot run te same tests with ThrottleStop enabled. But I'm sure there is someone here who has a bigger brick and a Watt meter... Once I get my own big brick, I will re-run the tests.

 

If AC adapter draws 95w at wall, how many watt request tha laptop?
If the AC adapter has an efficient about 80%, 76w is the power that the laptops draws....

Sorry for my "english"

 

I'm not sure how to calculate the power draw at the laptop side. Unfortunately I cannot measure that.

If your method is correct, and the 1645 really does need 90W, it would need at leas a 112,5W power supply. That would be the dumbest mistake ever: to ship the laptop with a power supply that exactly fits its needs, only to forget about its efficiency

But since people are (or were) reporting power draw of over 120W this will probably not be the case.

Anyway, I would still love to hear Dell officially say: The 1645 does/doesn't* need more than our 90W power supply can deliver.

* cross the one that's not relevant.

 

What's the power draw of the RGBLED from minimum brightness to max with everthing else the same? Also curious about keyboard brightness? (one of these days I will get my own meter...)

 

I'm not at my meter right now, so I can't test it right now.

I have checked the keyboard before, and that was about 2W per step, so about 4W in total (at the wall). I will test the screen later tonight.

I think al the little bits add up, and I also think we should be able to use those bits.

 

To solve this performance situation, they are upgrading me from 4GB RAM to 8GB and 720QM to 820QM.

 

Here are my frustrations after speaking with a Dell tech -

1. They seemed to be aware of an issue regarding throttling, but had internal notes that advised specifically against giving new power supplies to customers requesting them.

2. They seem to think a revised BIOS will be able to fix these issues and they asked me to hold onto my laptop until this BIOS was released.

3. They insist that the S-XPS is not a gaming machine and that if the system performs other, non-gaming tasks without issue, it's performing as intended.

Now, in all of my years working with computers, I've never seen once case where a revised BIOS could assist in power draw for components, so I'm skeptical (to say the least).

The last day I can return my laptop for a refund is the 11th. Hopefully the BIOS comes out before then so I can test it out. If not, this thing is going back.

 

If you go to the Dell XPS 16 webpage or the Laptop/Notebook front page there is a "Learn More" or "Help Me Choose" link to help the consumer identify which processor is best suited for them. Click on that link and they say the core i7 is rated for "Extreme Gaming" while the others are rated for regular "Gaming". Dell also says the XPS 1645 provides "Ultimate Performance" on their order page.

It would be interesting if someone would play with Dell Chat pretending to shop and ask if they would recommend the 1645 for heavy gaming...

 

That will actually make it worse as you will be drawing more power

 

I really would be surprised if they didn't recommend it.

Why would they put a dedicated mid-range or high-end (I'm not sure where the 4670 is positioned, but it is definately not low-end) GPU with 1Gb RAM in a machine that is not intended for gaming or other graphics intense work.

I wonder how Dell feel about 3D modelling and other demanding tasks on the 1645. Do they consider those non-gaming tasks?

 

also the design page for the xps 16 specifically mentions gaming as well, nevermind the stuff about ultimate performance. Its not ultimate, the 820qm's + rgbled is a premium that will throttle even more, making them useless. Really whats the point of the 4670 if its not going to be used? or the rgbled screen, if its a premium that will just cripple the system more.

Really what other product out there are you going to buy with premium/high end components and not expect to use them? Tell them its a graphically intense program that throttles, doesnt really matter if its a game. They really have a lot of balls to say that to something like that.

 

That was my question. I didn't go as far as jlivengo did on their website, but I did mention this blurb in my phone conversation:

I asked specifically about the "graphics-intensive applications" and he said that they're talking about programs like "Corel."

*groan*

I just dug this up from the press release when the Studio XPS was first announced:
http://content.dell.com/us/en/enterprise/d/press-releases/2009-01-08-XPSLaptops.aspx

Sounds like they're doing more verbal dancing than a politician at this point.

 

how many hours on the phone and calls did that take?

Every time I call they tried to dial into my computer to adjust the power options.... which I have to explain to them each time would only affect performance on battery!

well good job on the upgrades man!

 

Just ordered my XPS 16 with the core i7 and RGBLED. I asked the sales rep about this issue and she said it will be fine and gave me some spiel about having already sold 59 this month with no problems.

I ordered mine with an intel 1510 (n) wifi card which wasn't on the options list so it won't arrive till 10th February by which time the so called BIOS fix should have been implemented.

As soon as it arrives I intend to strip the bloatware and give it a run for it's money by running some benchmarking on CPU/GPU and seeing how it copes and giving it a blast on Crysis. If it works all well and good and I'll be a happy bunny as it'll be a nice laptop for the price.

If not and I don't get offered a 130w power adaptor I think I'll ship it straight back because despite the sales lady saying dual booting with linux wouldn't void the hardware warranty, that's just the sort of excuse they'd use to waste a whole load of my time trying to step out of their obligations.

Anyway I've got my fingers crossed that by Feb 10th they will have fixed this issue one way or another. If not it looks like I'll be going for a HP envy 15; at least I know that comes with a decent power adaptor even if the screen is a lower spec and their support equally appalling..... Might actually go for an Asus instead.

 

I did some tests again with the Watt-meter. These results might not be 100% accurate (background tasks, etc). But they should give a ballpark figure.

- WiFi / BT combo off -> on: +/- 1W
- Keyboard off -> low +/- 1W
- Keyboard low -> high +/- 1W
- RGBLed low -> high +/- 10W
- Charging the battery: max W available. With an empty batter my system will draw up to 90W!

 

You're a bold man for believing a sales person with regards to defective products. Did you expect them to talk you out of buying it?

 

This is a re-post of my previous results:

Just finished some Wattage testing, found out how much each part is consuming and how much Wattage this system needs on full load...

Battery fully charged consumes 9W
Idle System/LCD off/No battery/Wifi On = ~27W
RGBLED 16W max brightness 7W min brightness
Wifi Chip connected to N network 2W
Speakers 7W
Video Card Maxed 47W + Idle System Wattage
Processor Maxed 60W + Idle System Wattage

 

When you put it like that it does look very stupid!

However I was just trying to demonstrate that they are selling this machine as being and i quote, 'excellent for gaming', according to their sales people. I bought it in spite of this however because I think for the price it's a pretty good spec and I'm happy to put a bit of effort in to see if I can get it working properly.

Of course if it turns up and it still has this issue and they're not willing to send me a more powerful adaptor or if that doesn't resolve the issue then I'll just return it as being unsuitable for purpose. I know I'm protected by UK consumer law in this regard so I'm happy to risk wasting my time as long as I know my money is safe.

 

Why don't you send this to Michael at Dell and see what happens?

 

Very cool. What is your power plan setting when the system is idle?

 

arnt those your old numbers, thought you had a math error or something in there siphen. As my tests showed somewhere near 50w for the cpu alone. other then that everything else seemed spot on.

 

your right, sorry about that... it’s what happens when you have to manually search for your old numbers and didn’t save them to txt ;p lol

I corrected my post above.

 

Im running a customized high preformance power profile. the system is not allowed to slow down for any reason when on AC, everything is allways-on and maxed.

 

i think i understand??

heres my results i had, on a idle system that draws 31-32 watts at the wall. no wifi, dim bright etc.
prime 95 went up to 81 watts at the wall
intel burn test went up to 90watts at the wall

subtracting the differences from the loaded cpu. the cpu took 58 watts for the intel burn test, and 49 watts for prime 95.

also to add to the measurements there was a 30 watt increased between a 7x multiplier at full load and a 12x multiplier at full load.

 

Darn, I didn’t edit the post fast enough look back, I corrected it. BTW, You might want to add these numbers to the new thread.

 

I added it, as your testing is pretty thorough. I just did enough to verify your findings. thanks for that, im sure that will help others to understand more as well. going to have to clean up some parts of the mainpage soon i think.

Since you have the 820qm + rgbled screen, the system that seems the most F'ed in this situation. Do you have any wattage numbers for normal gaming etc? My 720qm + wled screen doesnt hit much above 110w under normal gaming etc.

 

n/m - no longer selling

 

Isn't a PA-3E the original 90 watt adapter?

 

Looks like I'll keep my PA-4E.

 

I get around the same here, ~110W while gaming. It’s around the same because when we game, the video card maxes out before the processor. granted I play all my games at 1080p... if I lowered the settings and allowed the CPU to do more, there might be higher wattage intake... never really tested for that though..

I’ve been wanting to do a YouTube video of the issue got to find the time to do it...

 

youtube it and ill link it to the post, if you get any fps data with throttlestop vs no throttlestop im also building a database of user experiences and games(right above the faq section currently). I should be able to get a few games myself in by this weekend finally.

I was kind of curious the impact of the rgbled screen vs the wled, between our two computer. as 110w is a peak for me pretty much. The max i can draw with furmark + prime95 is 130w from the wall.

 

this got me concerned, some of you ordered their xps 16 right after xmas and it says "in production". You guys are lucky!! I ordered mine dec. 30th and now it says shipped. I planned on cancelling mine but now, I guess Dell is ignoring our concerns and is still trying to make money off some of us.

 

http://www.fileden.com/files/2008/3/3/1794507/ThrottleStop.zip
http://www.sendspace.com/file/gnkg7e

I decided to add a monitoring panel to ThrottleStop that combines data from i7 Turbo. In the future I will probably add an option to hide / disable this monitoring panel but for now hopefully it is useful to gather some more data about the throttling problem.

These new features are in early beta testing so give it a try, post some results and give me some feedback. A screen shot using a Core i7 mobile should fill up that blank space in the above pic. My Core 2 makes the data area look a little naked.

I decided to show the Clock Modulation values for all 8 threads individually now since it seems that sometimes they might not all be the same. They are supposed to be but who knows what the bios is really up to.

The log file includes more data now including the multiplier for the hardest working core. When you are gaming and playing a game that is mostly single threaded, a properly functioning XPS 1645 with 130 watts of power and ThrottleStop working will likely show multipliers of 15.0+.

 

So, according to this numbers, we would need 150W? are 130 not enough?

 

It looks good, thanks for the updated version. I will give it a go on my 1645. This is what it looks like on my desktop (which I cannot get to throttle no matter what I do )



Oh, and I like the More Data feature!

 

Have you added the backup to multiplier after a reboot ?

Thanks

 

Are you asking if it stores the last entered or used multiplier when you close it?
If so, the answer is yes.

 

Yes, but when you restart TS, the box of Set Multiplier is uncheck ?

 

ok, i see what your asking, it dont for me.... is it a safety thing? has to be. so you don't accidentally start on battery.

 

Personaly, I burn 2 times my battery with ThrottleStop but only when the power draw +120w (furmark + Prime95). I run now TS on battery with 115w max and for the moment no problem.

 

--[Deleted]--

 

This link works for me:

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

And I have tried it on the laptop now, that also works great!

 

--[Deleted]--

 

--[Deleted]--