Hey DEXGO quick Questions

I expected you to get better scores, something over 30 anyway. It's been a while since I've had an SLI machine, but I think you should set "use application profile" in nVidia Control Panel. If you've already messed with the application profile for Crysis, restore it to it's default and give it another try.

 

Those are old drivers but they shouldn't make that much difference. (Maybe 1-2fps)

 

get newer drivers from sager or modified ones or simply use the same ones that dexgo has.

patch the game up to latest build.

re-instal your nvidia drivers if you tampered with ANY of the settings. once the drivers are installed you should only enable sli only and do nothing else. you should not even view the profiles for games as again the buggy nature of laptop drivers may screw something up.

run the benchmark in high settings in wuxga and wsxga at all high settings with a non-modified game.

post results

-i am waiting for my d900c to ship so cannot remember where the sli bar would be enabled as i currently do not have an sli machine. but i know it is one of the tabs on top when you are in the nvidia control panel.

oh, the benchmarks appear relatively normal. if you have not patched to latest build you will get a few more fps. i am very curious to see what you get in wsxga resolution as dexgo claims about 33 fps with his single card.

 

"re-instal your nvidia drivers if you tampered with ANY of the settings. once the drivers are installed you should only enable sli only and do nothing else. you should not even view the profiles for games as again the buggy nature of laptop drivers may screw something up."

HUH??????

 

when i had my alienware m9750 with 7950gtx sli i tried alienware's drivers and a bunch of modded ones and i can tell you that sometimes if you alter a pre-set custom profile for the game it would permanently lose it's setting due to bugs. only fix was to re-instal the drivers. some drivers are so buggy that profiles don't even show up even though sli will work fine.

bottom line is when you instal the driver, it has custom pre-sets (directly from nvidia) for the majority of the games out there and you should not tamper with them. the more drivers mature the better support for sli with games becomes i believe, even if a game is not coded for sli, like supreme commander. i saw an 82% fps increase in the game's benchmark and in-game compared to single gpu mode. i got 43 fps average with wuxga and all settings maxed with no aa. my brother's desktop rig with an 8800 gts 320 managed 37 fps average. but the catch with that game (due to the buggy nature of sli drivers) was that i was forced to run with vsync enabled otherwise the fps crashed to like 2 fps for no reason and then recovers. enabling vsync prevented this from happening but i took a serious performance penalty with it. my benchmark went down to 33 fps average from the 43 and i got really fed up with sli in general because of it. i then decided that i need the power of the single 8800 gtx with the added bonus of it being in sli when and if my games supported it properly. so i look at sli as a bonus instead of a requirement because of the finicky driver nature that sli users fight with.

alternative frame rendering "2" (AFR2) is the most common sli setting, not AFR1. you should only start tampering with settings if you are experiencing stuttering, fps drops, or other problems with sli enabled. maybe crysis likes the AFR1 mode, i don't know.

this site has alot of knowledge for sli. i learned almost all my sli from there.

http://forums.slizone.com/index.php?act=idx

 

Okay I put the 174.33 drivers ..... I enabled sli... And updated crysis to 1.2....

Here are the new scores
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 75.50s, Average FPS: 26.49
Min FPS: 6.95 at frame 138, Max FPS: 35.27 at frame 859
Average Tri/Sec: -17115546, Tri/Frame: -646143
Recorded/Played Tris ratio: -1.42
!TimeDemo Run 1 Finished.
Play Time: 64.23s, Average FPS: 31.14
Min FPS: 6.95 at frame 138, Max FPS: 36.96 at frame 71
Average Tri/Sec: -20072002, Tri/Frame: -644612
Recorded/Played Tris ratio: -1.42
!TimeDemo Run 2 Finished.
Play Time: 64.08s, Average FPS: 31.21
Min FPS: 6.95 at frame 138, Max FPS: 37.31 at frame 989
Average Tri/Sec: -20123190, Tri/Frame: -644731
Recorded/Played Tris ratio: -1.42
!TimeDemo Run 3 Finished.
Play Time: 63.57s, Average FPS: 31.46
Min FPS: 6.95 at frame 138, Max FPS: 37.31 at frame 989
Average Tri/Sec: -20279112, Tri/Frame: -644560
Recorded/Played Tris ratio: -1.42
TimeDemo Play Ended, (4 Runs Performed)
==============================================================
Press any key to continue . . .

 

you got a 5% boost in fps. is this with your cpu overclock? is y our gpu overclocked as well?

can you do a run for wsxga please?

 

no overclock on the gpu....

I ran the gpu benchmark on crysis

what do I need to do to run a wsxga ?

Sorry I don't know

 

That's a nice bump up over your previous score. The single card rigs get 18-19fps in that test, so you're 65-75% higher with SLI. I don't know if that's the best the Sagers can do, but that seems pretty good to me.

You can compare your results with Justin's recent results for the Q9550 CPU with the 174.82 drivers here: http://forum.notebookreview.com/attachment.php?attachmentid=17221&d=1207345149

Aside from the fact that your average FPS is 10% higher, I notice that your average is closer to the maximum (84% vs 78%). This suggests to me that you have fewer occurrences of very low FPS periods, even though your minimum FPS is very low.

It would be interesting to see what 3DMark06 scores with this new driver.

Good work.

 

Thanks Wobble

 

If it is a non-overclocked score then yes there is a significant difference from the single card. If is is overclocked though then it is only 40% faster (and not 65-75%) than my score (22.48fps).

And Justin's score is only 27.8% faster

 

wsgxa = 1680 x 1050 resolution. i am curious to see what you get compared to dexgo's claims of 35 fps with his single card in this res.

wuxga = 1920 x 1200 resolution.

 

CesOne's machine has stock GPU clocks. The 65-75% increase I cited is in comparison to a stock single card 9262 running XP.

A stock single card 9262 running Vista (as Justin was) seems to get 17-18 fps in this test. Since Justin got 28.74fps in SLI, that seems to be roughly a 60-70% increase. I guess the 27.8% number is relative to your machine. If so that's interesting, but somewhat an "apples and oranges" kind of thing.

 

That's why I asked in my post if his scores were overclocked or not.
BTW I love apples but oranges no so much

 

1680 1050 was not on my resolution menu... so i did it on 1600 1024

Here are the scores ARGH


==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 64.36s, Average FPS: 31.08
Min FPS: 7.78 at frame 139, Max FPS: 45.39 at frame 906
Average Tri/Sec: -22158840, Tri/Frame: -713035
Recorded/Played Tris ratio: -1.29
!TimeDemo Run 1 Finished.
Play Time: 53.27s, Average FPS: 37.55
Min FPS: 7.78 at frame 139, Max FPS: 47.24 at frame 1003
Average Tri/Sec: -26698804, Tri/Frame: -711099
Recorded/Played Tris ratio: -1.29
!TimeDemo Run 2 Finished.
Play Time: 52.88s, Average FPS: 37.82
Min FPS: 7.78 at frame 139, Max FPS: 50.30 at frame 93
Average Tri/Sec: -26850400, Tri/Frame: -709991
Recorded/Played Tris ratio: -1.29
!TimeDemo Run 3 Finished.
Play Time: 52.52s, Average FPS: 38.08
Min FPS: 7.78 at frame 139, Max FPS: 50.39 at frame 75
Average Tri/Sec: -27092692, Tri/Frame: -711437
Recorded/Played Tris ratio: -1.29
TimeDemo Play Ended, (4 Runs Performed)
==============================================================
Press any key to continue . . .


I did not change the resolution on my notebook just on the game...

 

is 1680 x 1050 an option for your desktop resolution? i really hope the new sager drivers does not limit resolution choices. when i had the alienware m9750 the stupid alienware sli drivers only had like 4 resolution choices it was so annoying i could not stand it.

there is a way to create your own resolution in nvidia control panel so it looks like there is a way around it. i am eagerly waiting for my d900c to be delivered so i can see for myself. i might play the game in 1280 x 960 or 1440 x 900 in dx10 mode with very high settings. curious to see what the benchmarks would be. i'll find out this tuesday.

38 fps average is very playable for crysis with stock gpu clocks. sli 8800 is looking better and better.

 

Glad I could help

 

CesOne,

Your score at the lower resolution is a little over 50% higher than my score with a single card (38fps versus 25fps). Obviously that's a pretty big falloff from the 65-70% advantage at the higher resolution. The advantage of SLI falls off with lower resolutions, of course, but I don't know if the falloff you got should be considered normal or not.

ARGH,

The nVidia drivers provide tons of resolutions, and, as you pointed out, custom resolutions can be added. The problem is that neither Crysis nor the separate Crysis benchmark program seems to pick them up. However, there are different resolutions in each, so if you have both, you can use practically any resolution you want.

 

xp 174.31
single card
600 - 1600 - 1000

I finally figured out the VID MAP and managed to create my own for the .14 bios and it actually let me put in 1.1v

so I am on my way now

no freezing or anything

 

1900x1200 all high

notice that I am averaging 24fps in the benchmark tool now

the fraps didn't have time to catch up.

but if you look at the crysis benchark avg in the top left you can see.

 

Nice scores, dex! Your CPU tweak really put some wheels under the quad!

Why did you raise the voltage from 1.05 to 1.1...stability?

When are you getting your second 8800?

 

yes, the 1.1v makes it more stable for higher ceiling clocks.

I am going to get my 8800 soon.

 

Nice scores Dex. 16k is not that far off now.
If only I had my damn machine...

 

38fps is indeed an excellent score. 42% faster than my overclocked single card (26.71).

 

respekt dexgo man...come to amsterdam and bring the 7900 wit u forkin yenks lol