Is an 80x25 text display possible on the Prop using 2,000 bytes of hub RAM?
Dave Hein
Posts: 6,347
Has anybody done an 80x25 text display on the Prop that uses a 2,000-byte array to store the text? This would require either using the ROM font, or the font would have to reside in cog memory. 2-colors is fine for my application. It would be great if it could fit in one cog, but 2 cogs would be OK also. Ideally, it should be on a VGA display, but TV might be OK if it's just black and white.
I'm just wondering if this is possible to do on the Prop.
I'm just wondering if this is possible to do on the Prop.
Comments
Do you care how many characters are included? One option we've got is stuffing characters into the TV COG. It's reasonable to do 96-128 of them that way, keeping the HUB memory clean, but for the text screen buffer.
I've got a TV one that works that way somewhere. It's 64 cols @ 80Mhz, 80Cols at 96Mhz. Cluso has done a text in cog driver too.
Do you need redefinable characters?
Basically, put the video driver and the character definitions into one 2K memory segment and start it. Then use where it was in the HUB for the video display, avoiding the use of the ROM characters, which are very tall vertically.
It reuses the hub code space for the 2K buffer. The font is in the cog, and it has a limited VT100 command set.
The TV is actually driven by the VGA counters and is monochrome. Its not all my work (see the code for attributions).
Been thinking on the drive home, and if a VGA is preferred, it might take two COGs to do this using the ROM font. There is the WHOP method pioneered by Kuroenko, and his drivers, which I've not really explored much. The VGA sweeps are 2X what the TV sweeps are, and the ROM font is 16 pixels too. That's basically a 1280 pixel display!! Chip did it with 5 COGS, but it would do more than text too.
So, TV is easy, but it's TV. I assume it's monochrome, due to the 2K text screen buffer. On VGA that would mean 2 colors. On TV, it means monochrome as there are few color combinations that really look good without a lot of fussing, and a good TV. S-video will display very nicely --even with more color combinations. Is that an option? S-video monochrome will look great on many TV devices, and it's more or less pixel perfect on most capture cards.
May I ask what the display need is? For example, is it for debug? Is it to be mounted somewhere? Can a capture card be used, etc...?
I'm not sure about character cell sizes. 8x8 is super easy, just a byte, and the char defs fit into the COG nicely. 8x12 could round up to 16 bits, also fits into the COG, but not so many characters are possible... Would need 1.5K or so for defs @ 96 chars. Probably would need two COGS, loaded sequentially to hit the 2K HUB RAM goal. The second "text" cog won't take much code, so I think that's possible. Some new code has to be authored for that. We have existing pixel routines for 8x8, the ROM font. I don't think I've seen a 12 implemented yet. Anyone know?
(goes off to look at Cluso's driver)