PERQ Emulator Progress (of sorts)!

Tony Duell ard at p850ug1.demon.co.uk
Wed Nov 7 17:58:14 CST 2007


[...]

> 951).  Of course, I obviously still have a ways to go.  As evidenced by 
> the completely garbled text, I have yet to get the PERQ's unique (and 
> extremely complicated) RasterOp hardware emulated to any significant 

Good luck, you'll need it :-). The rasterop hardware is conceptually very 
simple, and practically very complicated. The timing of just about 
everything is critical, as you must know by now, the machine has a 
4-state counter (basically counting through the 4 16-bit words that make 
up a 'quadword', the memory hardware and video shift registers being 64 
bits wide, whereas the path to the CPU is 16 bits wide). It's possible, 
although not really documented, to do a read-modify-write cycle on a 
quadword _and the data transfers overlap_ -- IIRC there's a 2 clock delay 
between them. And that's exactly what the rasterop hardware does.

The Rasterop microcode is intrictate to say the least....

I asusme you've read the schematics and the techref. Have you seen the 
Rasterop microcode? And the filksong?

> degree.  And after that I have a ton of hardware details to work out.  
> But this is a major milestone, and I'm so excited that I just had to 
> share my excitement with someone... I never thought I'd get anywhere 
> near this far with the emulation, given how complicated the PERQ is.
> 
> And if I've seen so far, it's because I've been standing on the 
> shoulders of giants.  Or one giant, at least -- our own Tony Duell who's 
> been ever-so-patient with my questioning and who probably knows more 
> about the PERQ hardware than the original designers did...

That much I really doubt... Mind you, even if I say so myself, I think I 
understnad the hardware fairly well. And I have written microcode...

-tony



More information about the cctech mailing list