 |
|
|
CT60 and related things BBS
| Re: + & - CT60 PCI vs SuperVid |
|---|
|
Posted by: Johan Klockars
|
May,23.2006-23:53
|
> In Aranym (I use it very sporadicly) it's just horrific slow.
Then you have a very unusual system, or you do something very strange.
Possibly your lack of JIT makes a big difference, or the SDL and/or OpenGL native side drawing code is not up to scratch on the Mac.
On my five year old PC ARAnyM(JIT)/fVDI is faster than any other Atari compatible machine I've ever heard about (at most things).
The Kronos VDI test just clocked in here at about the same average numbers as the Hades060 configuration (from the default Kronos performance files, which is itself more than twice that of the CT60-100-25) when running with the SDL driver, and eight (yes 8!) times faster than that when using the OpenGL driver (more than four times faster than the PM8600 MagicMac 2.2 numbers).
Without JIT it dropped down to between the CT60 and the Hades scores in SDL mode, but stayed slightly better than the PM8600 in OpenGL mode.
Even AB040/Eclipse normally beats the various other graphics card solutions, at most operations (not circles and things like that with the current implementation, IIRC).
> The videl emulation with NVDI is much quicker. I'm on a dual G5. The speed makes it unuseable there.
Well, none of the core ARAnyM developers work on Macs, so there could be some strange problems with performance of the native side graphics implementions (did you use SDL or OpenGL?) there.
But what you are describing sounds absolutely insane!
> When I tested it on my Falcon some years ago, the speed was horrific as well (then without graphics card).
I don't know what graphics mode you tested there, but the monochrome driver used to beat NVDI at many things on my Falcon, and tie it on most of the rest.
There never really was a useful driver for the other bitplane graphics modes before late last year (only 16 colour mode ever worked, and that limited most drawing to black (but it did beat NVDI significantly on scrolling on my AB040)), so I don't think you would have tried that. The 16 bit driver, however, was (and still is) really only an example driver, implemented in 99+% fairly unoptimized C, so it should definitely be slower than NVDI (but I doubt there would be a "horrific" difference).
> With eclipse, I've seen it trying to run all sort of things, f.ex. gem apps
"Trying to"? As in failing?
> like Imagecopy didn't even run properly.
I wouldn't use the word "even" there. IIRC, imagecopy is a big user of the NVDIv5 blit modes, so it does indeed not have much chance of running properly until someone completes that stuff.
Calamus SL (only in 32 bit colour at the moment), HighWire, PapyrusX, etc all seem to work reasonably well (the vector font support is not yet 100% compatible), as well as just about everything else I've ever tried.
> They did on the nova however.
That's the thing I've been hearing about. It must mean that there is some kind of special thing the VDI can do to make NVDI support those new blit functions even though it does not control the screen itself. NVDI on top of fVDI simply refuses to do that currently (it does not even pass the calls along to fVDI).
Of course, not having any developer documentation for NVDI makes it somewhat difficult to figure out what can be done to make this work.
> Missing NVDI5 comaptibility is a big flaw as well.
Well, if I had also been making tens of thousands of dollars on my VDI, I might have been able to find more time to work on those things. ;-)
As it is, I try to find the time to spend on the issues that people actually ask me to look at.
> So in a few words: Speed and compatibility.
The speed problem is only in your machine (and possibly other Macs).
Why? I have no idea. Try it on a PC running Linux (I'd guess it should work well on a Windows build of ARAnyM too, but I haven't tried that for years), or for that matter an Eclipse/RageII.
> > You'll have access to just about every NVDI feature except for the NVDIv5 blits (and even that is likely fixable,
>
> This is just the problem.. "likely fixable". Eg, it isn't fixed, and nobody seems to be fixing it.
It probably is some kind of simple fix, but without any idea about what it might be... :-(
Perhaps someone has a NOVA board and would be willing to investigate what NVDI does there to make those blit functions work?
Tracing around in the NVDI internals trying to figure out what's going on isn't a job for the faint-hearted, though.
|
[All messages in this thread] [Start new thread]
|
Topic
|
Posted by
|
Date
|
|
+ & - CT60 PCI vs SuperVidel
|
raphael
|
May,21.2006-21:38
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,21.2006-21:56
|
|
Re: + & - CT60 PCI vs SuperVid
|
Crash
|
May,22.2006-12:23
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,22.2006-17:51
|
|
Re: + & - CT60 PCI vs SuperVid
|
Strider
|
May,22.2006-16:20
|
|
Re: + & - CT60 PCI vs SuperVid
|
earx
|
May,22.2006-16:53
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,22.2006-17:49
|
|
Re: + & - CT60 PCI vs SuperVid
|
CiH
|
May,22.2006-18:20
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,23.2006-00:20
|
|
Re: + & - CT60 PCI vs SuperVid
|
vlg
|
May,23.2006-11:51
|
|
Re: + & - CT60 PCI vs SuperVid
|
Peter
|
May,23.2006-12:42
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,24.2006-00:22
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,24.2006-00:15
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,23.2006-15:00
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,23.2006-23:53
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,24.2006-13:27
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,24.2006-14:59
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,24.2006-16:19
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,24.2006-19:41
|
|
Aranym Mac
|
CiH
|
May,24.2006-21:26
|
|
Re: Aranym Mac
|
Coda
|
May,29.2006-09:53
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,25.2006-00:06
|
|
Re: + & - CT60 PCI vs SuperVid
|
evil
|
May,26.2006-18:05
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,27.2006-19:36
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,27.2006-19:56
|
|
Re: + & - CT60 PCI vs SuperVid
|
Johan Klockars
|
May,25.2006-14:29
|
|
Re: + & - CT60 PCI vs SuperVid
|
hencox
|
May,30.2006-00:08
|
|
Re: + & - CT60 PCI vs SuperVid
|
kas1e
|
Jun,02.2006-09:26
|
|
Re: + & - CT60 PCI vs SuperVid
|
kas1e
|
Jun,02.2006-09:30
|
What's the anti-troll code?
That's your personal code to be able to add comments and messages on the dhs.nu site.
Don't have a code or forgot it? Fix it here.
|
|
|
