Conversation

i miss the days when consoles would name shit like Emotion Engine or Reality Display Processor or Blast Processing. they should bring that back

i mean i know that consoles are just a zen apu soc glued to a heatsink but at least call that the Quantum Synthesis Engine or smth

2
3
1

@eclairwolf games should be fun again, i think. playing them but also the whole experience around them

0
1
3

@eclairwolf wasn’t “blast processing” just a marketing buzzword that didn’t translate to anything

1
0
0

@eggbreakerdotexe it pretty much was just marketing, but it did actually mean something - the genesis was theoretically able to "race the beam" and draw directly to each pixel on the crt by directly writing to the palette on the fly

but nobody ever did this because it involves precisely counting instruction cycles to maintain sync with the crt beam lmao

1
0
0

@eclairwolf @eggbreakerdotexe so you're telling me the Amiga had blast processing in the copper chip?

2
0
0

@fluffy @eggbreakerdotexe not too familiar with amiga hw but from what i read about it i think so!

0
0
0

@fluffy @eclairwolf @eggbreakerdotexe YES! It did! I directly referenced this in my multiple articles on the subject. Games like Gloom use it. It's done for an entirely different purpose on the Amiga, but it's the same trick. On the amiga, the reason you'd use "Blast processing" (or direct color DMA as it's more technically known) is to get around the planar pixel format. Too expensive to plot arbitrary colors due to jumping around memory. You need to plot arbitrary pixels for a raycaster engine

1
0
0

@fluffy @eclairwolf @eggbreakerdotexe So, games like gloom instead render to a single bitplane, 1bpp, and use COPPER to change the plane's color per dma access slot. It's literally the same trick as blast processing, except owing to COPPER doing the DMA without blocking, it's far more useful. Source: I'm the author of the blast processing demo and wrote all the articles on it.

1
0
0

@fluffy @eclairwolf @eggbreakerdotexe Also, the reason nobody used it wasn't necessarily because it required precise instruction counting, it's because there was no known horizontal sync method. We didn't discover a method for stable line sync until oerg stumbled upon it. The long gist is it was possible to get within a 2 pixel error range using known sync methods. dram refresh cycles will halt DMA, and cost 2 pixels. Just nudge DMA to begin at a refresh slot and eat the error range for sync.

2
0
0

@GabeMoralesVR @eclairwolf @eggbreakerdotexe This is a lot of very fascinating technical detail written in response to what was essentially a shitpost, but thank you!

1
0
0