So I'm working on a review of the latest XPS 17" machine, which I got with the overclockable CPU. You can hitch it up as far as 3.17 GHz which is neat. However, some oddities with Speedswitch.
First, some background: Speedswitch is an Intel thingie which turns the speed of your processor up or down, automatically - based on CPU load and available power [battery, AC, etc]. Using the neat tool SpeedswitchXP, you can fully customize your processor speed; and another neat tool called CPU-Z will faithfully report the exact speed. OR, you can leave Windows to do power management and Speedswitch on its own, which is a lot less predictable (hence the utility for SpeedswitchXP).
So now Speedswitch has the following settings:
1) Max Performance: Locks your Core 2 Duo (or other Intel processor) at top speed, even when PC is idling. Net result: Lots of fan noise.
2) Dynamic Switching: When idling, your Core 2 Duo is at lowest speed. When load goes up, your Core 2 Duo runs faster and peaks if load remains. Net result: Optimal fan noise.
3) Max Battery - Battery Optimized: I don't see a difference between these two options, they both seem to lock your Core 2 Duo at lowest speed, even when system is under load. Net result: Least fan noise.
On every other computer than the Overclocked Dell, these switches work exactly as described above. So you know exactly what you are getting. On the Dell, its quite confusing.
Initially, the switches seem to work. However as the system load continues, even if you are under Max Performance or Dynamic Switching, the CPU starts throttling down...so lets say I overclocked my CPU to 3.0 GHz (actual example). As I compress massive files, the speed goes down to 2.66. Then down to 2.33. Then down to 1.0. Yes, 1.0. That's actually the speed setting for "Max Battery".
Now I am wondering why my system is underclocking itself as low as the lowest power state when I am telling it I need to run at maximum performance.
Is this a case of Dell choosing what is right for the consumer? Maybe its some sort of security setting built deep into the BIOS that automatically down-throttles the system based on some funny algorithm. At any rate, its a really bad thing that its this way. This severely hampers performance during lengthy tasks. You basically have not even the baseline Core 2 Duo 2.33, but a processor that is insistent on pushing itself back to 1.0 GHz. Needless to say, after paying nearly $400 extra for the overclock capability, I am not happy with Dell overriding my decisions.
I was wondering if anybody might care to comment on this?
Right now this system is basically slower than a regular 2.33 GHz Core 2 Duo CPU for any task worthy of mention. Sure, when the PC is sitting idle, the real clock speed goes as high as 3.17 GHz. But the moment I start doing anything, it slowly works its way down to 1.0 GHz.
Seems like a waste![]()
-
-
i'm really shocked
i really hate it too when i am processing something that requires lots and lots of time, the processor just clocks itself down. it was that time when i was using ACL 7 (old software, auditing) and trying out some 2GB and millions of logs to be sorted out. i left it sitting to do the work and it was a long way before it finished itself.
i thought it was that long and i tried it again. this time i was using the computer. then it was a bit longer. strange. then i tried it again, now only moving the mouse so that the computer recognizes that i am doing "something" and will not go down. same as that when i am using the computer while processng long data.
maybe you should try to talk to them why is this happening. 400 bucks is 400 bucks... like 4GB = 4 sticks of 1GB or so -
According to someone called zyb on notebookforums.com, it's best to set it to 2.83GHz since the fans only come on as needed.
http://www.notebookforums.com/showthread.php?t=179939&page=4
Someone else in the same thread also suggests just leaving it 2.83GHz, which seems to be the optimal setting, unless you like fan noise. As I understand it, the processor will automatically switch down to lower power state when under low load.
I'd guess that there is just some sort of conflict going on between Speedswitch, the CPU, Windows, which is resulting in slowdown.
I'm hoping to get an XPS M1710 within the next 3 weeks and will be very interested to read your review (with any benchmarks, if poss) on performance gains when overclocked (eg. SuperPi 2M, lame and ogg encoding, XviD encoding with AutoGK at 2.33GHz, 2.83GHz, and 3.16GHz). -
Also, check out question 1 on the FAQ of the SpeedswitchXP home page which mentions of a similar problem on an older laptop which is caused by the BIOS switching it to a lower speed for a set amount time:
http://www.diefer.de/speedswitchxp/faq.html
This could be a similar problem.
The home page does not mention support for Core 2 Duo processors either which may cause a problem. -
Hmm...in some ways, I wonder if it built in to downclock when temps get too high, which they will if you OC too much. However, the speed shouldn't drop down so far. I also do wonder if you just have to many programs trying to determine the speed and they are giving you conflicts. I am not sure. I would call Dell, don't you get special XPS support?
-
I don't think the issue is SpeedswitchXP. All it does is set Windows Power Profiles. So that can't be it. I think what's actually going on is the BIOS - or at any rate something other than the OS Power Profiles, with the power to override the OS Power Profiles - is overriding the speed settings. For the record, SpeedswitchXP works with Pentium M, Core Duo, and Core 2 Duo laptop's I've had.
So I have no idea what Dell is doing. Except that I'm exceptionally disappointed. This "overclocking" is for show only. It doesn't last any lengthy operation. In fact, you end up with a 1 GHz processor which is why this laptop performs worse than my 2.33 GHz Core 2 Duo in my tests. Shocking! And brain-dead design on the part of Dell. Either they are outright and intentionally lying to consumers about this overclocking thing, or they are really not strong in the smarts department.
PS: I don't want to call tech support - its going to be some Indian reading back some scripts and asking me to reinstall the OS -
Does your system throttle itself during lengthy operations if you just set the speed in the BIOS (eg. 2.83GHz or higher), with SpeedswitchXP uninstalled ?
This would help to identify if there is a conflict with SpeedswitchXP or not.
I can't find it at the moment, but one of the reviews I read on the M1710 with T7600G had them running at the max 3.16GHz for a week and running lots of intensive benchmarks/games. During that week they said they got a couple of lockups which was probably due to the CPU overheating, but other than that it ran without problems. -
Yes, it still throttles back. Basically the whole reason I installed Speedswitch XP was to prevent the throttling
But it has had no effect.
Can you direct me to that review? Maybe I can talk to those chaps and see what I can find.
They might not even been *able* to notice the issue, by the way. Most speed reporting programs detect the speed as the overclocked speed and are unaware of the throttling. The only accurate tool I can find is the CPU-Z tool - and the ultimate results are coming from the real world slow performance. -
I'll have a look for the review.
You may also want to post to the notebookforums.com thread I mentioned in post 3 to see if anyone else running the T7600G overclocked suffers from the automatic throttling problem. -
Below is the link to the review:
http://www.hardwarecentral.com/hardwarecentral/reviews/article.php/3659156
According to the review, it says that "Dell Tech Support will verify the full functionality of the CPU up to one speed step above the factory default setting," (ie. 2.5GHz), so try overclocking to 2.5GHz and see if it still throttles. If it does, then it sounds like there might be a problem with your system which might be worth a phone call to DELL support. -
BTW, one other benchmark I'd like to see at different OC speeds is for compression (7z or RAR), since the Merom Core 2 Duo seems to be able to match the Conroe Core 2 Duo CPU's in all benchmarks apart from ones that work on a lot of data, presumably due to the slower FSB of notebooks compared to desktops. It would be interesting to see how much of a boost a faster CPU will give to this operation.
-
I just remembered a MS hotfix for multi-core CPU's which might help solve your problem
http://www.laptopvideo2go.com/forum/index.php?showtopic=10272 -
Wow, thanks for all these great suggestions and feedback. I will definitely try them out and get back to you.
What's really upsetting in this whole issue is the extent of the underclocking. It goes down to 1 GHz, the lowest speedstep setting.
I just wonder - why -
So to clarify, you are using SpeedSwitchXP @ max performance mode and the CPU still downclock. The reason I ask is because I was messing around with someone's E1505 the other day and there was something that was a little misleading to me. I was trying to run the 3DMark05 benchmark so I was cranking everything up to max performance. In the Dell quickset I selected max. performance. Interesting thing is that on max performance it downclocked the CPU to 1GHz, instead of holding it at 1.83GHz. So I switched the profile to Home/Office and then it held 1.83GHz. Don't know if this would be your issue or not, but I thought it was very misleading.
-
Issue solved. I can't take credit for it - credit is wholly due to:
http://www.pbus-167.com/chc.htm
VERY neat proggie. It was actually able to do what SpeedswitchXP tried to do - it locked the CPU speed at my exact desired values, I guess overriding the Dell BIOS or whatever, giving me my exact speed.
I am running an extensive process now that normally takes 1.30 hours on my 2.33 GHz Core 2 Duo, lets see how long it takes on the 3 GHz Core 2 Duo
Can't praise the proggie enough. It even lets you manually adjust processor voltage (and monitors processor heat). Overclocker's dream.
Dell must be shipping this proggie with every overclockable XPSOf course they won't, but at least this little bit of information will make it into my review.
More coming later... -
Nice find mimarsinan!
I'd be interested to know if the Dynamic switching mode works correctly/effectively with this prog ?
Very much looking forward to your review! Can you reply to this thread when it's up so I get notified when you post it. -
do you mind posting a step by step on how you overclocked with that program? (i had it on my laptop before already for some reason). i am quite interested to see if i can clock my 1.86ghz core2duo to a 2.00ghz.
thanks -
To overclock, you can look at SetFSB:
www13.plala.or.jp/setfsb/
(never mind what evil Google says about the site harming your computer, the program either works or doesn't - I wonder who listed it there)
or you can look at CPU-Z:
http://www.cpuid.com/cpuz.php
These very neat tools work for most desktops and laptops and will let you overclock. However I must say based on past experience that they do not work with Dell laptops. Dell really does weird things with their BIOS. Some of my findings reported above were actually incorrect, thanks to additional limits Dell seems to have placed on their *overclockable* BIOS!!!
More on that below... -
"Previously on Lost..."
If any of you are watching Lost, you will know that it adds more and more layers of mystery, one on top of another, without really explaining anything at all. Well, that's so far been my - by now decidedly negative - experience with this "overclockable" Dell.
So in the last "episode" I reported that I had figured a way to lock the speedswitch in a high performance setting...I was wrong.
First, my test results were off again. The task that took 90 minutes on my 2.33 GHz system was again taking almost 180 minutes on the 3.0 GHz Dell. Argh. So something had to be wrong.
Even though CPU-ID was reporting a frequency of 3.0 GHz after installing the latest "silver bullet" proggie referenced above, it turns out that was incorrect. Dell's BIOS - or whatever custom thing they did to this machine - was actually lying to CPU-ID, a remarkable feat I might add, because this is one solid program that I haven't seen go wrong before.
After being alerted to this by my miserably failing test results, I looked for other CPU tools. I found one called RMClock:
http://cpu.rightmark.org/products/rmclock.shtml
This neat tool actually has two fields for each CPU core:
1) CPU Frequency
2) Throttled Frequency
Yep, it has two separate fields, and for good reason I guess. When the CPU-ID tool reported a CPU Frequency of 3.0 GHz (which I was able to obtain only after installing what was the "silver bullet" proggie), the Throttled Frequency as reported by RMClock was *still* not right.
In fact, during the lengthy operation, the 3.0 GHz base frequency was throttled down to 1.3 GHz.
Yes folks, you read that right. 1.3 GHz.
That's SLOW. Slower than any Core 2 Duo worthy of the name.
So that's why my performance was terrible. And for all indications, while the CPU pretended to be running at 3.0 GHz, it was actually running at less than half that speed. Leave it to Dell to impress!!!
OK...so I ran another test. I set the *overclocked* speed to 2.5 GHz. Then I started the task. As best as I could tell, no throttling occured at 2.5 GHz. This is of course using the "silver bullet proggie" as probably it still would throttle back without it.
The next test was at 2.6 GHz. Yes, throttling did occur at this speed. It didn't go far down as 1.3 GHz, but it did go down to 2 GHz and 2.3 GHz interchangeably. Test results were needless to say disappointing.
The next test - which is currently ongoing - is at 2.83 GHz. The first test ended up with the CPU hitting temperatures in excess of 95(C) and the system powered itself off. I am now running a second test wherein I changed the processor voltage, leveraging some more of the silver bullet's capabilities. Lets see if this one fries also.
I was thinking before the latest 2.83 test that the BIOS is manually and "un-overridably" interfering the moment temperature goes high. But like I said...this is just like Lost, one mystery after another. As far as I could tell, at all other speeds, the throttling occured the moment temperature hit 80(C), whereas with 2.83 it does not seem to occur.
Is the 2.83 a sweet spot for this *overclockable* Dell, where the BIOS - due to a bug or oversight - does not throttle the speed down?
Why did the BIOS throttle the speed down all the way to 1.3 GHz when running at 3.0 GHz?
Is 2.5 GHz the only real reliable setting for this laptop where the BIOS won't forcibly throttle the CPU down?
These are just some of the questions I've got today.
All in all, I'm extremely disappointed with whatever it is that Dell is doing with their BIOS:
Its NOT predictable, (downclock to 1.3 GHz? lie about real speed?)
Its NOT beneficial, (completely destroys system performance)
Its NOT for safety (system did experience a shutdown!)
At best, this BIOS looks buggy to me. At worst, its Dell outright lying about the true performance of the computer.
"Dellienware" has a similar model with the same CPU that is capable of running at 2.66 GHz. I bet that can easily outclass the Dell, since the Dell keeps shooting itself in the foot. It costs only $100 extra whereas I paid $375 extra for the Dell version of the CPU, in hopes that it might be overclockable more - and give me the choice of what I want to do.
They're both Dell's anyways though, so the "Dellienware" version might also have "unadvertised" features like the ones I have described above. And the only way to find out seems to be spending a lot of money...not advised for either.
So I will report my final findings either in the review or in this forum...but suffice to say, I expect things to get only stranger.
<To quote the Mac-vs-PC ad>
Vista UAC: You are coming upon a sad realization. Cancel or allow?
PC: Allow. -
I can only think that you have a problem with your machine, since as can be seen from the following post, someone with the t7600g did some benchmarks while at 2.33 and 3.16, and the 3.16 benchmarks are about 33% faster.
http://www.notebookforums.com/showpost.php?p=2475222&postcount=58
If it was throttling it, then it would not be 33% faster.
Can you try the same PerformanceTest 6 benchmarks aand see if you can get the same results. If you can't, then it would definitely indicate a problem with your machine. -
ok I got pulled into this thread buy the guy just above me
I currently own an M1710 with the T7600G and I noticed a couple of things about throttling
1. throttled down to 2.33 when comming out of standby.
2. Throttled because of heat
solution.
installed NHC (notebook hardware control) and didn't ahve a throttling issues with the exception of comming out of standby. -
Thanks Bal`thzar!
-
How did you discover that you don't have throttling issues? Which program did you use to find out your CPU speed?
-
He says: "The CPU only hits 80c if it a 100% utilization (constant) for games and stuff your fine."
This is definitely not the case with my machine - I've seen it exceed 95c. So I think he has a cool environment...and the laptop definitely seems to be intent on watching its heat through the BIOS, but the behavior seems buggy as described above. -
as far as identifying CPU speed.
NHC monitors your clock speeds and displays them in the taskbar. -
Unfortunately, NHC reported speeds are incorrect. Install RMClock, and you will be able to see the dynamic throttling in action.
-
-
Yes, unfortunately this overclockable laptop is plagued by a constant throttling problem; I'm afraid my review will be quite negative, a type of "buyer beware" thing.
-
Hi,
I just got one, a few days ago, what I did,(besides installing SpeedswitchXP) was installed Diefers I8KfanGUI
Keep the Fans set to full & when I do things which require heavy load, with mine set at 3.17Ghz it stays at that speed stable...
What I did was purchased a separate HDD with XP Pro SP2 on it, so I can use it when I want, & I just power off & swap the original HDD back in to run Vista Ultimate.
In Vista Ultimate it runs at 3.17Ghz all the time, when you set it to!
PCfixinman
Overclocked XPS m1710 Issues...
Discussion in 'Dell XPS and Studio XPS' started by mimarsinan, Feb 23, 2007.