Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.

All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Questions and Answers for AROS AMIGA OS

Building for AROSpage  1 2 3 

Kamelito Loveless

Posts 245
18 Jan 2020 07:51

As FBlit source code is available it could be directly be integrated in AROS source tree. I guess it could then be build using the original C or the Asm version using the CPU. I wonder how all AmigaOS fonctions that use the blitter could be replaced with CPU version that can be use multiple Time {read in paralell) without having to do ownblitter() disownblitter()

Wawa T

Posts 695
18 Jan 2020 13:16

this has been discussed and rejected long ago. fblit is asm. afair aros already uses cpu to blit stuff if it is appropriate, that means on faster cpus, actually on those you would want to use aros with in the first place.

maybe it needs to be reviewed, but fblit as such wont be integrated. what else functionality it provides you would want to have implemented in aros?

Kamelito Loveless

Posts 245
18 Jan 2020 13:25

Faster execution time than C produced code? It could also be optimized by using better 080 operands and or AMMX. Since the CPU is already doing the job there’s no need for own/disown() blitter code. If many programs use the Amiga calls that  use the blitter and the blitter is replaced by CPU code there’s no reasons to lock those calls right? So it should speed up the performance I guess.

Wawa T

Posts 695
18 Jan 2020 14:24

i understand the saga hidd is already asm ammx optimised. what else would that achieve.

Ronnie Beck
(Apollo Team Member)
Posts 199
18 Jan 2020 15:14

Martin White wrote:

Ronnie Beck wrote:

  I don't think you know who is in charge of the project(you mean SDL right?

I don’t care who’s in charge of that.

Ladies and gentleman, I give you Francesco Schettino.

Gunnar von Boehn
(Apollo Team Member)
Posts 6125
18 Jan 2020 15:31

Nixus Minimax wrote:

Gunnar von Boehn wrote:
APOLLO CPU can fully saturate the memory interface.
  So if you do pure "memcopy/blitting" stuff Apollo does this today already perfectly.

  Yes, sure. But the CPU can't do numbercrunching at the same time as doing soft blits. If the blitter does its blit jobs on the memory interface and the CPU does register-intensive numbercrunching executing code that is in the instruction cache, more work is being done in the same time than if the CPU does both the blit jobs AND the numbercrunching.

Yet, Hyperthreading does allow to do exactly this!

You can with one thread do memory operations..
and with a 2nd thread do math.
So yes.
A CPU can do exactly such two jobs in parallel.

But before you ask now to add hyperthreading to AMIGA OS.
Lets go back and review carefully if what you want is REALLY used?

Can you give a real world example benefiting of such paralism?
AMIGA OS in generally would not benefit because normally after a Blitjob / Waitblit is called.

Kamelito Loveless

Posts 245
18 Jan 2020 21:08

If one blitter functions is called by a task/process and another one is also doing this instead of forbidding a blitter job because the 1st task/process is using it (own blitter) you could add a flag in blitter functions and if another call is done you could do another blitter job in parallel. Disownblitter() decrease the flag and so on. So in theory instead of waiting you could do 2 blitter operations in parallel right?

posts 47page  1 2 3