Personal computing discussed
Moderators: renee, JustAnEngineer
awakeningcry wrote:NOT windows ME
axeman wrote:Another classic piece of hardware would probably be the TNT2.
Prestige Worldwide wrote:Vooodoo 2 FTMFW. I think that 12mb monster is still somewhere in it's original box at my parents house....
l33t-g4m3r wrote:awakeningcry wrote:NOT windows ME
Win ME had some key improvements that made it superior to 98. The reputation is ill deserved, because problems mostly stemmed from other sources, like drivers and pebkac. The bloat is easily disabled, too. It's a smoother experience overall, with more updated software and features. 98's more for dos mode, and retro for the sake of retro.
Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
drfish wrote:Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
Never thought of that, now I'm really curious!
jihadjoe wrote:drfish wrote:Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
Never thought of that, now I'm really curious!
I don't know how FCAT would have worked, because back then SLI was literally two cards each rendering half of the screen. In a way it is superior to current SLI, because two 12MB cards would combine to have an effective 24MB ram, and you could run resolutions that would otherwise be impossible without SLI.
BIF wrote:I'm actually happy NOT going back in time anywhere before the summer 2013 system guide, although my system is already an ancient 6 months old, and not including the GPU is using tech a year older than that. I think it's almost time for an upgrade. So there.
Concupiscence wrote:As for my ultimate legacy gaming config, I'd hop into the wayback machine and snag the following parts:
* Windows 3.1 (why not, there were some nice adventure games that'd run on it)
* Slot A Athlon - clock speed almost doesn't matter, it'd shred through anything made before 1997
* 256 MB PC133 RAM
* solid motherboard (I have a fondness for the weird-as-hell FIC SD11, weird AGP port and all)
* Gravis Ultrasound
Scrotos wrote:Concupiscence wrote:As for my ultimate legacy gaming config, I'd hop into the wayback machine and snag the following parts:
* Windows 3.1 (why not, there were some nice adventure games that'd run on it)
* Slot A Athlon - clock speed almost doesn't matter, it'd shred through anything made before 1997
* 256 MB PC133 RAM
* solid motherboard (I have a fondness for the weird-as-hell FIC SD11, weird AGP port and all)
* Gravis Ultrasound
Not Windows 3.11 or WFW 3.11?
I had that same mobo and Athlon 500. Why not 512 MB of RAM? That's the max that Windows 9x could handle without hackin', right? And I think the SD11 only took PC100 but it's not like going faster would be a problem in that system. Not like really old stuff where 60ns versus 70ns RAM would totally jack your system up. What was weird about the AGP?
Last comment is that you want either a SB16 + WaveBlaster and a GUS ACE with the memory expanded to 1 MB or a SB16 + WaveBlaster and a GUS PnP Pro with 8 MB or 16 MB of RAM. Feel free to substitute the WaveBlaster with a WaveBlaster 2 or one of the compatible Yamaha cards, the DB50XG or the DB60XG. I never had any experience with the Pro Audio Spectrum 16 (PAS 16) or Turtle Beach cards. I don't think they were supported well with games, they were more for doing DA and MIDI stuff. Likewise I'd probably skip out on a Roland LAPC-1 or MT32. Tried it once, it sucked for gaming.
Concupiscence wrote:Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
It was really dependent on what you expected it to do.
The good: Framebuffer size was doubled, so you could actually run in higher resolutions, or at lower resolutions with Vsync + triple buffering enabled. Raw fillrate was also effectively doubled, so framerates would jump accordingly in most games of the time.
The bad: The implementation of SLI required that texture memory contents be identical between both cards, as they'd be used to work on scanlines with identical data. Thus, a 12 MB Voodoo2 SLI would function like an 8+8 MB card, with 8 megs of framebuffer space but only 8 MB of effective resource memory. There was also no notable improvement to geometry throughput, if I remember correctly.
The ugly: This may have been an apocryphal rumor, but the implementation of multitexturing via separate TMU chips might have effectively halved available texture memory again when multitexturing was enabled. I read that somewhere once but haven't been able to source it in years. Also, the image quality of Voodoo2s wasn't the greatest, and all the usual pre-VSA100 3dfx limitations applied - maximum 256x256 textures, basically no texture compression except for palettized textures, 16-bit color, 16-bit z-buffer, no anisotropic filtering, no trilinear filtering, mediocre blending mode support, &c.
Overall: For games released when the Voodoo2 itself was viable SLI definitely made a difference, but by any reasonable objective measurement you'd be better off with a Voodoo3. I remember my 12 MB Voodoo2 SLI setup being near its wit's end trying to handle Serious Sam: The Second Encounter with the WickedGL driver about a decade ago. It might be interesting to see if it works any better with that crazy MesaFX driver released a while back.
jihadjoe wrote:drfish wrote:Prestige Worldwide wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
Never thought of that, now I'm really curious!
I don't know how FCAT would have worked, because back then SLI was literally two cards each rendering half of the screen. In a way it is superior to current SLI, because two 12MB cards would combine to have an effective 24MB ram, and you could run resolutions that would otherwise be impossible without SLI.
Many people, broke nested quotes wrote:How well did SLI actually work back then, though?
Would be pretty interesting to see if frame pacing could be tested on FCAT to see if it's actually worth it.
Never thought of that, now I'm really curious!
I don't know how FCAT would have worked, because back then SLI was literally two cards each rendering half of the screen. In a way it is superior to current SLI, because two 12MB cards would combine to have an effective 24MB ram, and you could run resolutions that would otherwise be impossible without SLI.
Negative- they did not combine. It's actually no different than what we have now, except that the output combining happens internally, and digitally, and that today's games are so very friggen' complex that there's no 'good' way to do multi-GPU scaling; all methods have their drawbacks. The 'best' setup is with a GTX690, which would still be awesome if they'd put more RAM on it.
Forge wrote:Actually, you're wrong, jihad joe was much closer to the mark. original SLI (the only REAL Scan Line Interleave) would do, say, 800*600 by having each card render 800*300, and have card one do the odd lines and card 2 do the even lines. In this way, you did have to have the same texture memory (8MB of the 12MB total), but you could use a smaller framebuffer, or more of them, since you only needed to have an 800*300 buffer instead of an 800*600 one. Back in those days, with tiny little memory, things like the size of a framebuffer for a given resolution could actually limit you. It's hilarious now, but it was Srs Biz back then.