Do we really need to wait for rosetta with this?
graphics accel!
graphics accel!
by slowdive » Tue Jun 21, 2005 10:37 pm
I would like to hear updates on the graphics accel things going on. someone posted terrific things on the devel-list.
Do we really need to wait for rosetta with this?
Do we really need to wait for rosetta with this?
- slowdive
Re: graphics accel!
by Guest » Tue Jun 21, 2005 11:52 pm
This is what Jens van der Heydt writes:
[quote:027233e7e6]
Hi,
as pretty much all of the PearPC Users are eager to have hardware accelerated graphics, I'd like to announce that I was able to finally write an acceleration module for OS X.
I am receiving surface -blit and -fill calls finally. As far as I'm concerned this is great news because a basic kind of acceleration is possible now. I've teamed up with Daniel who's writing the Framebuffer stuff (video.x replacement) while I'm writing the Host Interface for the Acceleration. Nothing"s finished yet, so it will take some more time but the biggest problem to actually write an accelerated Driver for OS X is done.
We still don't have hardware accelerated Open GL, but everything the software Renderer produces (Surfaces) will be blitted by the Host Hardware with these new drivers. So 2/3 of the whole Driver Framework is finished :-) Boy, am I glad to have found the solution.
Jens
[/quote:027233e7e6]
Hope that gives you the idea.
PearPCUser
[quote:027233e7e6]
Hi,
as pretty much all of the PearPC Users are eager to have hardware accelerated graphics, I'd like to announce that I was able to finally write an acceleration module for OS X.
I am receiving surface -blit and -fill calls finally. As far as I'm concerned this is great news because a basic kind of acceleration is possible now. I've teamed up with Daniel who's writing the Framebuffer stuff (video.x replacement) while I'm writing the Host Interface for the Acceleration. Nothing"s finished yet, so it will take some more time but the biggest problem to actually write an accelerated Driver for OS X is done.
We still don't have hardware accelerated Open GL, but everything the software Renderer produces (Surfaces) will be blitted by the Host Hardware with these new drivers. So 2/3 of the whole Driver Framework is finished :-) Boy, am I glad to have found the solution.
Jens
[/quote:027233e7e6]
Hope that gives you the idea.
PearPCUser
- Guest
It's not finished yet
by Jens » Wed Jun 22, 2005 4:31 am
I am still writing it. The current Alpha that I am working on
does not draw to the screen at this current stage. You wouldn't
want to have OS X tell you
"rect update at 10,10 size 500,400"
"blit surface ID 3 at 15,20"
But having finally written this driver in the first place,
even without having a host implementation yet, is a very big
step. It took about one year to find these things. So, for me,
it's a big step. I think that we will have a beta in around 1-2 weeks.
Maybe more, maybe less.
Jens
does not draw to the screen at this current stage. You wouldn't
want to have OS X tell you
"rect update at 10,10 size 500,400"
"blit surface ID 3 at 15,20"
But having finally written this driver in the first place,
even without having a host implementation yet, is a very big
step. It took about one year to find these things. So, for me,
it's a big step. I think that we will have a beta in around 1-2 weeks.
Maybe more, maybe less.
Jens
- Jens
- PearPC developer
- Posts: 244
- Joined: Fri Jan 07, 2005 11:22 am
by Jens » Wed Jun 22, 2005 3:16 pm
[quote="Anonymous":2792111809]
Good going Jens!
Was this a process of trial and error, or did you finally find some documentation on the subject that didn't require an NDA with Apple?
[/quote:2792111809]
It was a mix of trial&error, reading deprecated Documentation,
and finally finding some hints in the darwin sources. You could
say, that it was all there for everyone to read, you just had
to find it and figure out how to use it. It's still
not working 100% as expected but I'm getting there.
Good going Jens!
Was this a process of trial and error, or did you finally find some documentation on the subject that didn't require an NDA with Apple?
[/quote:2792111809]
It was a mix of trial&error, reading deprecated Documentation,
and finally finding some hints in the darwin sources. You could
say, that it was all there for everyone to read, you just had
to find it and figure out how to use it. It's still
not working 100% as expected but I'm getting there.
- Jens
- PearPC developer
- Posts: 244
- Joined: Fri Jan 07, 2005 11:22 am
by Jens » Wed Jun 22, 2005 5:40 pm
[quote="PearPCUser":bf2a61cd67]
That sounds awesome, Jens! - Is there already an estimation how much this will speed up PearPC? Nothing guaranteed, of course, just to have an idea ...
PearPCUser
[/quote:bf2a61cd67]
Well, it is really hard to tell because I never saw it drawing because
that is the part that I am currently writing.
From my understanding of PearPC's biggest problems - meaning
slow Memory Access - I tend to say that it will be a big difference
to what we see now. The Acceleration Module will reduce the number
of Memory Accesses significantly because we can do surface-to-surface
copies on the Host Hardware, 2nd Filling Surfaces is going to be
done in Host Hardware too.
I think we will have very useable graphics performance with this.
That sounds awesome, Jens!
- Is there already an estimation how much this will speed up PearPC? Nothing guaranteed, of course, just to have an idea ...PearPCUser
[/quote:bf2a61cd67]
Well, it is really hard to tell because I never saw it drawing because
that is the part that I am currently writing.
From my understanding of PearPC's biggest problems - meaning
slow Memory Access - I tend to say that it will be a big difference
to what we see now. The Acceleration Module will reduce the number
of Memory Accesses significantly because we can do surface-to-surface
copies on the Host Hardware, 2nd Filling Surfaces is going to be
done in Host Hardware too.
I think we will have very useable graphics performance with this.
- Jens
- PearPC developer
- Posts: 244
- Joined: Fri Jan 07, 2005 11:22 am
by PearPCUser » Wed Jun 22, 2005 5:59 pm
[quote="Jens":26eea191ec]
Well, it is really hard to tell because I never saw it drawing because that is the part that I am currently writing.
From my understanding of PearPC's biggest problems - meaning
slow Memory Access - I tend to say that it will be a big difference
to what we see now. The Acceleration Module will reduce the number
of Memory Accesses significantly because we can do surface-to-surface
copies on the Host Hardware, 2nd Filling Surfaces is going to be
done in Host Hardware too.
I think we will have very useable graphics performance with this.
[/quote:26eea191ec]
Thank you for these detailed informations, Jens - that's more than I could ever hope to read - and it sounds great! :-)
PearPCUser
Well, it is really hard to tell because I never saw it drawing because that is the part that I am currently writing.
From my understanding of PearPC's biggest problems - meaning
slow Memory Access - I tend to say that it will be a big difference
to what we see now. The Acceleration Module will reduce the number
of Memory Accesses significantly because we can do surface-to-surface
copies on the Host Hardware, 2nd Filling Surfaces is going to be
done in Host Hardware too.
I think we will have very useable graphics performance with this.
[/quote:26eea191ec]
Thank you for these detailed informations, Jens - that's more than I could ever hope to read - and it sounds great! :-)
PearPCUser
- PearPCUser
Who is online
Users browsing this forum: No registered users and 2 guests