|
|
 |
|
|
Pages: 1
3D Mark 2001 Benchmark between Radeon and Geforce3.
(Click here to view the original thread with full colors/images)
Posted by: Chako
I know the subject title is rather stupid, but I must say, I just did 3D Mark 2001 benchmarks on my system. The only thing that changed was the video card.
Radeon 64DDR VIVO: 3040 3D Marks
Pine Excalibur Geforce3: 5837 3D Marks
Nice boost for only the video card..and the Radeon was no slouch either.
Posted by: tkron
Which Pine Technology GeForce3 did you have the TI 200 (EXCALIBER GEFORCE3 TI 200 AGP 64MB DDR NVIDIA TV/DVI OUT {PVT20KMA}) or TI 500 (PINE GEFORCE3 AGP 64MB DDR NVIDIA GEFORCE3 W/TV OUT {PVT20ART})
Did you do the standard default test.
My Geforce 2 MX 400 64meg card scores:
a score of 2365 at default 1024x738 32bit compressed, D3D hardware T&L and antialiasing none.
a score of 3475 at 800x600 16bit compressed, D3D hardware T&L and antialiasing none.
These were run on an Intel 933Mhz, 133Mhz bus, 512meg 133 Mhz SDram, Asus CUSL2 running windows ME
t'k
Posted by: SKYHN
Nice scores Chako. What resolution and settings were you running?
Posted by: Chako
I have the Excalibur GeForce3 AGP 64MB DDR w/ TV-out option (PV-T20A).
The test result was at: 1024X768, 32 bit colour, D3D Pure Hardware T&L, no antialiasing.
1024X768, 32 bit compressed texture, D3D hardware T&L, no antialiasing: 5748 3D marks
800x600, 16 bit compressed texture, D3D hardware T&L, no antialiasing: 6229 3D Marks.
Got this on a Thunderbird 1.2 with 516 Megs of PC 133 on an Asus A7V133.
Posted by: Null Actor
You get a big jump in scores because the GeForce 3 is able to complete some tests that the other card cannot.
Posted by: SKYHN
Id love to come close to that score. I ran a 3d mark earlier today after defraging and fooling with some stuff and I got 1379.
Is D3D hardware T&L option only available on some video cards? I noticed I can only use D3D Software T&L.
Posted by: tkron
Absoulutely correct Nova-Z.
Chako, if you want to compare apples to apples you have to turn off the following tests when running the Geforce3 if you wish to compare your scores with mine. You may even have to unselect the same tests to compare your Geforce3 to your radeon card. click on the change button in the selected test box on the left side
Unselect the image quality box under the Pro version tests.
and then hit the custom tab and uncheck the tests.
these test were not run with my card:
7 game 4 nature
12 enviornment bump mapping
15 Pixel shader
All four tests at the bottom under Pro version tests should be unchecked.
I also remember awhile back that Wargy said that the test results changed when he added more ram, but I do not recall if it was 3D Mark 2000, or 3D mark 2001 benchmark. In any event adding more CPU ram should not really effect your score, and the fact that is did confused both me and wargy.
Posted by: Foogar
Wowsers, I thought my score of 3763 was low for the default test...
PIII 550
256MB RAM
100MHz bus
GeForce 3 ti200
Posted by: SKYHN
3763 is a good score. I would be that if your CPU was faster that score would be a lot higher. GF3 Ti 200 is a good card, probably the one ill end up buying.
Im gonna copy 3DMark to a cd and run it on my P3 machine, it has a little better video card than my TNT2 but not too much better. Its only got 128mb of ram though.
Posted by: SKYHN
Well....
Looks like my P3 is kicking some butt. My 3dmark on it was 2356, about 1000 higher. Thats expected though. My p3 has a GeForce 256 32mb AGP in it.
AGHHHH! I NEED A GEFORCE 3!
Posted by: Chako
Ok Tkron, I followed your instructions and here are the results.
1024X768, 32 bit compressed texture, D3D hardware T&L, no antialiasing: 5183 3D marks
800x600, 16 bit compressed texture, D3D hardware T&L, no antialiasing: 5380 3D Marks.
Good point actually, one I had never thought about.
Posted by: SKYHN
Doesnt compressed textures make a video card slower?
Posted by: tkron
Actually I believe texture compression utilizes cpu memory and also lowers the graphics quality of what you are seeing rendered on the monitor.
To a big extent what texture compression does is not that important in what I was trying to acomplish.
In order for tests results to be meaningful, so that you campare one cards test to another, you have to be sure that both tests are being run with the same set up, especially if changing the options selected when setting up the test will impact the final test score number.
Posted by: MADhix
tkron, I totally agree. If both tests are not run with the SAME EXACT setup, then the scores are useless. These people on the internet are getting 7000 and crap, and some people are like WOAH! But, I can do that too with my PNY Verto Geforce3 Ti 500 by setting the res to 640x480, crompressed textures, 16-bit colors, etc.
Posted by: MADhix
Hey, does anyone know why when I look at my system info in 3dMark2001 it shows AMD Athlon(tm) @ 565MHz when if I look at the system information in system tools in Windows XP it says @ 856MHz (I have a 850MHz). I think MadOnion is crap.
Posted by: Null Actor
 Quote:
Originally posted by tkron
Actually I believe texture compression utilizes cpu memory and also lowers the graphics quality of what you are seeing rendered on the monitor.
|
Texture compression doesn't use CPU time or extra memory.
Yes, it can lower image quality, if you compare the same size texture very compressed as uncompressed. However, when most games use texture compression, what they do is use a higher resolution texture map in the place of a smaller one, AND they save memory doing it. Or they use a lower compression to save texture memory, and you don't see a difference.
Consider a 512x512x32bit texture. This is a decent texture by todays standards, but not really high res anymore.
This texture takes up 1048576 bytes. About a meg.
Now consider a 1024x1024x32bit texture. This is a high res texture, and looks by far better than a 512x512.
This texture takes up 4194304 bytes. About 4 megs.
Now, with texture compression you can between 1:3 compression (almost no visible quality loss) to 1:6 compression (minor visibible quality loss).
So now, either you can compress that 1meg texture down to about 333k, with no quality loss (save a bunch of texture memory), or you can replace that low res one with a high res one, compressed decently, and use about the same amount of memory as a texture 4 times smaller.
Pretty sweet huh?
Not only that, but compressed textures render just as fast as normal textures, and in some cases, faster.
It's a nice win win situation.
Posted by: SKYHN
Every program can read it different. SiSoft reads my PC as AGP fast writes are off, but in Bios they are on. Some programs read my hard drive as different sizes. SiSoft says mine is 60.9gb, windows says 55.6, and another program read it as 62.4, I may never get a right reading on it, accept maybe in good ol DOS.
Posted by: Chako
I checked everything. How was I supposed to know that the Geforce 3 card would use some test that the Radeon would not. I didn't notice it on my original posting. However, I did ensure that all else was consistent and the same. So there. 
Posted by: Null Actor
Personally, I think the test is totally valid with all the tests being used, regardless of card.
If a card can't do something required for the test, then it's an inferior card, and by that logic, should receive a lower score.
Posted by: SKYHN
Here, check these out:
Click here
Click here
|
|
|
|
|
