My previous rig, was a full-tower case with an Abit BX6-r2 (sucessor to the BX6 and BH6, the diff between the BX and BH was number of RAM and ISA slots, primarily). Had a PII-300 SL2W8 @ 450 (the one guaranteed to OC to 450), which blew away the "other" OC'ing champ of the day, the 300A, especially for things like emulators. (At some later point in time I had a 300A in there, multi-tasking just wasn't as smooth. Cache helps more than you think.) I had both an Aureal Vortex2 and an SB AWE64 value ISA (replaced my SB16 SCSI-2 ASP card), chained together via external line-out/line-in. I used the ISA card for emulators and old DOS demos, and the Vortex2 for Windows-based stuff. I used the internal CD-ROM inputs to wire up three optical drive's analog audio outputs. I had a RagePro AGP 2X card in there too. Only 4MB VRAM, but the neat thing was, I ran with 1:1 FSB:AGP, and actually ran a 100Mhz AGP bus, with SBA enabled! I overclocked the video card's memory clock to 100Mhz too, so everything was running more-or-less sync. It was "well-tuned", even though it was still slower than a Voodoo2 for 3D. DVD playback and TV-out for emulators rocked though. I also had a WinTV PCI to take S-Vid/Composite/Tuner in. I had a Promise Ultra66 PCI IDE controller, and for a time, had both an IBM 13.5GB 14GXP and a 30GB 75GXP, both of which supported TCQ, and I think, the Promise drivers supported that too, because they were hella fast. Combined with my PCI UltraSCSI card and some SCSI opticals (Toshiba reader, Yamaha 6416S burner, and some others), I could burn + rip at the same time to the same HD. Amazing! (I can't even do that now on my current AMD XP/KT400 rig.)
Sadly, that all went horribly wrong, when I was working on a slotket-based Tualatin mod, and couldn't get it to work, so I started randomly changing jumpers around on the slotket (never a good idea to randomly change hardware jumpers...), and combined with the wire-mods that I did, ended up shorting CPU vcc to gnd somehow, and I really toasted the CPU VRM section of the board. I have never seen a CPU heatsink get so warm, so quickly. Surprisingly, in a different board, that Tualatin CPU survived and worked fine... those things were tough buggers. I still have the board, in the hope that someday, I will be able to repair it, and return it to its rightful place in history as the i440BX greatness that it is, along with three 256MB PC133 Micron DIMMs that I've saved for the occasion.