Video Card Thread (Mostly Gecube x1950xt)

Well just to chime in, the last Nvidia card I owned was a 8800GTX, and it's 3dmark score was about 11,500, while with a HD 3850 I am getting 10,000.

 

Nope, the new standard is the HD3870.
As all 8800GT fanboys state, the 8800GT is faster. The fact that it's only 5-10% faster in almost all games is irrelevant, simply the fact that it sits on top means it's god. if you think properly about it, that difference on 30-40fps simply isn't noticeable, and even when it is, I'd still consider it slight. However, I think having games look jagged or blurred regardless of what resolution you run them at is unacceptable, especially for a card that expensive (and by this I mean the fact that the 8800GT will not support DX10.1 and therefore will never be able to run Anti-Aliasing with DX10 titles). I like my smooth quality looking games. In this day and age, buying cards that render a shoddy image just shouldn't be necessary, not for frame rate gains you're never going to see.

As for the fans, the Arctic cooling fans with the rubber suspension are the ones I was talking about before. They are quiet, but because they push so little air due to an inefficient blade design, they work out no better than any other fans.
With the 900, yes it has 3 speed settings, but even at the lowest setting the 120mm fans run at 1200rpm, that's high for a silent PC, when you consider that the Nexus real silent fan runs at only 1000rpm, and most people undervolt them to reduce their noise further. I run my case fans at only 750rpm.
With regard to the noise from the extra devices, that's only if you get noisy ones. If you want a silent PC, you'll make sure you quieten every one of those sources. Buy a proper quiet Corsair PSU, get silent enclosures and/or anti-vibration mounting for hard drives, use a low speed silent CPU heatsink, get a fanless graphics card or use a Thermalright cooler so you can run it fanless, and if you buy a side fan, make sure it's a silent one. It's perfectly doable, because I've practically done it. All that remains is to get the Corsair power supply. Even with the Thermaltake one in my PC at the moment, the system is utterly inaudible from about 6 feet away. With the Antec case's fans on low, you have to be about 15 feet away before you stop hearing the roar of the air through those uncovered front fan vents.

And that's only an HD3850, let alone an HD3870. Cut the bull and buy a 3870, Maccer. Seriously. Why would I offer you false advice?

 

Also to chime in the drivers for the ATi cards aren't that good right now. Maybe later on it might even be FASTER than the 8800GT.

Anyway it seems you already bought it and I don't want to rain on your parade. So by all means enjoy, as the 8800GT is still an excellent card. Have fun and don't forget to post your 3Dmark score.

 

Well crossfire does well at some things, scales nearly 100% in some games but does really bad in others. When a GT outpaces the single 3870 so easily what is the point of a single 3870 purchase!? At least with the GT you can run Crysis in full details and get over 20fps!
Just look at the DX9 performance here;
http://www23.tomshardware.com/graphics_2007.html?modelx=33&model1=1060&model2=1057&chart=277
most games are 20% faster in this rig, (40% in 2142 in that link!) better and in the couple that the ATi is better in are by 1-2%, and the card shown there is not as fast as the SSC I'm getting and it's shown against the 825mhz ATi!
This review, although a little on Nvidia's side, has some other stuff and some CF tests too;
http://www.driverheaven.net/reviews/3870-XXX/testsystem.php
Tweaktown has closer results;
http://www.tweaktown.com/reviews/1222/1/page_1_introduction/index.html
but the 8800 is still the DX9 winner and best for Crysis. DX10 is a little more swings and roundabouts but the margins are marginally better for the 8800 on games it prefers, and it does better at the lower res 1280x1024 so I think I'll be happy with it.

 

It works better for crysis because the nvidia driver for the 8800Gt automatically detects Crysis.exe and scales down the image quality to improve FPS. Try renaming the file to anything else and you will see better quality and lower frames.

 

Yep, Crysis will look crap with the 8800GT since nvidia edited the drivers to make the game run faster, simply to show off higher benchmark scores. The render quality is terrible though, and it shows.

 

what would you rather choose: one of the best and fastest card of this time or a card that produces the cleanest, most crisp graphics and is more future-proof? because that's how it looks like between the two cards.

the HD3870 may be slightly slower, but the graphics quality makes up for the speed. there's always a compromise.

 

Maybe I should change the name to "Gecube x1950xt and 8800GT Vs 3870 Thread"?

3870 FTW!

I feel like nvidia cut some corners with there new drivers (the Crysis performance vs quality) and therefore might have lost some customers. I think that if nvidia didnt do that, the 8800GT would perform the same as a 3870, if not less.

 

Change it to the video card thread. God knows we need one lol.

 

Haha you changed it!

 

yep, I did

Anyways, I was playing around with RivaTuner and noticed a "AGP Fastwrites" option with "on" or "off", and it was set to off. I looked around the web, and found that it basically makes the card run faster (in short).

Not to sure how much faster as I havnt really gotten a chance to play some games yet, but I'll find out!

Can you clarify this?

 

That's kind of the problem though, I don't think they really have lost many customers, when they should have, because that compromise is unacceptable.
I spotted the AGP fastwrite option back when I had an X800, I had no idea what it did, and I still don't now...

 

this is what i found

 

BTW, I was just reading up on the 3870x2 that is going to be out in January, and was thinking this would be better than the my original 2x 3870s.

Question: I plan on getting 1 3870x2, but if I ever wanted 2, would the Maximus support it? I mean, the Maximus supports regular CF, but not Quad CF, even though this would be regular CF since it's just 2 gpus on 1 PCB.

Also, is it worth the wait? With Intels next gen CPUs being released as well, I think I will just see what happens...

btw, heres some stuff on the 3870x2:
http://www.theinquirer.net/gb/inquirer/news/2007/11/17/amd-showcases-radeon-hd-3870-x2
Shown in CF:

Quad CF:

Card without cooler:


EDIT: More stuff: http://www.fudzilla.com/index.php?option=com_content&task=view&id=4251&Itemid=34

Finally, will this card work in games that dont support CF, even though technically its not a CF card but does have 2 GPUs?

 

As for your board not supporting Quad CF, you'll be fine, just don't use two 3870X2s. As for crossfire support, this is one of the big reasons I never recommend crossfire. When SLi isn't supported in a game, you just won't see a difference. When Crossfire isn't supported in a game, the game will usually crash, or you will get millions of rendering errors. This will still apply to crossfire-on-one-card solutions like the 3870X2. You will need to disable crossfire and use only half the card to run games that don't support CF (i.e. most of them)

 

Wow, thats dumb. I don't see why ATi and nvidia would even bother making cards like that then.

Do you think their drivers will help so that CF support is better on a wider verity of games?

EDIT: So if I were to use only half 3870x2s speed, then I would basically get the same results as a regular 3870 right? So then this would be a better buy, as I believe they will be between $350-$500, and it costs exactly $500 to crossfire 2 3870s, not guaranteeing CF will work.

 

Like I say to people, crossfire hasn't improved in that regard for more than a year, so I wouldn't hold your breath.
Personally I think a single 3870 would be a better buy.

 

Do you think more future games will have crossfire and SLi support?

Anyways, getting back to the Gecube since I still have mine

I ran the Crysis benchmark test a total of 3 times each, with all high settings @ 1024x768 with shadows on low and physics on medium.

The first run was with the mysterious AGP Faswrites on, second was without it. Third was with fastwrites also off and a 702/828 overclock.

12/13/2007 10:44:49 AM - XP
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1024x768, AA=No AA, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: Custom
Custom Quality Values:
VolumetricEffects=High
Texture=High
ObjectDetail=High
Sound=High
Shadows=Low
Water=High
Physics=Medium
Particles=High
Shading=High
PostProcessing=High
GameEffects=High
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 195.74s, Average FPS: 10.22
Min FPS: 2.96 at frame 691, Max FPS: 19.49 at frame 1753
Average Tri/Sec: -9092410, Tri/Frame: -889867
Recorded/Played Tris ratio: -1.03
!TimeDemo Run 1 Finished.
Play Time: 135.07s, Average FPS: 14.81
Min FPS: 2.96 at frame 691, Max FPS: 20.55 at frame 989
Average Tri/Sec: -13061450, Tri/Frame: -882120
Recorded/Played Tris ratio: -1.04
!TimeDemo Run 2 Finished.
Play Time: 135.79s, Average FPS: 14.73
Min FPS: 2.96 at frame 691, Max FPS: 20.55 at frame 989
Average Tri/Sec: -12986067, Tri/Frame: -881710
Recorded/Played Tris ratio: -1.04
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

12/13/2007 10:44:49 AM - XP

Run #1- DX9 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 14.77


---------------------------------------------------------


12/13/2007 10:56:54 AM - XP
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1024x768, AA=No AA, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: Custom
Custom Quality Values:
VolumetricEffects=High
Texture=High
ObjectDetail=High
Sound=High
Shadows=Low
Water=High
Physics=Medium
Particles=High
Shading=High
PostProcessing=High
GameEffects=High
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 195.11s, Average FPS: 10.25
Min FPS: 5.45 at frame 270, Max FPS: 20.18 at frame 991
Average Tri/Sec: -9161329, Tri/Frame: -893746
Recorded/Played Tris ratio: -1.03
!TimeDemo Run 1 Finished.
Play Time: 136.13s, Average FPS: 14.69
Min FPS: 5.45 at frame 270, Max FPS: 20.44 at frame 987
Average Tri/Sec: -13049589, Tri/Frame: -888221
Recorded/Played Tris ratio: -1.03
!TimeDemo Run 2 Finished.
Play Time: 136.29s, Average FPS: 14.67
Min FPS: 5.45 at frame 270, Max FPS: 20.44 at frame 987
Average Tri/Sec: -13028278, Tri/Frame: -887795
Recorded/Played Tris ratio: -1.03
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

12/13/2007 10:56:54 AM - XP

Run #1- DX9 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 14.68

---------------------------------------------------------

12/13/2007 11:07:56 AM - XP
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1024x768, AA=No AA, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: Custom
Custom Quality Values:
VolumetricEffects=High
Texture=High
ObjectDetail=High
Sound=High
Shadows=Low
Water=High
Physics=Medium
Particles=High
Shading=High
PostProcessing=High
GameEffects=High
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 186.08s, Average FPS: 10.75
Min FPS: 6.29 at frame 270, Max FPS: 23.61 at frame 984
Average Tri/Sec: -9588595, Tri/Frame: -892099
Recorded/Played Tris ratio: -1.03
!TimeDemo Run 1 Finished.
Play Time: 122.69s, Average FPS: 16.30
Min FPS: 6.29 at frame 270, Max FPS: 23.66 at frame 85
Average Tri/Sec: -14408025, Tri/Frame: -883827
Recorded/Played Tris ratio: -1.04
!TimeDemo Run 2 Finished.
Play Time: 135.19s, Average FPS: 14.79
Min FPS: 6.29 at frame 270, Max FPS: 23.90 at frame 81
Average Tri/Sec: -13091521, Tri/Frame: -884894
Recorded/Played Tris ratio: -1.04
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

12/13/2007 11:07:56 AM - XP

Run #1- DX9 1024x768 AA=No AA, 32 bit test, Quality: Custom ~~ Overall Average FPS: 15.545

 

As I've tried to say on many occasions, I think that's doubtful. SLi perhaps, but Crossfire probably not. Realistically I think it's pretty much going to stay the same way as now until a new system is invented.

 

Just a final note on the 3870 vs 8800gt check out this review;
http://www.vr-zone.com/print/5392
they have some shots re: picture quality in crysis. I must concede that the gt does look a little washed out compared to the 3870 which does eem more vibrant, but really the difference is not that noticable, and for an extra 5 fps I'll have the GT.
It seems also that availability of the 3870 is even less than the GT too, so there'd be no hope of me getting one before xmas if I did choose that one anyway!
I'll be back with a 3dmark for everyone soon (I hope)