**The Official Studio XPS 1647 Intel Core i5/i7 "Owners' Lounge"**

---------Official 1647 Information---------
What's the difference between the 1645 and the 1647?
1645s run quad-core Core i7 processors.
1647s run dual-core Core i5 or Core i7 processors.

How do I know if I have a 1647?
If your XPS 16 has a Core i5 430M, Core i5 520M, Core i5 540M, or Core i7 620M, you have a 1647.

Throttling
This laptop throttles with BIOS A03 and below.
Update your BIOS to BIOS A04 or later to resolve any throttling issue.

Power Adapter
This laptop ships with a 90W adapter. Call Dell and tell them your 90W is overheating. They will provide you with a 130W adapter.
This will provide better performance for your laptop on BIOS A04 and above.

Miscellaneous Information
This laptop does not support switchable graphics.
This laptop has a backlit keyboard.
Many people are reporting extremely long production delays with the choice of an RGBLED screen.
There are no production delays on this laptop involving any other part, including Bluetooth.
TurboBoost is disabled by default. You need to download this driver to enable it.

 

Are your specs the one stated in your siggy?? Looks like a 1645 if according to your siggy:
Dell Studio XPS 16|Core i7 720QM|6GB DDR3-1333MHz|ATI 4670 1GB
16" RGBLED|Intel X-25M 80GB|DVD+/-RW + BluRay|Intel 5300 (A/B/G/N)
Bluetooth 370|3-year (Warranty+Accidental)|Windows 7 Ultimate 64-bit

 

Let me change that! I'm returning that system.
EDIT: Fixed!

 

How do you like the temps? Any better than your XPS 1645? Idle/load?

 

I'm still waiting for mine to ship. I ordered one of the "mistake"s with the 5370 from the Dell Canada sale so who knows if/when it will go though.

 

I idle at about 40, and I haven't even changed the thermal paste yet. Compared to my 720, thats a drop of about 15 degrees -- AFTER I reapplied the thermal paste on the 720. I don't remember what my 720 was at load, but I get a rock-solid 67 while running Prime95.

 

Nevermind. Ignore this double post.

 

Do you have the 1600x900p screen? Do you like it? Thinking about picking up the Best Buy model, I can't stand high res on such a small screen. ;p

 

I have the 1080p one. I like it a lot.

 

does it throttle?

 

I think I know the answer to this question, but I'm going to hold off until I can get concrete performance numbers.

 

I second that

 

Excellent, do that. Keep us posted

thanks

 

Can anyone confirm if the Dell Studio XPS 1647 supports switchable graphics?

 

Nice.

I haven't changed the thermal paste on mine, but I'm idling at 50/49/50/51C with Firefox open. Load hits mid-80s, but I'm too lazy to check now for the exact numbers.

Very sweet upgrade, I like!

~Ibrahim~

 

At this point, I can confirm that there is no support for switchable graphics. Maybe Dell will make it available later, via patch like they did with TurboBoost.

Your 720 with stock thermal paste idles close to 8 degrees lower than mine did, and still 5 degrees after I changed the paste. What the hell, I clearly had a lemon of a chip.
But yea, I'm really happy with it, it really is better than the 1645 (for me, anyway). Especially the wireless. I'm not sure how much better the 6300 is versus the 6200, but it's a noticeable upgrade over the 5300. I used to only get "Good" signal strength in my dorm room, and now I get "Excellent". Obviously, nothing else has changed. Same Windows install, same antennae locations in the chassis, and obviously my dorm room is still in the same place XD.

 

I eagerly await the info on whether or not this laptop throttles, not because I want one (I really want a smaller system as a replacement when i send my 1645 back) but because I want to KNOW that Dell has addressed the issue in the newer systems.

 

Thanks for the confirmation. It only seems logical that they would add a patch. Hopefully. Otherwise, the IGP on the Arrandale chip would just gone to waste.

Anyway, time to place my order. ; P

 

Alright, I've been waiting for UncleWebb to respond to my PM on how to test. Tired of waiting, so let's collaborate on the best way for me to test. I propose the following:

In all trials, record idle temp before start, temp at full load, avg c0 state, and average multiplier.
I think 3 trials and a control is appropriate.
1st, with only Prime95 running.
2nd, with only Furmark running.
3rd, with both Prime95 and Furmark running.
Control, do nothing, just record base data.

What do you guys think?

 

Throttlestop is all you need. run games etc and have it log.. if your multipliers fall below the minimum that is required to maintain the rated frequency then its throttling. if clock modulation doesnt read 100% its throttling..

A real easy to test just for throttling is set your multiplier high. and see if clock modulation will kick in.. (im not sure how this will work if the multipliers are different on the 620 but the logging stuff should still work i think?)


If throttlestop doesnt detect the new cpu's to well, i dont know if it does or not. But realtemp itself can do this same monitoring that TS does,if TS doesnt work out.

Edit: running a game and viewing the fps difference between battery and AC is a good bench. Make sure the powerplay etc is set high on battery.

 

Sorry, probably should have said, ThrottleStop fails with my 620, doesn't recognize it. I'll just run my tests. Back in ~5 minutes.

 

realtemp shouldnt fail.. it can log multipliers, and will warn in case of modulation. perhaps try that?

 

Prime95:
Avg C0: 99.81%
Avg Mult: 21.7
Lowest Mult: 21.3
Highest Mult: 22.3

Furmark:
Avg C0: 27.26% (note 1 virtual core is C0=99.4, the rest are in the single digits.)
Avg Mult: 23.5
Lowest Mult: 21.7
Highest Mult: 24.79

Prime95 w/ Furmark
Avg C0: 100.00%
Avg Mult: 20.0
Lowest Mult: 20.0
Highest Mult: 20.0

Control:
Avg C0: 4.2%
Avg Mult: 10.4
Lowest Mult: 9.2
Highest Mult: 16.7

 

doesnt look like throttling with C0 of 100%.. If furmark + prime95 cant throttle the system, real world app's wont have a chance..

You measured these states/multipliers with realtemp i take it?

that looks like promising news

 

Yup. I'm confused though. I guess I was thinking of C0 backward? Especially when you look at the C0 values of my control test.

 

The C0 state is how hard the cpu is working, (like processor load in windows) takes in to account things like clock modulation etc when calculating the full load. If your system was being throttled, windows etc would still report the load at 100% of the cpu. In contrast C0 state wouldnt report 100% as when throttling its not using at all 100% of the capability of the cpu if that makes sense..

Basiclly if you were throttling, your multipliers wouldnt of been the base 20x that its supposed to be for the furmark+prime95 test and wouldnt be the 24-25x you saw in the furmark test.. And if it was throttling the C0 state would of said something less then 100% load.

bottom line, C0 is cpu load taking in account throttling. If being throttle liked modulation etc, it will never report 100%. Edit: i think with multipliers it will show a 0-100% range for the muliplier setting (unclewebb will have to clear this up)

 

Wow, if you're right, then super-yay. Would you mind running a TS-free copy of my Prime95+Furmark test on the 130W adapter on your 1645, just to confirm?

 

Actually i sent back the 1645 a day ago. But I have done this before and i will tell you it gets locked down to 7x multiplier (it should be 12x). and even some modulation starts kicking in. In fact running prime95 alone was enough for the cpu to start throtlling from the 12x it was supposed to be to the 9x-11x it was throttling too. If i minimized brightness it would go back to 12x :-/

Congrats, seems you made the right choice with the 1647. This is good news as my bro was very interested in these style of laptops.

The only thing i can think of is to test a 1647 with and rgbled screen, as those tend to suck more juice. But even then i doubt it would affect everyday use

 

prime95's multiply is still slightly higher than prime95+furmark though.

 

That is the turboboost kicking in. As you can see under prime95 the C0 state isnt exactly 100%. Turbo boost will switch multipliers at a ridiculously fast rate, the way to measure this is to do statistics that average these multipliers. Realtemp, i7turbo, TS monitor these things accuratley..

Technically there is no decimal place in actual multipliers. The 21.4 or so multiplier you might see, is just the average multiplier as over the sampling period there was higher and low multipliers used..

Which is one reason you wont ever really see the max multiplier of 25x on single thread with turboboost, as there is always other processes on the operating system workin in the background using the other core.

fully loaded, if it were to throttle it would have to drop under the 20x multiplier (2.66ghz). With one core loaded like furmark, it should reach 25x multiplier for the turboboost but like i said before there are always other processes running in the background causing more than one cpu to be loaded.

 

"Average(mean)" doesn't mean "median".

 

Umm, this was prime95 with all "4" cores loaded 100%. Why would other processes bring DOWN my multi?

 

Rest assured its fine.. Its the turboboost kicking in due the C0 state not being 100%

Its not 25x because thats not how turboboost works. The nominal frequency of the i7-620 is 2.66ghz which is a multiplier of 20x.. If only one of the cores was to be used, turboboost will boost the frequency up to 3.33ghz wich is a multiplier for 25x.

With all 4 threads 100% loaded your only supposed to get a 20x multiplier that correlated to 2.66ghz. With just one thread loaded such as furemark does, it will boost up to the 25x or the 3.33ghz.

You are seeing 21x ish because you can see the C0 percent isnt 100%. meaning prime95 isnt fully loading 100% all cores. One core is probably processing other tasks etc in the operating system, causing turbo boost to quickly engage/disengage making an average mult around 21x.

You multiplier should be 20x for all 4 threads being loaded, as that is 2.66ghz. If it were to fall below 20x with all threads loaded that would be a sign of throttling, but that isnt the case.

edit: i think i know what your asking, and the way unclewebb explains it to me, you cant see what the individual multipliers were exactly. its all averages as it switches so fast and a lot more complex to read then the core2duos

edit edit: ill ask unclewebb if he cant explain it better, as hes almost the only one to come out with a program accurately monitor the new core chips. but its not under 20x, thats good

 

DeathWalking: Great news, thank you for testing and sharing with us.
Im very happy, I really liked the Studio XPS and I was waiting to buy one when this hole throttling issue is fixed. As you stated, the 1647 has no problem, maybe because of the 35W CPU.
I thinkIm going to order the shipment in the next few days..

 

Sp 1647 has no throttling issue huh. Sound good.

 

I knew if these new CPUs didn't throttle that they were going to be great chips, especially for gaming. This little Dual Core is running way faster than the Quads when fully loaded. Most games don't fully utilize 4 threads let alone 8 so with a good GPU, these things would be great for gaming.

With all 4 threads fully loaded you can still receive some turbo boost. On the Core i7-720QM Quad core chips this is not possible because the power consumption is too high which triggers turbo boost to be completely turned off and you only get the default 12X multiplier.

For comparison, on the desktop Core i7 CPUs like the i7-920, you get the full +1 turbo boost even when running 8 threads of Prime95 Small FFTs. Some motherboards like the Asus P6T series will limit this when you are overclocking but that doesn't happen until about 4000 MHz. Most other boards ignore the built in TDP limit or give you an option to increase or ignore this limit.

Each CPU has a power and temperature limitation that controls when turbo boost will be available. When you are on the edge, that's when you will see multipliers between your default multiplier and your turbo boosted multiplier. That's a sign that turbo boost is rapidly engaging and disengaging hundreds of times a second based on load.

I need to go read all the details of this thread but first I'm going to go update ThrottleStop so it will recognize your new CPU.

 

Why? If it truly doesn't throttle then it shouldn't need TS?

 

Well, correct me if I'm wrong, but isn't it possible that the 1647 can actually throttle with a different system configuration?

Your system is using a SSD and WLED screen. Certainly, it seems that the throttling, if any, would be a non-issue for this system... but ThrottleStop could be useful for logging exactly what's going on.

 

ThrottleStop 1.90
http://www.sendspace.com/file/mzbu22

If you don't need ThrottleStop, that's a good thing. Not good for my donation collection but good for users that they finally have a laptop that lives up to the specs.

Hopefully this version will work with your new CPU and hopefully I didn't screw anything up for the older CPUs. If there are bugs then let me know.

Even if you don't need its anti-throttling features, it's still a good tool for monitoring and logging purposes.

 

Wow, I'm tempted to return my 1645 for a refund/demand an exchange for the 1647.

 

Do it! Join the dark side. We have cookies.

 

Here's some background info about the new Dual Core i7-620.

It has a default multiplier of 20. When only a single core is in the active state the multiplier can go as high as 25 and when both cores are in the active state and turbo boost is enabled, it can go as high as 23.

With 4 threads of Prime95 Small FFTs, that would have both cores fully enabled 100% of the time.

The amount of turbo boost you get depends on core temperature but mostly on load / power consumption. With the Prime95 load, you are right on the edge so sometimes you are getting full turbo boost for a 2 core load and sometimes you are getting none. The CPU will constantly cycle like this.

Basically the multiplier cycles back and forth between 20 and 23. If you were averaging 21.5 that would be a sign that the CPU has turbo boost engaged about 50% of the time. An average of 21.7 is like having turbo boost enabled about 56.7% of the time. That's an excellent result considering you are running a very stressful program. I think most games are going to be running with a 23 times multiplier and above.

If ThrottleStop works now and you want to play around, you could set Clock Modulation to 87.5% which will take some of the load off the CPU during this Prime test and you should see your multiplier jump up to a solid 23.00. Less load should give you the full turbo boost. For best performance you are obviously better having Clock Modulation at 100.0% and have the turbo boost cutting in and out. This should help prove if you are hitting the turbo boost TDP limit.

DeathWalking: Sorry I didn't respond to your previous PM. It must have got lost. I was ignoring things do to the idiots that have been spamming some of these good threads here.

Edit: The Prime95 + Furmark test shows that you are over the TDP turbo limit so now no turbo boost is available. You end up with the default 20.0 multiplier. That also proves that a properly designed laptop can run both of those programs at the same time unlike the nonsense being spread recently about Furmark being a power virus.

Furmark is a great testing tool. Manufacturers need to use it more often.

 

Guys please correct me if im wrong; The i6-620 is not throttling which means technically the 520/540 will not either?? Or it could be that dell also might have messed them up somehow? Do we have any benchs from any 520/540 users? Tnx.

 

In theory, if the i7-620 is not throttling then none of the lower models should throttle either like the i5-520 or i5-540.

Maybe Dell finally listened to all our complaining and got tired of the abuse. Good for them.

 

you're right, if i7-620 isn't throttling i5s wont either

edit: too late XD

i just hope they enable switchable graphics

 

I would be nice if they did it across their entire line of Arrendale chips.

 

I agree. I'd really love the added battery life.

 

Im very happy this hole throttling issue is not affecting the 1647, but Im still scared
So..
Im buying the 1647: i5-430, Ati 4670, 900p WLED.
Do you guys recommend me to buy the 130 watt adapter?
I cant test the system and then buy the adapter, Its really hard and expensive to find a Dell 130watt here in my country , so.. what do you guys think.. Should I buy the 130watt adapther while I am in the US or just stick to the 90 watt?

 

more test needed pls!.. lol.

 

The only fears i have anything close to throttling would be if someone got the rgbled screen, some how with everyday use im sure that would be in the clear as well. With furmark+prime95 is way beyond what any application will ever pull on a laptop, it would seem with death's test its safe to say you wont need the 130w adapter.

This is making a lot of sense actually, the TDP may be 10w lower. But the cpu's are also halved compared to the 720qm. The 720qm's may had a tdp of 45w, but drew much more power at max load. They were power hungry beasts and probably the main problem.