Overclocking on older systems (e.g. 286) could involve changing the frequency used across various components (including ISA BUS), where they all derive their frequency from the one crystal.

Overclocking ISA bus affects all cards, and yes - disk controllers and NICs can get faster this way.

Also, does using an MFM HDD with RLL controller count as overclocking?

I'm going from distant memory here, but AFAICR, on the sound card front :

- it was apparently possible to overclock the DSP on at least some MWAVE variant based sound cards
- Some SB Pro compatible cards apparently allow one to set an effective sampling rate of more than 22KHz in stereo mode (specific software is required). YMF724 based cards may be capable of this
- Either the GUS Max or Interwave based cards (or both) could be set to more than 48KHz (64KHz possibly). I don't recall whether the low pass filter actually negated the potential usefulness of that

Anyone who remembers more/better, please correct me or corroborate, as needed.

I accidentally overclocked a YM2203 to 7MHz one time. The sound played an octave too high. I over-volted a fan once or twice. And to really stretch the concept, I 'overclocked' a furnace by replacing the drive pulley on the blower with one of smaller diameter so it would run at a higher RPM...

PCI IDE cache controller by Tekram DC-690CD

Another example of unusual overclocking would be a Hercules clone graphics card.
BUT!
The goal wasn't to make it faster.
The goal was to make it compatible with SVGA monitors.

Normally, such cards use 16 MHz pixel clock, which produces 18 kHz HSYNC and 50 Hz VSYNC.
SVGA monitors need 30+ kHz HSYNC.
The solution is obvious: replace the 16 MHz crystal with something faster.
Sometimes it works, sometimes doesn't - I guess it depends of the memory speed.
Also, it's necessary to add a very simple DAC between the card and monitor, to convert 2-bit TTL to analog 0.7 V.

My Sega Mega Drive is overclocked. Getting a sweet 10 MHz outta that baby!
Way back in the day, I overclocked my Palm Pilot - think it was a III or a IIIx. Got a sweet 25 MHz outta that baby!

Just a funny idea, please don't take me seriously: how about under-clocking the RTC and/or the PIT to get a boost on all the benchmarks that depend on them to measure time?

EduBat wrote on 2024-06-04, 20:58:

Just a funny idea, please don't take me seriously: how about under-clocking the RTC and/or the PIT to get a boost on all the benchmarks that depend on them to measure time?

I think sombody actually did that... not sure... some disk controller with BIOS slowing down the clock while disk operations?
Anyway, there was plenty of tricks to cheat common benchmarks.