« first day (589 days earlier)      last day (860 days later) » 

quartata
quartata
00:54
@KZhang What are the possibilities in the future of some kind of paging and/or address extension so we can have more than 65536 bytes of RAM
I know it's pretty hard and we don't need it for now but just curious how much it would take
K Zhang
K Zhang
I'd imagine that we would need to do some sort of paging technique, but I don't really know how that would work
It doesn't seem as hard as some of the other things I have done, though
quartata
quartata
trying to see if there's any architectures whose MMUs you could take inspiration from
K Zhang
K Zhang
01:09
I think I have part of an idea: make a multiplexer take an input from some sort of memory (modified synchronizer?), and have that memory be set by a new operator
 
2 hours later…
quartata
quartata
02:59
@KZhang You mean like load words making up the full address into a special place in memory?
That sounds nice but it would have to be in addition to the current addressing
If we take up 4 instructions for busses that would only leave room for one instruction
FETCH offset, length, destination?
like
if we want to move address 65537 to address 2 assuming big endian order
PhiNotPi
PhiNotPi
Would it be worth it just making the computer have a larger bit width?
quartata
quartata 
. MLZ -1 2 3;
. MLZ -1 65535 4;
. FETCH 3 2 2;
loads words 3 and 4 as the address, moves its contents to 2
@PhiNotPi you say that so casually
PhiNotPi
PhiNotPi
Like make it an 11-bit computer or something. Or maybe 13-bit.
quartata
quartata
I mean I'm mostly future proofing here. There's no way we can fit any significant libraries in 16 bit ROM or RAM so if the computer itself stays 16 bit we need tools to help out
btw does my fetch example make sense
the answer is no because I did the numbers wrong
my example loads address 196607, pretend that's what I meant
 
2 hours later…
Anonymous
05:19
What if we had multiple 16-bit RAM blocks, and a page selector?
Anonymous
So we could do e.g. PAGE 1 to swap to the second page, and PAGEMOVE 0 1 5 to move the value in page 0 address 5 to page 1 address 5.
Anonymous
(or PAGECOPY or something like that)
Anonymous
Or FETCH page_num remote_addr local_addr to fetch the value in page page_num at address remote_addr and put it in local_addr (on the current page)
quartata
quartata
14:30
@Mego Yeah, although if we go the paging route we're also kind of obligated to add in protected mode
 
2 hours later…
Anonymous
16:10
@quartata Why? We don't have to follow Intel's lead. We certainly don't when it comes to the rest of our architecture.

« first day (589 days earlier)      last day (860 days later) »