Dell's Graphic Driver vs. Modified Nvidia Driver for 445M

Fantastic RedlanceEQ!! I am agree with you!!
Thank you very much!! Got to get the drivers NOW!!!

Do you think I would need to unistall the previous driver before to proceed with the installation of the new ones?
You know just to avoid trouble...

 

Cat in a Hat- All Futuremark(3dmark) software will switch from Aero to Windows basic automatically, the reason it does this is the Aero desktop takes up RAM and GPU/CPU cycles to run and would kill performance levels if left running in the background. Daily usage doesn't really show it but the idea of benchmarking is to see the absolute peak of what your computer is capable of so it switches to Win Basic.

kroenen04- I've been sick, I'll get back to you on your pm sometime today when I get a minute.

All- Yes the Nvidia drivers will work just fine with no modifications necessary, the only thing I had to mod in the 266.44 package were the id strings but ther wer a few additional things modded to make sure it ran best as it was originally a desktop driver. Now that Nvidia is directly supporting the machine I would suggest running their drivers, they should be stable and give awesome every day usage performance. Unless you are chasing extreme overclocks and benchmark scores there is absolutely no need to run modded drivers.

 

NICE report! and Dane Answered your question before I so he is the "king of the lab"!

Noted. Thanks Dane!
I will compile my report on the 266.58 in few so lets get some more results posted!
Rooster

 

I would, I have posted the best way to load drivers on the mission statement on first page.( its the standard that I do with all driver testing) use driver sweeper and follow the instructions that Dane supplied us. This is a suggestion, not an order.
Rooster

 

Thank you Rooster!!
Need to do some reading tonight then!

Love this forum!

 

Regarding the variance (or rather - lack of!) in my test; is this due to a) 3Dmark bypassing any Nvidia panel 'mods' or b) shows that they don't actually impact performance at all?

Given that (according to 3Dmark) you should have all settings maxxed to get the highest results, I'm left a little confused

 

Yes, that is exactly what I wondered for a while, and your scores confirmed it. +1 It most likely has preset settings that override the global settings. Also, is that the A03 version of the Dell drivers. I was getting close to 12k with those, although I don't know if you are overclocking.

I did notice I can't overclock as high with the newest nvidia.com vs the newest Dell but so far only by 10MHz GPU or less, so that shouldn't matter much. The temps all seem the same on the new nvidia.com, but the fan isn't running on medium anymore when not gaming.

 

Yes, these forums are one of the greatest things next to the laptop. It's great to gain so much knowledge and then spread it around to others.

 

Per your request I did, though it might appear underwhelming but the post-its were just for the Graphics Driver Thread, I had already compiled the other post-its on to a document for neater storage.





For history all to see...
Rooster

 

Haha yea, once you get a considerable stash of post its, you can bury them in a time capsule.

 

When I reviewed your post , I thought to myself that it looked different but didn't put much thought into it because the scores were there and looked correct.
But putting more thought into it, I noticed that you used the Nvidia application to adjust your settings, which until now, I hadn't thought of doing. I would just install the driver, made sure that it installed correctly and went about testing with 3DMark6 and Furmark stability and ran a game or two for performance notes and thats it. I will have to investigate further into this procedure. As to 3DMark affecting the settings, I would agree with compwiz0620 that 3Dmark probably does reset for performance.

When you noticed that you couldn't overclock as high, are you saying that it would crash with higher settings or you couldn't physically set the overclock higher due to the driver restricting the limit for overclocking?
And are you saying the fan is running slower or faster when you are not gaming?
Just trying to get it right...
Rooster

 

I was able to use 750/800 for a while with no crashes before I switched to testing VRAM overclockability. 750/800 crashed pretty much instantly on the first 3dMark06 test after I installed the newest nvidia.com driver

Everything seems the same in Civ V. Maybe it runs like 1C higher. When I am not gaming, the temp is still high 40C/low 50C, but the fan doesn't run at medium anymore like it did with the other nvidia.com drivers.

 

This my results from installing Nvidia driver 266.58
Observations before I ran the tests:
I've notice that before activating Afterburner(AB) that the idle core temp is lower (46C) and you can run, though not required for this thread, the benchmark test for Furmark 182. After activating AB(even with factory settings of CoreClock(CC) 590MHz, ShaderClock(SC)1180 MHz, MemoryClock(MC) 800MHz), you cannot complete the benchmark test for Furmark 182 and have to end the program through Task manager. The idle core temp also increases by 5C-15C (60C)

Settings for Furmark 182 Stability test:
1280x1024 resolution
MSAA 4x
Xtreme burning mode enabled
Post FX enabled
Ran test for minimum 15 minutes

On to the results...
after clean install no changes made

3DMark6 score 10123
Furmark 182 stability test (FramesPerSecond=FPS) Min FPS 16 Max FPS 19 Max temp during test 71C
Furmark 182 Benchmark (optional) score 1238

With AfterBurner Overclocking utility
Settings CC 712 SC 1424 MC 912

3DMark6 score 10888
Furmark 182 stability test Min FPS 24 Max FPS 27 Max temp during test 79C
Furmark 182 Benchmark (optional) no score due to conflict with AB

Overall impression of performance of 266.58
meh....
nothing really jumped out or showed improvement over last driver 266.44
games that I tested with (ME2, SC2, Warhammer) performed fine, no stuttering but I didn't expect it to.

For comparison and just to refresh my memory I will re-install Dell's A01 (259.39) just to see if the idle core temps will drop or if I need to re-paste the GPU

Curious to hear everyone else's opinion.
Rooster

 

+rep for that dude! You rock as a librarian Thanks man

 

Can I ask the reason why with an OC'd 445M, I'm still hitting +600 on your score with an un-OC'd one?

 

good question.
it changes from driver to driver, and my scores compared to everyone else's has always been lower than theirs. The scores on new systems tend to score higher than the older sytems and no graphics card is the same. These are observations that I have made since I started recording the results. If you look at the first score I recorded it's higher than I have now. and when I overclocked it with that driver(259.39) the score was alot higher than when I overclocked with this driver. It's something that I am still investigating.
we will see when I revert back to 259.39 later on.

Rooster

 

You mean you think you should get a higher score? Which driver are you using, and what are your overclock speeds? With 266.58 at 740/800 I got 11187. If I can find the stable overclock for this driver before a new one releases, I will post again. I had over 12k with some of the other nvidia.com drivers, but I don't remember the overclock speeds. I will have to peruse the threads, although Rooster probably has it on his post-its the exact post #

Here it is!

 

Yea, and it all comes down to practical application. If driver A's score is 2k points higher than driver B, but driver A crashes your favorite game, but B is rock solid stable, of course you will pick B

Or B could have some optimization for your favorite game that makes it run faster than A in that game, so then again you would pick B.

Same thing with those stress tests. I can run Furmark all night and not crash, but then I will play a game for like four seconds and have to reboot

 

Get well soon..and sorry when I stressed you.

 

No truer words have been said, and you hit the nail on the head.

Rooster
p.s I actually didn't have the post # number of what you posted earlier, but I will, because that's a juicey bit of info to keep around for this thread.

 

No, exactly the opposite.

I posted my clock speeds for a reason - to compare to OC'd cards. e.g. a vanilla 3Dmark score vrs the increases you see with OC'ing the card.

This allows a % comparison.

e.g.

With driver series X you get Y 3Dmark
With "" X + 10% OC of 1,2,3 modules you get Y+n*3 3Dmark
You then give what +C temp you're getting from the increases.

And so on. It allows people to accurately gauge where they're comfortable OCing to.

e.g. My system has to last 3 years, I don't run high end games, and I don't have the luxury of spending too much time fiddling, so, ergo - I'm a inbuilt conservative OCer, so will always have a low acceptable temp cut off, as I don't want to cook off my thermal paste too soon. If I can get +400-600 3Dmark with a mild increase of +3-6 C, I will.. otherwise, I'll forget it.

However, another user (say Rooster) is fully confident in re-applying thermal paste, and is happy to push his rig to the max - so might settle with a +20-30% 3Dmark total, whilst monitoring temps. They might go for +10-30% mhz speeds on the chips, and might not mind a total +10-20 C increase, because they're confident that a) they can install better cooling and b) they can replace cooked paste when it happens.


See why my question was so apt? There's no point in OCing if you're ending up with lower scores...


[edited for correct labeling of stuff]

 

I agree. As long as you can get the quality and performance you want in games, then it is too much of a hassle to overclock. I honestly don't notice too much of a temperature difference with overclocking. I see more of a temperature difference with different drivers.

 

Actually, just to be fair to the ones that are true tweakers, I haven't pushed my system at all. Yes, I am confident I can re-paste my GPU but I am in no way wanting to do so on a regular basis. I haven't even come close to causing my system to crash nor have I had a system crash since I installed MSi Afterburner. I purposely use the same settings for Afterburner (CoreClock 712,ShaderClock 1424,MemoryClock 912) which has proven to be stable and in no way a large increase on the GPU. I also use this setting so I can have correlative information to compare my previous tests to. I only do this to offer unbiased information for we L701x users to refer to. I am also conservative when it comes to extending the life of my laptop. I only apply Afterburner when I am playing a game that will tax a graphics card, and most of the time I don't even have the application "turned on".
I mentioned re-pasting only because I noticed that the idle temp for my GPU seemed to be increasing and if the cause is not due to the driver then I probably will have to re-paste. But the kicker here is that the lowdown on the pasting the factory does seems to be in question and there has been reports that on average it's poorly done. But then again, I've read it takes over 200 hrs of running time for the paste to actually cure and become more effective so I may be jumping the gun on stating that I need to re-paste, but I thought it relative to mention it as a possible issue.
I understand the reason for your question, but I just wanted to clarify why I do what I do and not be misrepresented as someone I am not.
no harm done.
Rooster

 

Gotta agree with Rooster on this one, there aren't really many people in this thread that have gone the way of the true GPU tweaker. A true GPU tweaker would have squeaked out some ridiculous bench score more along the lines of 13-14k with the XPS 17. I didn't even go to the full extreme with my XPS 14 and I turned a 7k(stock) score into a 9.5k(gpu oc only) score. You want to see a true GPU tweaker go check out the M17x, X7200, etc threads and you'll see what really pushing the limits is. The reason I repasted is I've seen a 10 for 10 track record in the last year of botched paste jobs from Dell/Alienware. The stuff they use is more like molasses than TIM and is horribly applied in mass amounts to top it off. Even stock temps are improved on average of probably 10 degrees Celsius with an amateur paste job, which equates to parts lasting a lot longer. I can understand where you're coming from Cat in a Hat but in reality a properly tuned system lasts longer with better performance than the crap Dell puts out from the factory.

 

Hi guys, i had to do some exams recently, today i installed new nvidia driver 266.58, and did some 3dmarks...

this is old 266.44



and this is the new driver...



not very much..if temp stays at 50° i could return to 266.44,i don't like this... but let's try this stuff for now...

 

266.58 drivers on vanilla core speeds:







Not as bad as I expected - hitting 30 FPS on new DX11 titles will be possible if you don't max all tesselation / HD effects etc

 

my 266.58 results on different speeds..













the results were good but i noticed that during the tests there were more jaggies than usual hence the better frames i was getting..

also checked a couple of games.. weren't as smooth as 266.44 so i switched back. didnt notice much of a difference in the fan speeds just that the lappy was idling at 53-54c.

 

A baseline and another speed to add to the collection.

 

Suggestions? Tips, tweaks, GPU Frequencies? Really interested in

 

wow.. may I ask which driver you were using with those tests? i got a crash for 800/960 with 266.44 after 3 or 4 sec on the first test.. did a test with 750/950 and the results:




also its clear to me now that 266.44 delivers better quality compared to 266.58.

266.58 has better performance but alot more jaggies.

 

I used all the drivers at this speed and haven't noticed any change in score between drivers, each run varies too much. I had the original dell, then used the 266.44 but the score I posted is with the 266.58. I only noticed less lag in gaming with the 58 as opposed to the 44, but I think the 266.44 is a major step up from whatever dell had preinstalled. My computer doesn't crash first time around at this speed, but by the third or fourth time it will, the score also goes down with each run. This score is a first run at room temp. The highest I can run on 3dmark06 is 790 without ever crashing, but gaming has a ceiling of about 770 (ram stays maxed at 960). Heat makes a difference, I saw each consecutive run go down a little bit till i tried it in the garage. It was almost freezing outside and the garage was probably halfway to room temperature and after scores had gone down to 124XX from heat I got a 13090 when run at lower temps. I wonder what it would do with a little repaste.

 

LOL, kill all background applications to where you're literally running nothing but a striped windows.exe basically. Overclock to the sky if you're gpu is capable and run 10 times to actually get a run to go through lol. Overclocking is kind of an art that you have to learn by safe trial and error and a lot of experience with crashed runs and drivers. The other thing is you have to play with drivers as well, the 266.44 should be a good place to start if you don't have an i5/Optimus rig. PM me if you want me to answer questions.

 

How in the "bleep" did you do this? ....LOL
that is amazing....
have you had any crashes, stability issues?
you have the highest recorded score on this thread!
amazing!

sorry, you already answered but still amazing!
<!-- / message -->Rooster

ps for all the nickpickers , I am just curious for the history of this thread and am not recommending that everyone try and and "blowup" their pc...
just sayin...

 

Open question to all...
when using 3DMark products, what are or have been your experiences with the different packages offered and was/is it worth it?

The shareware version of 3Dmark6 has worked fine as an initial score producer but I would like to know what you, the readers and contributors, would suggest as an optional and dependable benchmark for testing the GeForce GT 445M graphics card.

I know that a suggestion has been made for the use of a different benchmark procedure, but I am trying to be within the norm of the general user, who's knowledge would be limited to the mainstream of information that's available at first glance, much as it was when I first started though the products available was way more limited then as to now.
Rooster

 

I admit you have my attention with this benchmark utility. I have only vaguely heard of unigene and was sadly unaware of it's utility. But as to be used as our mainstream results producer, it will take more requests for this test to be used. But I will experiment with it and suggest that others look into it as well.
Heaven Benchmark

Rooster

 

I ran Nvidia driver 266.44 to compare to Cat in a hat's run with 266.58 with the Heaven benchmark and did so with no alterations to the GPU nor even having the overclocking utility "on".
results as following:

shows that what I think will be the general opinion about the 266.58 driver as not being as good as the 266.44 and so the wait is on for the next best driver

Rooster
P.S
as to my previous statement about reverting back to the A01 dell driver to test the core temp difference, there was none and I went back to 266.44 and will probably in the future re paste for better heat reduction, though what I have now is not that bad.

 

I'm not a benchmarker myself, so no knowledge of other testing tools.

What I know, 3dMark6 tests the DirectX9 featureset.
Our gfx card is DirectX11.

As 3dMark6 will be a good way to compare gfx speed, I suspect that to have a good indication how new games will benefit of a driverupdate and/or OCing we should use a DX11 benchmark.

 

Well, I guess you hit the head of the nail that I was doing everything but ignoring and trying to find an alternative that made sense.
You, Xmyox, said it most simply and to the point.
I should be testing what we were given.
"the right tool for the right job" so they say...
I guess I was letting my wallet and my stinginess cloud my judgment of the very question I posted here not long ago. I have invested a good amount of my time as well as yours, the contributers and readers, compiling all this information for L701x users and I guess it's about time I do some financial investing as well. You can download 3DMark11 and use it once for free, but for repeatability, ya gotta pay. So I did. $19.99(US)...the pangs of buyer's remorse have already set in, but that's nothing new for me, as I will review and or cancel every purchase I have made for this laptop at least 5 or 6 times, and could extend over a period of a week to make the final decision.
I ran the first test run under the basic option(performance setting) with no changes or overclocking involved an here is the result for the 266.44 driver:

Now on to figuring out what that score means...
I did notice while the test was running that it was "chugging" alot, though the level of visualization was superb.
EDIT***********************************************
I ran the test again with Afterburner set on my standard settings (CoreClk 712,ShaderClk 1424,MemoryClk 912) with the results being:

The "chugging" was definitely less noticeable and the core temp was never above 65C, so overclocking has it's use.
The "P" in the score stands for "performance setting", one of the three "beginner's" settings you can chose from with out creating a custom benchmark, the others being "entry" and "extreme" . According to 3DMark11 records for this setting and laptop setup, this score is not bad, same for the first score without it being overclocked. hmm.
If time allows, I will post scores from the "extreme settings" before and after overclocking.

Rooster

 

It looks like the best way to compare is to use the graph. The stock speeds put you in the 1300s along with pretty much everyone else, then the OC put you in the 1500 with only three others, and only one got in the 1600s.

 

Can somebody show his furmark curve with the 266.58 only?

 

The only part of the graphic I took was just this

The core temp started @ 59C and had steady curve to and hovered around 70C-71C with no spikes and a steady line thru the 15 min test.
I hope that helps...
Rooster

 

Thats the drawback to this test, though I paid for it so I am going to use it if it kills me
Hardly anyone with our set up has submitted enough tests to really give weight to the benchmark. But I guess that just shows the newness of our setup. When I set up the search perimeters, they don't even have the i7 740QM processor listed, just the 820 and the 720 with no choice in between.
Question:
Has anyone found out why the drivers we have been using for 3DMark6 and now 3DMark11 are not "FM approved"?
Anyway,
I ran the extreme setting for 3DMark11...
Man...it was painful to watch...lol
it reminded me of the "flip-books" we played with in elementary school.
FPS averaged like 2. :laugh:
run with extreme no changes to GPU nor AfterBurner "on":

run with extreme with AB standard settings (CoreClk 712,ShaderClk 1424,MemoryClk 912)

sometimes, curiosity can be painful. There is no graph available, apparently I am the only one dumb enough to run this test for this setup.lol....

Rooster

 

Thanks Rooster...
What settings did you used for furmark?picturesize,heatmode?

 

good question...
let me look in my Post-it notes...

1024x768, extreme burn mode.
curious, why are you needing these?
for me the 266.58 really didn't work all that well, so I switched back to 266.44
Rooster

 

Hey guys question, I just updated my drivers and did an overclock and my GPU is running at 66C while gaming, is this too high for the 445m?

 

No, not at all. In fact that is very good.
You should be good upto ~90C

 

I agree as well, no danger with that temp.
Rooster

 

How much did you overclocked your card?

 

Thanks Rooster.I need it only for comparing.
Because your card was steady.

 

If anyone has a suggestion or request for me to retest one of our previous Drivers with 3DMark11 put in your requests here, and if enough are curious, I will retest.
Rooster