This page is a mirror of Tepples' nesdev forum mirror (URL TBD).
Last updated on Oct-18-2019 Download

MMC5 CHR banking

MMC5 CHR banking
by on (#192092)
According to the Wiki:

Quote:
Registers $5120-$5127 apply to sprite graphics and $5128-$512B for background graphics, but ONLY when 8x16 sprites are enabled.

Otherwise, the last set of registers written to (either $5120-$5127 or $5128-$512B) will be used for all graphics.


Has this been confirmed via testing on original hardware? It seems very unlikely to me. How is the MMC5 supposed to know whether sprites are 8x8 or 8x16?
Re: MMC5 CHR banking
by on (#192111)
AWJ wrote:
How is the MMC5 supposed to know whether sprites are 8x8 or 8x16?

It can spy on writes to $2000.
Re: MMC5 CHR banking
by on (#192413)
Can. But does it? It could also log sprite reads and…yeah, that would take a lot more silicon.
Re: MMC5 CHR banking
by on (#192414)
The big question I have from that is why would that be desirable ?

It feels like it "has" to be a side effect of how they implemented this extended sprite banking, but I don't have any ideas as to what original constraint would make this simpler than just always having sprites come from a disjoint banking space than the background, regardless of whether the sprites were 8x8 or 8x16.
Re: MMC5 CHR banking
by on (#192419)
Playing Devil's Advocate…
  • Having concurrent banking helps when you're doing raster effect banking that includes sprites.
  • Having 8x16 sprites means you're using significantly more data on sprites.
  • 8x8 banking already permits BG-disjoint banking of sprites, by PPUCTRL.d3-4 = {10,01} and having different banking registers for $0xxx and $1xxx.
  • Having a mode where it acts like a traditional mapper possibly reduces programmer learning time.
Yes, weak reasons.