S-XPS 1645 AC Power Throttle Issue Investigation

what do you mean with ati 4670 is not playable in any game? i play modern warfare 2 on highest with no problems....

 

Jesus if it did exist, it would be at such an angle you would need to be a giraffe to use it lol! U wouldn't have to worry about heating issues with fans etc...

 

No problem TStop 1.7 works good, I haven't found a bug but just the backup of the multiplier is missing

 

@romills: A Dell rep reported here that using a 130w brick could damage a 1645. Only when challenged did he clarify that he had in fact only been told that the 130w brick was not validated for use with the 1645. That's not quite the same thing. By all accounts, the 1645 will recognise the 130w adaptor, but still throttle to 90w unless you use Throttlestop.

Where I'm coming from is that even in non-gaming, and possibly even non-throttling, use, my 90w brick gets far too hot for comfort, and I would be much happier with a brick that ran cooler, as I would expect a 130w brick to do.

@eblock12. Thanks for that.
http://www.hardcoreware.net/reviews/review-356-4.htm
confirms your comments on the Wii power consumption - "The Wii is amazingly efficient, using no more than 18 Watts in any of our tests (although it really shows in visual quality in games)".
Though note that last caveat. Do you find this?

But even if the parts that a 1645 and a Wii had in common ran at 18w, the battery charging rate (with the 1645 switched off) of 65w would still leave only 7 watts for the screen, keyboard and so on

@unclewebb

Thanks for the clarifications.

I presume that since the 1645 BIOS detects and reports the 90w and 130w adaptors, there must be some smarts in the adaptor to transmit this info, and some smarts in the 1645 to make use of it? As I also noticed that a non-Dell car/airplane power adaptor being sold in our local Staples (UK branches of Office Depot, I think) was flagged as able to power Dell laptops but not to charge their batteries.

Re switching to battery power; I spend a deal of time in Spain, where the mains power is less reliable than in the UK. So power cuts are a very real issue, and I would have lost quite a bit of work if my computers had not been laptops that fell back onto battery power when this happened.

So it's quite essential to me that my 1645 can do this safely. If I have the stark choice between a throttled 1645 and one with a dead (even if not exploded!) battery, then either way, this machine is not fit for purpose to me.

 

I received mine the day before X-Mas, but was unable to test it due to my own workload (I'm not throttling )

Anyway, I did some quick tests just now, and noticed the following:

When just running P95, my system does not seem to throttle (I'm assuming the <100% on Th0 & Th1 is due to overhead?)



When I'm running FurMark on its own it does throttle though. Unaltered the system shifts down to 9th gear instead of 12th:



When I lock the multi, but do not alter throttling (I'm still on 90w) it shows severe throttling. This is still without P95, and with only a single core really busy:



I'm going to run ThrottleStop and / or i7Turbo 'during my regular work to see what's going on.

I also noticed that ThrottleStop sometimes reports different from RealTemp (see image above and below). The Throttling percentages are sometimes off. Also ThrottleStop seems to not return to 100%, when RealTemp does.



I haven't done any gaming yet, so cannot report on it.

Overall I do really like the system (RGB screen is great when calibrated, keyboard is nice, SSD is *fast*.

 

khaledseif actually said: 'there isn't any game that is playable by 4670 and not by GT230M'

I misread it at first in the way you did, but then realized that he meant 'if the 4670 can play it then the GT230M can'.

This explains your observation that you can game on your 4670

 

By looking at your screen shots, in your first and last test you seem to be right on the edge of the start of clock modulation throttling.

ThrottleStop samples your CPU more often than RealTemp does so it should be more accurate. Neither of these two programs are as accurate as i7 Turbo is.

i7 Turbo uses high performance timers within the CPU which are capable of accurately measuring even the slightest hint of throttling.

In your first and last screen shot, i7 Turbo shows that 6 of your 8 threads are not being effected by clock modulation but the first two threads on core 0 are. That's why their C0% has dropped below 100.00% This core is throttling and the clock modulation value is being cycled between 87.5% and 100.0% on that core. Sometimes ThrottleStop will catch this if it samples at just the right time and sometimes it will miss it. With RealTemp only sampling once per second and with 6 of 8 threads reporting that they are OK, it will most likely miss this too.

One test you can try is to lock your multiplier to 11 while running just Prime95 Small FFTs. That will reduce the load on your CPU enough so that thread 0 and thread 1 should stop throttling and the C0% will jump up to 100.00% which would confirm this.

The i7 Turbo C0% helped 5150Joker better understand the throttling his M15x was doing. Intel designed their CPUs so it is very efficient to calculate the C0% so I think I should include this information in ThrottleStop.

 

I have run one session of P95 where all C0% values were at 100%. This was my first run and I don't have a screenshot of it. From what I can tell - I only did a short set of tests - the throttling seems to kick in sooner when the machine has been on for a while. My next tests would all throttle quicker.

This makes sense to me. Even after stopping the tests ThrottleStop keeps displaying 87.5% for as long as I cared to wait, while RealTemp consistently showed OK. Quickly enabling and disabling the Clocl Modulation option in ThrottleStop made it show 100% and it stayed there.

I will do this test as soon as I finish my work (have too many things open right now, and would like to do the tests in a 'clean' environment).

I do find that information usefull. Slowly I start to understand what's going on, and how the i7 (and Dell ) works.

 

I really like the C0% data because it gives a much more honest look at how hard your CPU is really working. Your third screen shot is a good example of this.

The widget with the nice Intel logo on it reports that your first thread is happily running at 100%. The Task Manager would show the exact same thing and make most users believe that everything is OK.

ThrottleStop clearly shows the clock modulation going on and the C0% on thread 0 also confirms that your CPU is not working anywhere near its capabilities. Maybe Dell's "engineers" are using the wrong tools to monitor these new CPUs when building and testing them but you would think that Intel must have better tools available than I've built readily available for one of their top customers. I look forward to your future tests.

 

Learn how to use linking thumbnails.



then it doesn't screw up the forum columns. and if you have more than three or four, stack them

image1 image2

image3 image4

 

Part of Dell's FUD, or maybe it's just the sales reps' RPG*, is that the 1645 is not a gaming machine.

For an interesting take on that, visit
http://www1.euro.dell.com/content/products/category.aspx/notebooks?c=uk&l=en&s=dhs&cs=ukdhs1

and use the 'Narrow Your Selection' option on the left. Tick only 'Gaming' under 'How fast does it need to be?'

You don't even have to click anything else for the screen to immediately change to show the one result available.

And you'll never guess what that result is

* RPG: Random Porkies** Generator
** Porkies: Pork Pies. Cockney rhyming slang....

 

Does it really matter whether you are gaming or not Midnight_voice? throttling under full load is throttling under full load...regardless. What does it matter if I am doing intensive photo editing, 3d rendering, running a crap load of virtual machines, etc F*** dell and their excuse on this. Not a gaming machine... screw that. They are blatantly selling a crippled product!

 

It matters when they specifically mention gaming on their "design" page of the xps 16 on the dells own website. With all the talk of "ultimate performance" as well. Really think they are playing us for chumps, this has been brought up a lot and they havnt taken any corrective actions to stop. Can only assume they dont care.

 

Don't eat everything they feed you. The Dell website is not set up by anyone who knows anything about computers, it is set up by MARKETING. If you say 'DURR I WANT A GAMING PC' of course Dell will direct you straight to their most expensive systems (Alienware) without any regard as to specifications.

Read any advertising on the XPS16 and you'll see "POWERFUL" and "GAMES" and "PERFORMANCE" and everything indicating that we were mislead.

 

The problem is, there is no rating system that is universally accepted or controlled for a computer's performance. Sure you can bench it with any of a million different products, but every result is almost impossible to compare to another test.

if there was a standard where you could say "This cpu scores 1000" then we could run that benchmark and prove (read: shove it up dell's arse) that we were sold faulty equipment and mislead and lied to about the performance we were told we'd get (especially when our CPUs would score 300ish based on the throttling).

unfortunatly we cannot do this. So what we must do is return the products.

I would encourage anyone who has the time to write a nice long letter to dell and post it up, and we can include it with our returns printed and signed. It would be nice to also start an online petition indicating we have problems.

The sad part is no matter how much evidence we stack against Dell I have a feeling we'll never see a resolution. I'm going to call again on Tuesday and see if i get any better luck.

 

Unclewebb, just out of curiosity, how difficult would it be to create a % on TS that would indicate how throttled we are?

Lets say it would calculate the CPU state and the modulation and say "You're computer is throttled to 25%" indicating that we are only getting 1/4 the performance we should be getting if it wasn't throttled.... is this possible? would be nice to show Dell agents...

or perhaps have it say "Your performance is being decreased by __% due to Throttlegate" lol.

 

Guys...there is some hope, I think. My system is still being captured (the new one is delivered, but I'm out-of-town now) and they told me that in this situation, they've taken steps they've never taken before. Make of that what you will.

~Ibrahim~

 

Not that hard. ThrottleStop already logs the clock modulation percentages. If the multiplier is throttled down to 7 when it should be 12 then it's not too hard to multiply 7/12 by the clock modulation percentage. A 12.5% clock modulation reading combined with a 7 multiplier takes you down to a computer that is operating at about 7% of what Intel designed these chips to run at.

Instead of worrying about this or waiting for Dell to acknowledge this problem or to come up with a fix, just run ThrottleStop so you'll always be running at close to 100%.

 

I wonder if it would not be better to give an equivalent mhz reading intstead of throttling percent. just caust on turbo boost the multipliers should be much higher then just 12x. so 7x is 930 mhz equivalent, with 50% clock modulation thats 465mhz equivalence. and so on

I just thing that might be a better measure, really gets the point across if ya ask me.

 

I'm a bit confused, so ThrottleStop is not recommended to work with the 90w power supply?
I still kinda want to return the laptop, but the my 21-day return ends right before the 2010 CES. I won't be able to know what the SXPS 1647 is about or any other new laptops to make a decision.

 

I would be really interested that as a feature, and seeing a number that represented (as realtime as possible) what my CPU was running at.

Obviously, with TS enabled it should always be right at 100% in theory, but it'd be really nice to be able to turn TS off and have it assign a number to the overall throttle on the computer. Then when i disable it for battery or testing reasons I can have an idea of exactly what is going on behind the scenes

 

Simple solution: initiate your return on the last day possible, and then you'll have so many days to actually return it. It'll give you essentially a few extra days to hold on to it.

 

I agree atlstang. A low MHz number is something that more users could easily relate to. The only problem is that the Clock Modulation percentages are not exact numbers. If you still have the clock modulation testing picture that you sent me a while ago, could you send it again? I might still have it in the original PM message but the PM box was filling up quickly during the early days of development so I'm not sure if I still have it.

To take full advantage of your XPS 1645 you need a 130 watt adapter combined with ThrottleStop.

Does it really matter what will be shown off at CES? Fancy specs and marketing BS don't mean anything these days. Manufacturers are in this habit of churning out new and improved junk so quickly that they never get around to getting things sorted out before it's time to ship. The end user is the poor sucker that gets to deal with the mess.

 

Exactly what I did!

I suspect it will take a while to get the return paperwork and if I change my mind and don't send it back - I've already paid for it.

 

Thanks for the reply, I still have the Dell 90w adapter.
Never requested or bought a 130w one.

Don't wanna go off topic, but people have been talking and documentations have shown a 1647 model, which could be a replacement of 1645 or a Core i5 model. And Sony might announce the new F model (to replace FW) with Core i5 and NVIDIA 330M graphic card at the 2010 CES.

 

I can't find which settings I need to put in Throttlestop to run it with a 90w adapter (and I know it can be done). Can you please help me?

 

Mitchell2.24v,
Where did you get that intel multi-core gadget?

 

I got it here: http://blog.orbmu2k.de/sidebar-gadgets/intel-core-series-sidebar-gadget

The page is in German, but you can download it at the bottom. As unclewebb pointed out this does not very accurately reflect the actual performance of the CPU, but it does give you an overall idea of how the system is loaded by the software you're running.

 

I just had my powerbrick (90W, old version without the blue light) shut down on me while doing some tests. This was without ThrottleStop active. It was enabled, but no boxes were checked. So basically running the system unmodified can load the powerbrick to the point of shutting down.

I only noticed some time after it had happened, so I will have to redo some of the tests.

After unplugging the brick (it was not hot or even warm when I noticed) and plugging it back in, it is now charging my system and puylling a constant 88W from the wall.

I wanted to do the 11x test unclewebb suggested, but it ran nicely on 12x for all 8 threads. C0% 100% all around.



As was to be expected after the above result, the 11x test also ran on 8x 100%


I will redo the other tests soon, but first need to do some work.

 

After 2 hours on the phone with Dell they will be picking up my laptop Wednesday and issuing me a full refund. It took a while to explain that the laptop is not functioning at 100% due to the power supply and BIOS limitation and the technician refused to send me a 130W adapter for free. Their loss.

 

Guys I don't know if it matters, or whether its just more dell 'porkies' but I was talking to Dell ANZ support today and they said a fix for this problem is imminent (end of the week), the guy said January 8th (Friday) Australian time will be the day its released, he also is sending me out an 130W Adapter, I asked for a 150 W and he said no as they don't make them here - but hey they said no to a 130 W before - I guess they have changed their tune. I thought I would share this and ask any one if Dell in the USA are saying anything about this.

Regards
Reece

 

Did any one get information from dell canada?
Thanks

 

My dell rep (norway) told me that they have a fix ready for next week with bios update A04, and he is also sending me a 130w adapter. He did mention something about wanting to replace my mainboard too. Not exactly sure why that was, but at this point i figure it's worth one last try and let dell fix this properly, or i see no other option than to return the machine for a full refund.

Another interesting thing i noticed is that the dell precision M6500 notebook is delivered with a 210W adaptor. The specs are pretty much the same as our 1645 except it's got a 17" RGBLED screen and a i7 920XM upgrade option. But does that really make out a 110W difference?

 

For the United States:

I just want to inform everybody that on 1/02/10 I had emailed Michael Dell about the Throttling issue and when would there be a fix.

About 5 hours later I received a phone call from the Executive Escalation Department and they stated that the Engineers are working on a new Bios that should be released in the 3RD week of January that will fix the issue.

I also had asked about supplying us with the 130W adapter and they stated that will not be needed and that the Bios fix will be enough to fix the issue, as well he did mention that he knew about other forums talking about the 130W adapter and he knows that many people are using the 130W adapter but stated that this computer is not designed to use the 130W adapter.

For now I left it at that and will wait to see if the Bios fix alone will fix both issues, (Throttling/Adapter).

 

Well that what i found in dell canada.
Still the A03. this way we can check if they made real changes

 

I think you missed some differences. 16gb RAM and a 2nd HDD for instance.

 

@LPTP-LVR

You're right, i did miss that, and to be honest i don't really know how much power the extra RAM or HDD requires, but thanks for clearing that up Just thought 120W difference was alot.
That being said, the M6500 is starting to look very attractive. Is the precision series available for regular consumers?

 

Yes it is, but you'll have to finance it yourself because you have to buy it from the Business unit and they can only do Business financing.

 

Yeah anyone can buy it and in my experience the Quadro cards perform pretty much the same in games as the geforce cards they are based on. I think i remember the M6500 could be specced with higher Mhz RAM too...or is 1333Mhz the max now for DDR3?
The M6500 is huge and heavy though anf the power brick is literally a brick

 

I'm glad that they feel that they have a fix on the way but I have little confidence in them at this point and am on hold with them waiting for a return. Not only can I not get my display to show reds in a manner that doesn't make my eyes bleed but my new laptop is throttling like crazy which is a huge disappointment

 

1333Mhz is to date the fastest DDR3 RAM i've seen available for notebooks at least. Yea, it is a bit more weight to carry around, but it will mainly be a DTR, so that probably won't be a deal breaker for me, but i'll have to think about it. Thanks for the info though!

 

DDR3 RAM requires nearly no wattage. Call it 1.5W/GB, to be conservative.

 

Hey Guys,

Was close to ordering this and just posted a request for advice on the owners lounge, but I did want to ask specifically related to throttling two questions.

1) Has the dimming of the RGB-Led screen made a reliable difference to users (it was mentioned a while back but never again)?
2) For non-gamers but heavy HD-Movie watchers, would the throttling issue come into effect?
3) Would you recommend a purchase for a user who already has the 130W power-brick or take a wait and see approach for CES 2010? Great dell offer expires on the 6th.

Couldn't resist throwing in the 3rd again, sorry!

Thanks!

 

Just to answer your 3rd question.
Dell prices never change - they always find a way to be the same regardless of what the special is. The max difference is usually 100$ unless you are lucky and get an elusive 50/75% mystery coupon.

If you could wait, I would hold off buying the xps 1645 just to be safe.
I have no issues with mine, besides the battery discharging while plugged in on heavy loads sometimes, but as I don't game it's not too much of an issue.

 

I think the M6500 comes with a 210w adapter because it costs about 2000-3000$ more than the xps......so they can throw in little dangles like that.

 

New quoted date on the BIOS update is on or around the 1/11/09. This coming from a western-located tech agent I'm on the phone with now.

He also stated that the 21-day return policy is actually 21 weekdays/non-holidays, which is a relief. This gives me a bit of time to test the new BIOS before making a decision on what to do.

 

Haha, I'm on the phone complaining about the issue again trying to get verification of the (hopefully) forthcoming BIOS update. The Dell tech just told me he'll be connecting to my computer remotely and he knows how to fix the problem. This should be worth a laugh.

 

Side note: I got to someone that handles returns and requested a refund for my laptop....they offered me $100 to keep my system.

 

@zimmyntrn

The studio XPS with the specs i have is actually 4.000 USD in Norway if i build it from the dell site now. Looking at the precision M6500, the price is "only" about $3000 with the same CPU, RAM, RGBLED (17" though) and 1GB ATI firepro 7740M , but like gaah mentioned, there might be more costs involved with the business financing deal that i'm not aware of. Anyway, it was just a funny observation.

 

The only way I can play any modern game is by reducing the screen brightness to the minimum and turning off the keyboard backlight (when using the 90W adapter).