The funniest thing is, I have had zero (0) problems with my Radeon yet. Several weeks of smashing on it in 98/2000/linux, and I have yet to see one error of any type. I installed the AiW and the drivers available on ATI's site (out of box, the Win2k drivers worked with suck-ass performance on Win2K, but great stability). I left my AiW capturing video while running a defrag on the raid, while running a DVD player looped from the single drive, while running 3Dmark 2000 demo looping/nosound. It ran for four days. (Only 3 hours of video, though). I then wiped out the test video (after watching it for errors) and repartitioned for linux. I installed linux, then did a DRI Radeon branch pull, and got usable 3D. I then half considered ordering XiG's Accelerated-X server for Radeon just so I could say I have the fastest linux 3D around, but passed. This is about ATI's drivers, not XiG's. I installed 98SE and ATI's Radeon stuffs (off the disc this time). I then produced a repeat of the earlier stability test, and it finished 3 days of torture before I *had* to use my PC again. ATI's drivers aren't #1 yet, but they are far from suck ass. I had to explain to a friend: He installed Win98SE fresh (a must, Nv drivers tend to destroy a registry, and loose files can linger forever.), he then installed the drivers on the CD and proceeded to rock the socks off of any 3d app he fired up. He then fired up the VIVO stuff and it rocked his socks as well. I then installed WinDVD and the 2.3 patch for him, and he watched the 'smoothest, cleanest DVD video' he's ever seen (his words). He then went and got the latest Special Purpose (very beta, for specific problems) drivers he could find, and completely destroyed the whole system. I got him to take his ATI card back and get a GeForce2. He's happier with it, even though the last 9 beta releases have been 100% incompatible with his system (he's running 5.32, it's the latest that system will boot with.) I rest my case.

Actually, I think that nVidia and 3dfx are about tied as far as driver quality goes.

nVidia has more "leaked" drivers and sometime that can equate to a good thing, depending on the user AND their hardware. I have found myself switching drivers to get certian games to run better though, and that can get annoying after a while.

3dfx hardly ever has leaks, but thier WHQL are of good quality (contrary to popular belief). Even the beta drivers that they post on 3dfx gamers are pretty good. My V5 will run anything i through at it though, and thats a good thing(TM).

Matrox, I haven't owned one of thier products in a few years so I can't comment on thier stuff of late.

ATI, my Radeon is a good card and i'm kinda impressed with it compared to the old Rage128, but the hardware is no good without good drivers. And thats about the bottom line on that.

Right on #8. The ATi silicon is decent and getting better but the drivers are still second rate. I think NVIDIA is to blame for people thinking more drivers leaks equal improvement. Yeah, there are a few leaked gems there but one or less in each driver set. But I think it's good for NVIDIA owners because it shows at the very least that NVIDIA hasn't given up on a particular card, especially when they're all backwards compatible. Let's see ATi give a crap about their old stuff...pfft

Here's my vote for overall driver set quality

1.NVIDIA (not perfect but you can see results a good % of the time)
2.3dfx (room for improvement)
3.Matrox (major room for improvement)
4.ATI (shame, shame)

ATI still can't get decent drivers out

#6, isn't that what ATI is for, now?

Originally Posted by wmtj
I hope this thing really flies. Competition is a good thing!

Wonder if this is their first one chip solution (GPU) or will the whole card equal the size of delaware? Proper/Complete AGP support this time? better Linux drivers? Ahhh.....

I still have my doubts about 3dfx but we shall see in time.

One quick question...... M-Buffer and T-Buffer... What will they do when, if they're still around 24 chip generations from now, they start running out of letters to name the new Buffers after? Will we have AA-Buffer? DD-Buffer ("hey man, my Buffer is a double-D")? XYZZY-Buffer?

If this isn't a mistake, could someone explain how the one texture lookup per pixel could work?

For effective supersampling, how can you *not* need more data than for a single sample?

With one texel per pixel, how can multiple subsamples remove aliasing artifacts that are due to high frequency (sub-pixel size) patterns in the textures?
How can edges be anti-aliased with only one texture lookup per pixel, when by definition the subsamples have to get texel data from different triangles and different textures?

Originally Posted by ronald
My, my. This from a \"mild-mannered\" technician. He\'s feeling it, baby.

"this Spectre might just make 3dfx a player again.

http://finance.yahoo.com/q?s=TDFX&d=t

Hmm, $2.69 a share. Nope, no player here. Just slightly aged and now rotting beef that will likely be gobbled up in a hostile take over.

Well that or they'll just be left to be devoured by their own maggots.