Slashdot Mirror
120+ GeForce FX Reviews Collected
Peter writes "We just finished at 8Dimensional our list of GeForce FX reviews. It tries to show all reviews of these video cards currently online, 120+ are listed at the moment." Hmmm, time to upgrade from an Xpert@Play98 ...
ATI did not cheat (although they have cheated in the past), they reordered the instructions in a shader but it remained mathematically and functionally identical. This is what optimizing compilers do all the time. It's called optimization, not cheating, and it is legitimate, they looked at the shader INSTRUCTIONS to see if they were suitable for this optimization. It made about a 2% difference, and IMHO ATI should leave this optimization in and broaden the scope if possible.
If it was a really narrow path optimization then it's borderline, but far different from the wholesale cheating of NVIDIA who: rewrote shaders with completely different results, didn't clear the screen at certain points and added hidden clip planes to eliminate pixel fill. All very underhanded and why you would make excuses for this I don't know. Sure it's a synthetic benchmark, but if it don't matter then don't cheat.
but their drivers have a history of being crap. I have never encountered a single game that wouldn't run on my Geforce 4 4600. My girlfriend owns a 128 mb ATI card and I can think of over 10 games that wouldn't play, strictly due to the video card.
Benchmarks are meaningless if you can't play games with the thing
If it wasn't cheating, and they were only looking at shader instructions, then why did the results change when they renamed the file from Quake3.exe to quack3.exe? It is also important to note that the Quake3.exe results were of less quality than the quack3.exe results.
Um... he was talking about 3dMark, not the Quack issue (btw, there are no shaders in Quake3), which although people insist on constantly bringing, up was most likely not a cheat; ATi not only fixed the 5 problematic textures, but also increased the performance past the level of performance attained using the "cheat," in the next driver release.
Uttering logically derived and empirically supported truths to the disciples of the orthodox establishment.
Try Ut2003, new Wolf demo, Splinter Cell.. Well, any relatively new FPS game, in any resolution over 1024x768, with any kind of image-improving feature (AA, AF).
Do you honestly believe that a million people bought R9700 in the first few months after it came out just to get "extra 20fps", or to win a "pissing contest" (another slashdot favourite)?
Never underestimate the bandwidth of a 747 filled with CD-ROMs.
Actually, I had a GF3ti500, and U2K3 and BF1942 really started to drag on the FPS at 1024x768+ Res. And then if I turned on AA/AF it dropped to sub 20's. Time to upgrade.
Picked up a ATI 9700Pro, and it has been about 9 months of great use for about another year before I need to upgrade.
BTW, to see if your system is worth an upgrade, check out future marks online browser, you can search for a CPU near yours and then check to see if a faster video card will improve your FPS. Good reference.
If memory serves, the difference is the Quadros and FireGL (etc.) cards are better for CAD/CAM because they're more exact, whereas your Radeon and GeForce series chips are engineered to be fast. You don't care too much if a wall in QuakeIII is misrendered for a second, but it'd be a big problem if you were rendering something like an office building or such. There was an article on Slashdot semi-recently that covered the differences, in fact. Gist is, you might be able to render things fine on a GeForceFX , but if you're doing it for your livelihood, then it's a Nice Thing to know that the rendered result is accurate, and that's what the workstation cards try to provide.
mrg
www.futuremark.com
I was using 3dMark 2001, but 2003 is out. Download the demo (if they still offer it) and sign up with the profile manager.
Then you can do 2 neat things, Compare by CPU and Compare by GFX Card. So if can see what a faster CPU does with the same GFX card, or the same cpu with Faster GFX cards. Really ace.
Actually... This is easily explainable.
:-\ - Still fun reading material though :)), I'd hardly call using it cheating. The performance DOES relate to real-world situations, as the majority of people playing Q3 Engine games don't change the name of their executable :)
John Carmack released a document YEARS ago explaining to hardware vendors how to optimize their OpenGL implementation for the Quake 3 engine. Such things as Vertex Array Client States (i.e. GL_VERTEX_ARRAY) never changing (always enabled), specifics for multi-texturing, vertex structure size, etc...
Given that anyone who cares to search Google for a minute or two can pull up this document (The original doc is gone
What would be nice, however, is if developers had some access to this driver layer. I'm not saying anything as major as Q3's optimizations, but if a developer could tell the drivers that it won't be making any state changes to a particular attrib stack subset (i.e. GL_ALL_ATTRIB_BITS wouldn't actually push the entire stack), it'd increase render-time without any fancy 1st party vendor magic.
Also, using languages such as Cg, shader optimization is possible at run-time. I've never used it myself, I'm biased to hardware specific ASM at the moment.