About Ememcopy

new topic     » topic index » view thread      » older message » newer message

Micheal Bolin created ememcopy after i asked for it, i asked him to
make it so you give him a list (not se_mem_copy but e_mem_copy). Now i
will explain why Micheal Bolin's e_mem_copy is so brilliant. And why it
is in most cases (and always in the important cases) the fastest.

    First of all.. you know it can copy a region instead. You needed
have to copy the whole width or cut every line.

    Then the ability of the command lists. Command list are stored in
memory and can have multiple commands...
    You only have to poke 4-bytes to execute *all* commands.... and
those for bytes can be stored in a sequence.
    And if you only have one command list you needed have to poke
something (only once in the beginning) at all.

    nextGFX will be totally based on the concept of command
lists....afterall...the important stuff (that has to go fast) is the
stuff that you want to do 200 times per sec. It's the same stuff 200
times per sec.. so all the overhead only has to be done once.
 Then you can just call ememcopy. With the tricks by Pete Eberlein svga
can now also be done by ememcopy (you cut it in 4 command lists and bank
in between).
    nextGFX will maintain the same flexibility and will only work with
command lists.. having all the speed overhead of the flexibility when
creating the list, instead of when you execute it. So the flexibility
will no longer be of any speed slowdown.

And then, finally then, we never have to worry about speed anymore....
(NOT!)

Ralf Nieuwenhuijsen
nieuwen at xs4all.nl

new topic     » topic index » view thread      » older message » newer message

Search



Quick Links

User menu

Not signed in.

Misc Menu