Joe DeFuria
Legend
As many know, I've been pretty critical of the past 3D Mark releases.
Of course, my final judgement of 3DMark '03 is reserved until I can actually use it for a while, and see the "3D Community's" results, comparisons, etc.
However, after reading the "white paper:"
http://www.beyond3d.com/forum/viewtopic.php?t=4272
I have to say that I am very impressed at the scope, goals, and selected tests and methodology of this benchmark.
A few initial comments:
The combination of "frame based" rendering and image qaulity tests is awesome. No more trying to "snapshot" images at the same time.
The filtering tests is also greatly needed.
Though it looks like Basic/Colourless's AA programs are safe for the moment.
Making the benchmarks more GPU dependent is "a good thing."
Having PS 1.4 support with fall-back to multi-pass PS 1.1 is also a good thing.
The feature tests (like PS 2.0, Ragtroll, etc.) should provide useful information. (I hope it's possible to select PS 2.0 color depth support for these tests...fp16 or fp32 for the FX) Multitexture test (64 textures per object) results should also be telling / interesting.
3D Sound test! Yay! 'About time somebody did something like this!
Again, congrats to the 3D Mark team! Now, my downoad should start in about 5 minutes, and then who knows how long the actual download will be....and we'll see how this thing chugs on a PIII 700 / 64 MB Radeon 8500. 8)
Speaking of 64 MB....
This is one area I found surprising in the negative direction: that both DX8 game tests require 128 MB boards to run without texture swapping during the tests. I would have thought it would be more "realistic" for ONE test to require a 128 MB board, and the other 64 MB....There's lots of 64 MB GeForce 3's, 4s, and Radeon 8500/9000 boards with 64 MB, and I don't think the majority of DX8 games would require 128 MB to avoid texture swapping.
But hey...that's a nit-pick.
Of course, my final judgement of 3DMark '03 is reserved until I can actually use it for a while, and see the "3D Community's" results, comparisons, etc.
However, after reading the "white paper:"
http://www.beyond3d.com/forum/viewtopic.php?t=4272
I have to say that I am very impressed at the scope, goals, and selected tests and methodology of this benchmark.
A few initial comments:
The combination of "frame based" rendering and image qaulity tests is awesome. No more trying to "snapshot" images at the same time.
The filtering tests is also greatly needed. Though it looks like Basic/Colourless's AA programs are safe for the moment.
Making the benchmarks more GPU dependent is "a good thing."
Having PS 1.4 support with fall-back to multi-pass PS 1.1 is also a good thing.
The feature tests (like PS 2.0, Ragtroll, etc.) should provide useful information. (I hope it's possible to select PS 2.0 color depth support for these tests...fp16 or fp32 for the FX) Multitexture test (64 textures per object) results should also be telling / interesting.
3D Sound test! Yay!
'About time somebody did something like this!Again, congrats to the 3D Mark team! Now, my downoad should start in about 5 minutes, and then who knows how long the actual download will be....and we'll see how this thing chugs on a PIII 700 / 64 MB Radeon 8500. 8)
Speaking of 64 MB....
This is one area I found surprising in the negative direction: that both DX8 game tests require 128 MB boards to run without texture swapping during the tests. I would have thought it would be more "realistic" for ONE test to require a 128 MB board, and the other 64 MB....There's lots of 64 MB GeForce 3's, 4s, and Radeon 8500/9000 boards with 64 MB, and I don't think the majority of DX8 games would require 128 MB to avoid texture swapping.
But hey...that's a nit-pick.
