My Far Cry and Splinter Cell Benches 9700Pro -> x800Pro

[RussianSensation]

Originally posted by: Acanthus

Did you expect it to be slower with a card 2 generations newer?

Yeah, of course -- I am introducing a new term - Videocard Inflation. What I think i was trying to say was that I didnt know if my 3.2 ghz p4 would be fast enough to tax the x800pro (and I dont want to spend $399 to have a cpu limited rig). As you can see all resolutions besides 1600x1200 for him are same performance. So I was thinking of just getting a 9800Pro but since I'd love to play at 1600x1200, it seems the cards do get used up pretty well so now it's not a bad idea.

 

[fritzdog]

UPDATE:
I posted my findings from the first post on the official Far Cry forum as well. It was pointed ou that my framerate with the 9700 seemed low. After looking at the scores others are getting with similar systems it seems as if that is true. Last night I reinstalled the Catalyst drivers that came on the CD. I still have not checked the version number. I then retested the x800 Pro in Far Cry using the Bench?em All tool.

Far Cry with CD Drivers
Using Max settings.
Map: Fort Demo: BenchemallDefaultDemo
640x480
run# 0: Average FPS: 57.25
800x600
run# 0: Average FPS: 57.22
1024x768
run# 0: Average FPS: 55.89
1280x1024
run# 0: Average FPS: 41.76
1600x1200
run# 0: Average FPS: 37.40

As you can see the lower resolutions were roughly the same but the higher resolutions dropped off significantly. I then discovered that the Cat 4.5?s were out so I installed them.

Far Cry using Cat 4.5
Using Max settings.
Map: Fort Demo: BenchemallDefaultDemo
640x480
run# 0: Average FPS: 64.95
800x600
run# 0: Average FPS: 63.14
1024x768
run# 0: Average FPS: 63.52
1280x1024
run# 0: Average FPS: 41.97
1600x1200
run# 0: Average FPS: 37.41

As you can see the lower resolutions increased a decent amount but the higher resolutions stayed the same.

The above was tested with the PC2700 Ram still in the system. I installed the Mushkin Level 1 PC3500, set the BIOS to DDR400 and CAS 2.0 and got the following results:

Far Cry Cat 4.5 with PC3500 Ram
Using Max settings.
Map: Fort Demo: BenchemallDefaultDemo
640x480
run# 0: Average FPS: 61.99
run# 1: Average FPS: 64.10
run# 2: Average FPS: 64.45
800x600
run# 0: Average FPS: 63.34
run# 1: Average FPS: 64.22
run# 2: Average FPS: 63.21
1024x768
run# 0: Average FPS: 62.65
run# 1: Average FPS: 63.92
run# 2: Average FPS: 63.01
1280x1024
run# 0: Average FPS: 42.45
run# 1: Average FPS: 42.63
run# 2: Average FPS: 42.25
1600x1200
run# 0: Average FPS: 37.39
run# 1: Average FPS: 37.45
run# 2: Average FPS: 37.44

Splinter Cell 1.2b
Demo name: 1_1_1TbilisiDemo.bin
800x600 Average= 63.950 fps, min 33.659 fps, max 97.465 fps. Effects quality= VERYHIGH
1024x768 Average= 62.915 fps, min 36.173 fps, max 97.435 fps. Effects quality= VERYHIGH
1280x1024 Average= 55.092 fps, min 32.247 fps, max 85.741 fps. Effects quality= VERYHIGH
1600x1200 Average= 49.810 fps, min 26.504 fps, max 87.798 fps. Effects quality= VERYHIGH

As you can see there was no real increase in framerate in Far Cry. Splinter Cell went down significantly from the scores in my first post.

My 3Dmark03 scores are as follows:
AMD64 3400+, 9700Pro, 1GB PC2700: 3840
AMD64 3400+, x800Pro, 1GB PC2700: 8479(CD Driver) 5954(Cat 4.5)
AMD64 3400+, x800Pro, 1GB PC3500: 5970(Cat 4.5)

I only ran 3DMark once while the CD drivers were installed. Needless to say I was thrilled at the outcome. Testing with the 4.5s then was a disappointment, especially after an improvement in the Far Cry test. My 3Dmark2001 score with the x800Pro/CD driver was 21738. I did not test this again with the 4.5s.

All the testing I have done has not been very scientific. As I said in my first post I did not shut down any background tasks. Perhaps I?ll try shutting done those items and see if there is any noticeable improvement.

For me I?m very pleased with the improvement of the x800 over the 9700. My expectations were met. What I need to do is try to figure out what is holding the system as a whole back and why I'm getting such inconsistent results. I?m very happy that I can now play Far Cry at a much better resolution, maxxed out graphic settings and still get better framerates than with the 9700pro. The 9700pro also had a ram problem that caused graphic glitches in windows. It never affected gaming but sometimes I?d have to reboot the system 2-3 times before it would boot up clean. So now I do not have to deal with that anymore.

 

[fritzdog]

Regarding the Splinter Cell drop. Is this game OpenGL? I Think it may have been affected by the OpenGL settings in the Catalyst Control Panel. I was also testing IL2FB in perfect mode to see if the 4.5s really did fix it(Not really: 7 FPS ->9) and the splinter cell test in my first post was probably set at 2xAA and last night it was definitly set at 4xAA.

 

[Regs]

Originally posted by: BFG10K

Once you get Pc3200 or PC3500 , you'll likely be getting 65+ FPS. Your ram is crippling your system right now.

PC3200 will provide 5%-10% gain while PC3500 will make no difference over PC3200.

Yes. Just which ever one floats his boat, doesn't make a difference.

 

[Curley]

Fritzdog,

Are the last set of scores in FarCry with the X800 or the 9700Pro?

What most people don't understand (including me) about the graphics settings in FarCry is that you must leave your video card settings to default settings. Then set the game settings to your desired level of play, AA, AF, etc... You will get the best eyecandy/gameplay results from this according to Ubisoft.

What I have found in Splinter Cell is that you do have to apply your graphic setting in the Card Properties and no the game to get AA, AF, etc... So it is the oposite of FarCry.

What I like about the Nvidia Drivers, is that they now come with preset game setting in the card properties.

With the ATI, you will have to make your own presets. The days of Voodoo2 are gone. One setting will work great with one game and look like crap on another.

Let me know on those last benchmarks if it was with the X800 or the 9700Pro.

 

[fritzdog]

All the benchmarks in my last post were on the x800. At this point I'm not interested in reinstalling the 9700 to retest, although it would not really be that hard to do so since it uses the same drivers as the x800.

I have the D3D settings in the control panel set to application preference and created a profile that way for Far Cry. The thing that I do not know for sure is that I think when I ran the benchmarks for the x800 in my first post the water was set to Very High in game. Last night it was set to Ultra High. When I run Benchemall I do not really know what it is changing to be considered MAX. I may retest tonight with the in game setting to Very High to see if that is affecting my results from last night.

ANother possiblity is that when I installed my MSI mobo and AMD64 CPU a couple weeks ago I did not do a fresh install of WinXP. Everytime I have upgraded these components in the past I have always had to do a clean install. This mobo installed without a hitch so I did not bother. Since I have now upgraded the 4 primary components in the system I should probably go ahead and reinstall the system from scratch to make sure that there are no conflicts anywhere. My old mobo was an nforce 2 Asus A7N8X Deluxe.

 

[Curley]

Thanks,

I can't wait until I get the X800XT PE.

 

[Pete]

FYI, Curley, AA doesn't work correctly in Splinter Cell, because of the way it performs some effects (same with Prince of Persia and NFS:Underground, IIRC). You're right that the preferred way to enable AA/AF is to set the Control Panel to Application Preference and enable them using the game's settings, when possible.

 

[Curley]

Pete,

Thanks for the info, I was just sitting here scratching my head trying to figure out what AA looked like crap in Splinter Cell.

 

[BFG10K]

A lot of console ports either render to a texture or don't render the entire scene from scratch, only the parts that have changed from the last scene. As a result standard control panel AA and/or AF doesn't work well and the games need to implement their own versions.

Yes. Just which ever one floats his boat, doesn't make a difference.

Somebody made the claim that his RAM was crippling his system when that simply is not the case. AMD processors have historically been less dependant on memory bandwidth than P4 systems and the A64 is no exception.

 

[ronnn]

I am confused, my version of Far Cry has only Very High - no ultra available? Oops found it.

 

[Curley]

Did you apply the 1.1 patch?

 

[fritzdog]

Yesterday I reinstalled both Far Cry and the Catalyst 4.5 drivers. I got the following results running Bench'emAll:
Far Cry

Using Max settings.
Map: Fort Demo: BenchemallDefaultDemo
640x480
run# 0: Average FPS: 65.19
800x600
run# 0: Average FPS: 66.61
1024x768
run# 0: Average FPS: 67.66
1280x1024
run# 0: Average FPS: 64.09
1600x1200
run# 0: Average FPS: 49.62

I played Far Cry as well for about 2 hours at 1152x864 at everything maxed, Low AA, 2xAF. Gameplay was excellent and my framerates were typically above 60 spiking in the 90's. Maybe as low as high 40's when in intense combat with 5-6 mercs.
Fritz

 

[fritzdog]

I just re-ran 3DMark03 with the 4.5 drivers and got a score of 8845. I must have had a bad install of these drivers the first time around.

 

[Jeff7181]

Originally posted by: Pete
FYI, Curley, AA doesn't work correctly in Splinter Cell, because of the way it performs some effects (same with Prince of Persia and NFS:Underground, IIRC). You're right that the preferred way to enable AA/AF is to set the Control Panel to Application Preference and enable them using the game's settings, when possible.

Unless you have an nVidia card and can set up all those options individually for each application

 

[Curley]

Jeff7181,

Are you running a Nvidia Card? Nice pics. I particularly like the new drivers for Nvidia where there are preset game specific drop down menu for your games. I plan on buying the 6800gt, and a high end ATI. They will sell very fast but I will make my decision on which card I will invest in.