Here it is, GTX 285M SLI benchmarks

Scook, it does seem that the 285M are weaker in synthetic benchmark but they do seem to be faster in some games as per Lozz bench...

The $50 premium is still not justified. The nvidia cards should be cheaper than the 5870...

 

Like I said, get them if you play one of the few games that lacks ATI driver support, otherwise they will not be worth it.

I just do not think the cards deserved the hype they received here is all...they are nothing new at all and bring nothing impressive to the table

 

I think that has been known and Lozz was aware of. It was just those guys who was about to purchase their m17x put it to hype.

 

I didn't realize there was any hype? the 285vs 5870 is full of me explaining that the 285's are not in any way 5870 crushers, and will probally be about as good as 280m's in the R1 and the 285's in the clevo's were.

I was reading from the notebookcheck page, which mis-read nvidia's page, they acutally post 576mhz on notebookcheck.

 

It just seemed like a lot of people were overly excited for these (not saying you). I fully agree with your expectations as stated above and in the other thread

 

o yeah.. but new stuff always excites people regardless if it's better or not.
edit;
here's a 5min furmark run

72C, I'll do the full postfx extreme burn shabang after I get back from the store.

 

And thanks to dell..."if it costs more it must be better!"

lol

 

I agree with you Scook and in no way I was stating that people should get the 285M over the 5870.

I think the next big hype will be the 6870 which may come sooner than some people might expect....

 

Then count me a "some people" haha

I am just not holding my breath, the 5870s are not even mature yet. If they do a 256-bit GDDR5 card I will be truly impressed, the 5870m was just too much of a compromise to make a laptop part (something nvidia would do )

 

well rumor say the desktop part that should correspond to m6870 is a 256 gddr5 card and i suspect the laptop variant to be in the performance range of a deskto 5830

 

That would mean all they are really doing is making the 5870m a 256 bit bus....that is all that is keeping it from 5830 performance

 

then it might be even faster as for now what is expected is a a bit downclocked bart to be ported to mobility as m6870

 

where are these rumors? I haven't seen any press at all on the mobility 6000 series gpus.

 

Lozz, can you see how BC2 and the new Medal of Honor play on your laptop if you have the games? Thanks!

 

right here http://forum.notebookreview.com/gam...0727-ati-mobility-hd-6000-series-roadmap.html

 

sorry, I have neither, the last one I will be able to do is lost planet, as I'm just not sure how to use the crysis benchmark tool.

I'm begining to believe 258's are unstable, I first had a total loss of screen power, and about 10 minuites ago I had a random total system freeze. I know R1 people have had the same problem and that there was a thread where someone suggested getting drivers from some mobile website that alters the nvidia mobile reference drivers, but I'll have to look for it later.

 

Have a kill-a-watt meter? The GTX 285m's pull more power than any GPU the R2 has seen so far. And the i7 quads pull more power than the qx9300.....could easily be hitting the power limit of the notebook

 

The performance of GTX 285's seems to be good to me, i really don't care about benchmarks, i mean you really don't play any benchmarks, they are actually bored, always the same.... In games NVIDIA is always better, if don't just try to play Batman Arkham Asylum w/ PhysX at high, or Mirror's Edge, Dark Void... Try to use some softwares that use the CUDA and get huge performance improvement, ok you can't with AMD's. I am just saying that benchmarks is only a small part of what a video card is, and ok, this part means just a number to me, it doesn't give any fun.

So go GTX 285M !!

 

I'm sure it's not hitting the power limit while I was surfing the web, in both cases I wasen't doing anything remotely cpu or gpu intensive. Getting the same graphical distortions I was getting in SCII on the ATi 10.7/10.8 reference drivers, so I'm just going to say the reference nvidia drivers are not stable at this point. I might nab a kill-a-watt meter though, they're on sale @ radio shack.

 

Obviously games that leverage PhysX will do better with the GTX 285's.....most games do not though

 

Ya they are great to have. That is odd though, that it was when under no load. If all else fails....call dell lol

 

if it still does it on 197.52s I definitally will. That would mean I haven't had a working computer since the end of june.

 

Thats what you think! haha lol. I'm quite the opposite. I'll spend all night benching. The game is to out do everyone else. And when you finally do, feels pretty great. lol, I have a ton of games, havn't played hardly any of them. But when I get home I might, if real life allows me time.

 

I have the same thing

 

benchmarking is fun, until you realise you spent $3000 in parts on an invisible number that will be beaten by the next 14 year old punk with the same hardware and a lucky streak.

 

I still enjoy benchmarking and am no longer current gen tech. I game too.....I am finally starting to game more than bench again....first time in a while lol (during school I gamed, breaks I benched - now I am graduated though so actually choose....)

 

Damn so it looks like these 285's are a fail. IMO the best and most stable M17x ever released was the 4870m Crossfire configuration. Once the initial bios hiccup was fixed, they remained rock stable and able to take any reference driver without issue. The 5870s are almost as good now with the new vbios + drivers but they will again become problematic when the Catalyst code base changes. Personally, if someone wants an M17x that's 100% stable, I'd recommend the 4870m Crossfire configuration and it performs only 10-15% less than the 5870s which is negligible in terms of in game FPS.

 

That is pretty much my conclusion as well. Which is also why I am not so eagerly hurting to get mine swapped up to an R2. I like battery life

 

But maybe some day later, we'll get that stable on 5870m Xfire and with a little bit performance more, it's a better hope right?

 

maybe.. but dell never addressed reference driver compatability on the R1, and they're stuck with that forever now.

 

reference driver compatibility has been great for me with the 4870s and the GTX 280m's.....

 

I've seen more than a few threads about reference drivers not working right with 280s in the R1.
like this
http://forum.notebookreview.com/alienware-m17x/524462-suffer-sli-low-performance-issue.html
and this, granted.. he's using beta mobility drivers for whatever reason..
http://forum.notebookreview.com/alienware-m17x/524025-just-about-had-nvidia-drivers-2.html

I'll try a few more reference drivers before I'll say whether they're another 5870 case but that won't be for a while.

 

Well up to February when I got the 4870s the GTX 280m had no issues

 

nvidia refernce drivers are definitally not stable in the r2, infact, the only one that even recognizes the graphics cards are the latest whql's.

also found out why 285's run so cool, Dell seems to have fixed the R2 HSF. There are gap pads on all the moffsets, and the power regulator chips as well, where as the 280M's only had it on the power regulators on the right side of the card. The 285's are packed with gap pads, and with the shim the hsf is nice and level with the card. I'll nab the size of the fat pieces later, the ram pads look like the normal 1.0mm stuff.

 

Had I have knowna bout these GTX 285m's coming out, I wouldn't have got the 5870m's because:

- Screen scaling issues with Ext Monitors
- Nothing exclusive like PhyX
- Never had it good with ATI

My first and currently, only nVidia card was the 9600GT. That was a beautiful piece of hardware.

Also what temps do these 285's get?

 

For whatever reason furmark will not work in SLI mode when extreme burn mode or post fx is used, so I was able to only run with displacement mapping on, that was on my cryo, i'll get one on the counter in a second.
edit; off the nztx produced nearly identical results, the secondary card managed to eek up another 4-5c, but nothing special. So in closing, they run essentially as well, and probally a bit cooler on stock hsf's compared to 5870s.

 

So they are only a little warmer than my 4870s, that is not bad at all

With FurMark My cards hit 68c

 

yeah, all that extra padding did its job, hope I can find some performance material that thick however. it had to be at least 1/4"

 

.25"=6.35mm

That is indeed thick thermal pad.....there are similar thick pads on my GTX 280m's but I do wonder how much they actually help it at all

The main things are the RAM and GPU itself

 

might have just enough ek water block material and AS5 left to do both cards tomorrow, we'll see. going to nab a killawatt tomrrow at the least though.

 

I need to do the cooling mods for my GTX 280m's

I want to finish up with the 4870s first. Still need to work on a couple benches before making the swap (and been benching less than I used to so going slow)

 

what voltage do your 280s run @?

gpu-z is telling me 1.01V 3d clocks and .85V idle.

 

I believe they were 1.0V stock. The vbios would allow upto 1.03V without being modified. That is why I was amazed to see that the 4870s were 1.1V stock lol

I took my GTX 280m's as high as 1.12V trying to get the most out of them that I could. It sounds like nvidia bumps the vcore up by .01V to keep stability over the GTX 280m speeds

 

Hi

Regarding Crysis benchmark tool, could you try it with 197.84 dell drivers (the most stable at the moment), with all settings on high on 1920x1200, with dx10, 3loops,

cheers

 

I'll try to figure it out for you today, in the mean time..

There was acutally a 1.0mm pad on the biggest moffset, so I applied one there too, in the pictures I had already removed the dell gap pad on it already however. There are a total of 3 different sized pads, but thankfully they had enough 'spring' that they're re-useable. I may still change them If I can find something that's cheap enough.

 

Bear in mind, this is with the 45W 740 CPU, everything at stock clocks, and the 940XM is a 55W CPU when it's turbo boosting as Intel wants it to (but as we know, it doesn't in the M17X R2 )

idle, sitting @ the desktop after 2-3minuites, this value fluxuated ~ 5W, peaking @ 70W or so



Full CPU load- surprisingly, the 740 hit it's 45W TDP value right on the mark, this was done running 8 threads in prime 95, on the in-place larget FFT's setting. This didn't really fluxuate very much, maybe 1-2W

Next, using furmark I extracted the gpu loaded result, furmark uses a little bit of cpu power during this test, but the nvidia demo I was using used the same amount and couldn't load the gpu's as much as furmark would, so this is the highest result I could produce. This value flucuated ~ +/- 10W


Next, using furmark again, I ran 8 threads large FFT prime 95, again, this value flucuated by ~ 20W, however this is about as high as I could get it, remember, this isn't even with the XM processor, so we could add another 20W or so since it doesn't clock throttle when all 4 cores are in use, based on this score we can plainly see that some of the previous value has the CPU power in it


bottom line; GTX 285's are power hungy, but it would appear that they don't draw a whole lot more than what they're ratted for; 150W. Obvisully though, either the flextronics 240W is very much under-ratted or I'll be going through 1 every 6 months or so.

I'm going to go try to nab some crysis numbers and monitor the meter to see what average gaming should produce.

 

Also give the OCCT Power Supply test a go and compare it to the furmark/prime95 ones

Aren't kill-a-watt meters fun?!

lol

 

here's the crysis run as promised

I'll run it in a few minuites, but lol yeah, it was worth the $30

 

Can you do Crysis again with very high?

I can compare to my 4870s then
http://i50.tinypic.com/2qd8j1y.jpg

 

sure thing, heres the killawat on crysis, 199 was the highest spike, it usually hovered around 185-190


OCCT3.0 was interesting, this was on the power supply test, it would peak to 270-280, and then dip back to 180.. 280W was the highest spike I monitored, or dared to monitor rather.


here's you screenie scook