Bit broke after the holidays, so focussing on working with what I've got rather than hoarding new stuff. Been cleaning up the spare room workshop a bit and finishing my backlog of soldering.
That said, last night I just gave up on a board while trying to recap. It was an Intel D865PERL - a quite nice black, last-gen AGP So478 board that did actually boot, but wasn't stable. No surprise: eight bulging and/or leaking caps. What was surprising: all the dud caps were labled "Nichicon". Normally Nichicon is one of the good brands, but obviously they had bad days too. Given that it's an Intel board fakes are unlikely. Regardless, bad caps are bad caps and need to go. But Intel seems to have chosen some diabolical ROHS solder with an astronomical melting point, combined with tiny-legged caps whose legs disintegrate at temperatures that actually melt the solder. Normally the worst that happens is that some of the legs detach completely from a disintegrating cap and you can desolder them separately afterwards (just pull lightly with pliers), but in this case they broke all over the place, sometimes both above and below the PCB on a single leg, forming a brittle little plug in the board that stopped the solder from flowing. If the legs - and holes - had been wider I would have considered drilling it out, but these things were so tiny I didn't have a fine enough bit. So given all that - and the fact that So478 is a bit later than my period of interest anyway - I've given up. This board goes onto a pile for a student who wants to decorate his walls with dead motherboards 😉
After that it was getting late, but I still had time to finally start on my UMC-build, with UMC CPU, UMC chipset, UMC I/O, UMC VGA and UMC NIC. Didn't get much further than installing the board in my AT desktop case and checking that it worked, but at least that went smoothly. Tonight I choose a HDD and get DOS running.