Problems? See this thread at archive.org.
Reading from the previous posts what I have figured out is: when plugged in with the 90W power adaptor, the 1645 is basically a 933MHz core i7. Would this be faster/better than a C2D 2.8 GHz 1640 system?
-
-
-
I got a replacement a few days ago, they have revised a few things like the heatsink and similar. Poking around, the graphics driver has been updated to "FirePro/FireGL" 8.650.*, and X-Fi MB looks newer than the disc. version by a long shot - readme indicates a date newer than June 2009.
Attached Files:
-
-
daniel88: Your log file tells a good story while gaming. I'll just go through a couple of lines of it so other users can understand a little better what it's telling you.Code:
DATE TIME CMULTI STDEV MSR LOAD% NOTES 01/01/10 11:26:26 8.000 0.000 8.000 28.12 01/01/10 11:26:27 8.000 0.000 8.000 26.00 01/01/10 11:26:28 7.999 0.001 8.000 26.30 01/01/10 11:26:29 7.660 0.048 7.000 25.52 01/01/10 11:26:30 7.000 0.000 7.000 40.14 01/01/10 11:26:31 7.000 0.000 7.000 41.36
The CMULTI column displays the Calculated Multiplier based on measuring high performance timers that Intel includes inside these CPUs specifically for this purpose. Each thread contains 2 timers which run at over a billion cycles per second when the CPU is loaded. This method tends to be extremely accurate and these timers respond instantaneously to any changes in the multiplier which can be changing hundreds of times a second based on load.
The first line shows a calculated multiplier of 8.000 with a standard deviation of 0.000. That shows that all 8 threads are reporting the exact same multiplier.
The MSR column is the old way to determine the multiplier by reading it directly from a Model Specific Register (MSR). When turbo boost kicks in, this method is not always accurate. In your example, it helps confirm that your multiplier is only at 8.
The LOAD% column is actually C0% which is a measure of the amount of time the CPU is in the C0 state executing instructions. Your readings of 25% to 28% show that you are fully using two threads so at most the CPU should be able to handle this load with 2 cores in the active state and 2 cores asleep in C3/C6 at any instant. When the CPU is like this, it should be able to mostly use an 18X multiplier. When you are only getting an 8X multiplier because of Dell's throttling scheme, you are getting less than half of the CPU performance that an i7-720 is capable of. How's that for a kick in the balls?
At the 11:26:29 mark, a calculated multiplier of 7.660 is a sign that the multiplier is transitioning between 8 and 7 during that one second sample interval. The following 2 seconds clearly show that is indeed the case and now the multiplier is only at 7.000 with 0.000 standard deviation so all 8 threads are locked at the lowest possible value. With a 40% to 41% C0 load factor, that is once again your CPU operating at less than half of what Intel designed it to do because of Dell's shoddy throttling scheme and decision to use a 90 watt power adapter.
With a 130 watt adapter and ThrottleStop, I can guarantee you that this game will run much smoother and be much more enjoyable.
There are older games that mostly use a single thread and for those, you might not see a significant difference. For any modern game that is multi-threaded, it's going to be able to take advantage of some more watts combined with ThrottleStop.
Even on battery power there is no guarantee that your laptop will be allowed to run at 100% of its capabilities as delivered from Dell. Post another log file of that if you want me to have a look.
Ch00kz: Intel designs their CPUs to run reliably up to 100C. If they weren't capable of running reliably at this temperature then Intel would reduce the thermal throttling point and prevent these chips from hitting that temperature. Thermal shut down isn't until approximately 125C so Intel obviously has a lot of confidence in their CPUs. Too bad Dell didn't understand better what these CPUs are capable of. Maybe they should read some Intel documentation and learn about these things.
-
Wow, thanks for putting this into perspective. I can't believe that they haven't released a new BIOS update and sent out the 130W adapters free of charge to fix their fault.
How many complaints will it take?
... and thank you for releasing the application to fix their fault. I find it astonishing that a multi billion dollar company has failed to do what one man has done. This predicament is worthy of a newspaper article. -
I'm not sure why it wasn't happening before but I got some 22.x multipliers on the single threaded Prime95 test when I ran it again.
Funny thing is, the Furmark test at 8x MSAA along with Prime95 doesn't seem to make things much worse than running the blended tests in Prime95 with 8 threads by itself. It seems to move between 9 and 12 spending more time around 12 than 9 (full screen brightness RGBLED).
edit: Just saw your comments in the other thread about how to properly read the i7 Turbo logs. Will run the tests again after I sleep.
So it's very disappointing to see this also affecting those of us who weren't even planning on using the machines well advertised gaming/multimedia capabilities.
Anyone with a 130W adaptor and Throttlestop let Prime95 run all night and then check inside their laptop for scorch marks anywhere on the motherboard? This whole situation just seems way too deliberate not to have been planned.
If Dell wanted to save a few bucks by sending us 90W power supplies then they probably also wanted to save by using parts capable of lower power handling inside the machine itself.
Of course if it can run from battery without throttling, then I guess we'd have to be concerned about the parts between where the adaptor plugs into the laptop and where that line joins up with the battery. Everything "downstream" of the battery should be fine powered by a 130W adaptor. Unless Dell meant to throttle the battery mode too but botched it. Maybe Dell's fix will make things worse?
Tom -
-
-
I'm not the one who wrote to Dell's executives. Not yet, at least
I'm just wondering wether to wait for XPS to be delivered to me or not before writing. -
-
@thewhitewizard
This is the machine i'm getting if these 1645 issues aren't resolved by Dell soon. It's a 17", so obviously a little bigger, but i think it's worth having a look
http://www.clevo.com.tw/en/products/prodinfo_2.asp?productid=237
It's got a 1920 x 1080 WLED display, and an i7 920XM upgrade option.
Was looking at the D900F too, but it was too pricey and an inferior screen. -
-
Are you sure, my HP DV6t works fine under battery or charger.
Attached Files:
-
-
![[IMG]](images/storyImages/throttlestop.png)
http://www.fileden.com/files/2008/3/3/1794507/ThrottleStop.zip
I was working with 5150Joker on the Alienware M15x thread and we discovered that his laptop uses a different type of clock modulation compared to the XPS 1645 so that new feature has been added to ThrottleStop.
I have no idea what laptops this type of clock modulation applies to but if the new Chipset Clock Mod feature is not checked, ThrottleStop will be able to monitor for this and will log that data to a file for you.
I've also increased the frequency of monitoring and resetting the magic registers. This might improve game play a little more. I might add an option in the future so this frequency is user adjustable.
I also added a new INI file option called:
MoreData=1
This will create data for the log file 8 times per second instead of once per second.
I made a few changes for the Core 2 owners as well.
Any feedback is always appreciated.
khaledseif: I believe it is the HP Envy 15 with the 7X multiplier on battery problem. Thanks for your screen shot. Any laptop you buy you almost need to take it for a test drive first and run some benchmarks on it to make sure it can deliver what it promises.
Here's the post I was referring to.
http://forum.notebookreview.com/showpost.php?p=5660613&postcount=4 -
-
Thanks for your update Have you added the backup of the multiplier after a reboot in this version ?
-
S-XPS 1645 AC Power Throttle Issue Investigation
Discussion in 'Dell XPS and Studio XPS' started by Zlog, Nov 26, 2009.
