S-XPS 1645 AC Power Throttle Issue Investigation

It takes a while for the BIOS to let up on the throttling. I believe once you hit or get near the 90W cap the bios takes you down to ~65W and keeps it there for a while and repeats the cycle 90-65 etc...

 

I have the same issue. It will very often never recover from the 87.5% status. In my case it difffers. Sometimes only one thread will stick at 87.5%, sometimes 7 threads will stick at 87.5%. I don't think I've seen 8 threads sticking at 87.5%. And sometimes they will all return to 100%.

To me this feels like a very buggy decision tree in the BIOS.

 

DO you guys think Dell's techy's are competent enough to solve this issue without the proposed BIOS update making things unexpectedly worse instead of better?

 

Hopefully not worse, keep fingers crossed. But I will not surprise me.

 

I've been told by Dell I can return my xps 1645 if they don't fix the throttling with the update when it eventually gets releases.

I was wondering if anyone knew if for everyday tasks the i7-620m will be faster then the i7-720qm?

Thinking about returning it and getting this option for the lower power rating!

 

Nope, in fact that the i7-620M is also only a dual-core. It'll be slower.

 

i7 720QM is a quad, but the clock speed with Turbo Mode goes like following.

1.60GHz Base
1.73GHz Quad Core Turbo
2.40GHz Dual Core Turbo
2.80GHz Single Core Turbo

i7 620M is like this:

2.66GHz Base
3.06GHz Dual Core Turbo
3.33GHz Single Core Turbo

See the clock speed differences? Even in heavily multi-threaded apps it'll be a wash because of the massive clock speed gap.

I'd go for the 620M, it'll be faster in most cases and competitive in the worst, and with better battery life and lower heat.

 

Thanks, I've asked them when the 620m is coming out in the UK and they can't tell me. Could save some cash sending it back and getting the 620m without the RGB screen (I'm on my 2nd screen and it's still flickering) and get a nice SSD with the savings!

The UK site says they have the xps 16 with i5 upto the quad i7 but i can't seem to configure anything other than C2D or i7-720qm.

 

Hi Guys with regards to the BIOS Update working with out a 130-150 W Adapter. My question is - is there some way that Dell can configure the laptop to be more energy efficient therefore rectifying the use of 90W Adapter?

Regards
Reece

 

I would take the Quad Core any day. More multitasking power, and most processor hungry applications will make use of it all, where the 720QM and 820QM will be far faster. The 620M also only supports 1066MHz memory, whereas the 720QM and 820QM support 1333MHz. If you want processing power, the Quad Core's are superior.

 

Not likely, because CPU rated as 45W, GPU rated as 35W, so only 10W left for Dell to work with.

 

Yep, my laptop is supposed to come on Tuesday and i'm contemplating returning it within the 7 day return period if the new BIOS doesn't fix the whole issue.

I cant configure it as well on the UK site, i guess it will be there within a week or so.

 

Does my head in that they can't just say 'we hope to have it on offer within a week'

I keep getting told they are in testing stage and they don't know when it will be avilable!

What a load of crap!!!

 

One thing I learned is that on the Core i7 mobile CPUs, when one thread is using a Clock Modulation value of 87.5% and the second thread on the same core is set to 100.0% then the core will use the lower value so both threads are actually using 87.5%.

I guess the Dell bios guys didn't read this paragraph.

8.7.13.3 Thermal Monitor
http://www.intel.com/Assets/PDF/manual/253668.pdf

"Typically, if software controlled clock modulation is going to be used, the feature must be enabled for all the logical processors within a physical processor and the modulation duty cycle must be set to the same value for each logical processor."

My personal opinion is that the present algorithm to control power usage by throttling the CPU with excessive clock modulation is horrible. Not enough development time was put into that scheme and it's been around for over a year now on the Latitude E6x00 series. I don't like the chances of this being all fixed up next week or the week after that or whatever the standard Dell of India response is.

The next problem is that Dell is introducing the Dual Core i5 and i7 now. These CPUs have a TDP of only 10 watts less (35w vs 45w) compared to the Core i7 Quad core chips. The Dual cores are designed to use a higher multiplier when loaded so there isn't the huge power savings that you would think there should be. When combined with an ATI 4670 and a 90 watt power supply, the new Dual Cores will likely be throttling too when pushed hard with Prime95 + Furmark.

 

Use this to compare would be better. There are many factors affecting the perfosmance. like number of threads, Cache, which I7 720Qm is better

http://ark.intel.com/Compare.aspx?ids=43560,43122,43124,

It depens on what you do with your laptops. For a Software Engineer like me whose daily jobs often run mutiple servers and powerful software applications at the same time, the Quad core is always the better choice.

 

Quad is the better choice for Video editing/Encoding as well, which I why I megaloathe the new i5/i3 lineup.

 

So if you are running multi threaded applications all the time then the Quads is the way to go since they have 4 cores/8threads vs 2cores/4threads in the dual core. If browsing internet, music, games I would say the i5 is sufficient.

 

Yeah I can agree with that, plus I think a quad core is more future-proofed than a dual (even with HT) games and intensive apps are becoming more multi-core friendly.

 

True. Now if DELL can only fix the throttling issue on the i7s

 

It almost sounds like you think the i5s and i3s won't be throttled lol I believed they are only rated for 10W less power consumption...still not enough.

 

We can only hope they wont throttle since the 1647 have a different bios then the 1645

 

Does it have a different PSU?

 

I guess I can go back to BestBuy and try to find that sales guy and ask him to show me the psu on the 1647, If he still works there.

 

Since I found that my question posted above hasn't been answered, I'll politely put here again.

To the experts around here: would it be possible that Dell's choice for the new bios will be: raise the power treshold to 115W for the PC parts (CPU, GPU, screen, etc) and leave the rest (15/35W according to the PSU, respectively 130/150W) to charge the battery?

It seems a good solution to my non technical eyes.

 

lmao he doesn't you got him canned

 

No one would know that other than dell, if anyone answers you it's just speculation.

 

Completely agreed, That's my jobs all about

 

The problem is that even all 4 cores are fully utilized, i720 is just a hair above i620.

 

hair above? You mean as far as performance?

 

It's a little more than a hair. And do you mean all 4 cores meaning all 8 threads? Or just 4/8?


On a slightly related note, how does the HT work? Does it mean that all of the 8 threads are rated at 1.6GHZ performance?

 

http://www.notebookcheck.net/Mobile-Processors-Benchmarklist.2436.0.html

There's a few benchmark on i5 540. So i7 620m should be somewhere between i5 540m and i7 720qm if not above.

 

This is probably a daft question but here goes anyway.

I'm running argus monitor and realtemp loadtester to see how well turbo boost works but when I pull the AC adapter out so im running on battery power the multiplier for all cores drops to x7. I've definately set minimum cpu state to 100% in the power profile for running on battery.

Am I missing something?

 

HP Envy 15 users are reporting the same issue when they switch from AC to battery the multiplier drops down to 7.0 and gets locked there on battery power. I'm starting to wonder if this might be partially a Windows 7 bug. W7 doesn't always use the Minimum processor state value but this could also be just a bios issue.

 

Is this just happening when changing from ac to battery while under load?

 

I Googled around and processor throttling has been reported even on desktops with Windows 7. They say it doesn't happen in Windows XP; maybe this is a Windows 7 problem? Time to test XP..

 

There was an issue on the DFI X58 motherboards during Windows 7 beta testing where at full load the multiplier was getting locked to the minimum value (12) on Core i7 systems. A new bios was released and this issue was fixed as far as I know. At the time I wrote a program called MultiBoost to help with that problem and it worked very well. That same code got reborn as ThrottleStop recently.

Same sort of problem, same sort of fix.

 

I wonder if anyone has installed windows XP or vista and tested the 1645 again for throttling. that would be an interesting angle to tackle.

 

Wouldn't it be the same? lol I mean the BIOS does it all. The Clock Modulation etc.

 

Of course. I just wanted to know if it's technically possible.

 

Maybe? you're still asking a question that only the engineers of this laptop could answer.

 

Where are you buying the laptop from? Did you insist that they send you a 130 watt adapter, or are they including it now with the orders that haven't shipped yet?

It seems that it is a mixed bag when asking dell to include a 130 watt adapter. I've seen posts where people have asked and were denied. I ordered a Studio 17 with i7 core and the R+BG 1080 Full HD LED. Which promises to be throttle city with the 90 watt AC adapter. Estimated DD of 1/20.

 

Couldn't it be caused by something in the ACPI tables that Windows is interpreting? My understanding is that BIOS doesn't run any code after POST, definitely not after Windows has booted, it's limited to when the CPU is in real-mode. And everything after that is done through ACPI, the OS interpreting AML code.

 

I saw a strange bug.

When I sleep the laptop and resume it (for fix the keyboard/mouse freeze issue) I saw that I have a little score up to the GPU in 3DMark Vantage & 06.

I decided after the resume to overclock my 4670. And I saw that I can overclock my GPU from 675/800 to 870/855 mhz ! Before a resume, I can only overclock the GPU from 675/800 to 790/855, after 790 mhz to the core, the laptop freeze and reboot.

So, I decided to test the new overclock with 3Dmark Vantage & 06 for look if the new frequency works or that it's just a bug and the result is very impressive

Another throttling (chipset ???) by Dell ?? Or just a bios bug which reduces performance ??

 

I just noticed that on youtube there are only positive videos about the XPS1645. The videos shows that it can play Crysis without lag etc. In order to convince Dell to fix this issue completely, I recommend that every user in here, upload a throttle-proofing video that clearly shows the difference, would be great if the 130/150W users would upload comparing videos "90W vs 150W".

1. This would get a LOT of people's attention, many users have watched the XPS1645 videos that shows gameplay on it, if there are enough videos that shows throttling, they would get more attention.

More attention = more users considering NOT buying the laptop = Dell looses even more money = Dell MIGHT actually create a solution that fixes this in a satisfied manner.

If we just let Dell throw us around with random release dates for a BIOS fix, when users on this forum clearly proofed that a bigger power supply is needed, then we would be like dogs without a mouth to bark with. All the users saying "I can live with this", NO! It's' not allowed, unclewebb clearly shows you that you loose more than 50% performance, 50%!

 

Ok, well that confuses me now. I was under the impression that the XPS 1645 ran on full multiplier when using battery power and throttles when the AC Adapter was in use?

 

Thats exactly whats is happening. The ac adapter is underpowered and it limits the system, together with the BIOS, to 90W thats almost 75% performance loss.

 

Yeah but on my system the multiplier seems to be locked on x7 when running on battery power, even with high performance power profile selected.

 

is it steady locked at 7 or fluctuates? Plus, how many threads do you have open? It could be that turbo boost is kicking in and thus reducing the multiplier.

 

I just ran realtemp xs bench again there and with the adapter in i get a score of 1038 and the multiplier sits around x20.2

If i run it on battery power, the highest the miltiplier goes is x11 and a score of 538.

I'm running intels turbo monitor gadget and it looks like when i'm running on battery turbo boost seems to be disabled! When I plug the adapter back in the gadget starts showing boost again and the multiplier shoots up to x20.

Is there a setting that tells turbo boost to shut off? I've already set minimum cpu to 100% in my power plan for battery only.

 

The bottom line is i guess, is that if you are using the laptop for normal day to day use e.g. using word, net etc...and you leave the gaming side to you PS3. then you shouldn't have any problem. You probably wouldn't even notice the throttling issue

However if you use it for gaming as it was advertised by Dell but where all of a sudden on the UK site any indication that the laptop could be used for gaming has been taken off magically then you would have problems with the inadequate 90w PSU.