Rockbox mail archive

• Next message: jeansebastien.tauth_at_mpsa.com: ".mod"
• Previous message: Psyco Dedman: "Re: false info are displayed by "low-quality" mp3"
• Maybe in reply to: Magnus Holmgren: "Optimized bitswap"
• Next in thread: Nielsen Linus (ext): "RE: Optimized bitswap"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Den 19 Aug 2002 skrev Linus Nielsen Feltzing:

> > I've also looked a little on an assembler version, and I believe it
> > would be possible to write more efficient code (like half the size
> > or so, compared with the output from GCC 3.1.1), but I'm not
> > familiar enough with embedded assembler in GCC (or SH1 assembler,
> > for that matter) to complete it. Some stuff must be in r0 AFAICS,
> > which complicates things.
>
> I have been thinking, and I will give this a try in the next couple
> of days.

I've done some work with bitswap (and fliptable) in a separate .S file (easier that way, IMO), and I have something that builds, but crashes at some unkown location (not in bitswap). Don't know where yet, but I guess somewhere in mpeg.c. It's difficult to pinpoint with next to no debug facilities...

From what I've gathered, I get data in r4 and length in r5. Do any registers need to be saved (doesn't work if I save or not though)?

  Magnus

• Next message: jeansebastien.tauth_at_mpsa.com: ".mod"
• Previous message: Psyco Dedman: "Re: false info are displayed by "low-quality" mp3"
• Maybe in reply to: Magnus Holmgren: "Optimized bitswap"
• Next in thread: Nielsen Linus (ext): "RE: Optimized bitswap"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]