27-Apr-2024 01:55 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 93 items in your selection [1 - 50] [51 - 93]
[Files] SDL 1.2.6 releasedANN.lu
Posted on 08-Sep-2003 09:41 GMT by Gabriele Greco93 comments
View flat
View list
I've just updated the Amiga SDL version to 1.2.6, you can download it in the usual home page and soon through Aminet and libsd.org. The binary package contains the 68k shared library and GCC 68k, GCC MorphOS and SAS/C 68k linker libriaries, StormC 68k and Warpos link libs will be available soon, I've not a suitable compiler to build them.
SDL 1.2.6 released : Comment 1 of 93ANN.lu
Posted by Peter Gordon on 08-Sep-2003 07:49 GMT
Is this the port with the bad busy loops etc. or...?
SDL 1.2.6 released : Comment 2 of 93ANN.lu
Posted by alan buxey on 08-Sep-2003 08:29 GMT
sixk's site seems to be absent right now
SDL 1.2.6 released : Comment 3 of 93ANN.lu
Posted by Gabriele Greco on 08-Sep-2003 08:59 GMT
In reply to Comment 1 (Peter Gordon):
Busy loops?

It's the application that busy loops, not SDL. A common problem is the fact that often people address to the library issues that are of the applications.
SDL 1.2.6 released : Comment 4 of 93ANN.lu
Posted by Rik Sweeney on 08-Sep-2003 09:16 GMT
This is good news, but the problem I have is that when I compile programs my graphics come out trashed unless they're a multiple width of 16. Has this been fixed?
SDL 1.2.6 released : Comment 5 of 93ANN.lu
Posted by justnotbill on 08-Sep-2003 09:18 GMT
@alan buxey

His ports are at http://dasixk.free.fr/
Forum: http://amiga.comscripts.com/
SDL 1.2.6 released : Comment 6 of 93ANN.lu
Posted by Peter Gordon on 08-Sep-2003 09:21 GMT
In reply to Comment 3 (Gabriele Greco):
OK. I was just told that the "official" Amiga SDL port was badly done, and contained busy loops and such.

I have not used it myself, but that is what I heard.
SDL 1.2.6 released : Comment 7 of 93ANN.lu
Posted by Bill Hoggett on 08-Sep-2003 10:52 GMT
In reply to Comment 6 (Peter Gordon):
@Peter

Gabriele is doing a sterling job of porting an Amiga version fairly soon after the base tree is updated, and I suspect he is right about the blame going to the applications for "busy loops". I've seen programmers complain about SDL related CPU usage before when it turns out they made mistakes in their programs causing it.

If anyone thinks Gabriele's port is sub-optimal at all, they should locate the problem and give him feedback about it. Perhaps such feedback might improve the base SDL tree overall...
SDL 1.2.6 released : Comment 8 of 93ANN.lu
Posted by hairy on 08-Sep-2003 10:59 GMT
Just thanks for the new release. Your work is higly appreciated here.

One thing it's busy waiting for sure... it's me!!! (for OpenGL support on MOS ;-)

Ciao
Alessandro
SDL 1.2.6 released : Comment 9 of 93ANN.lu
Posted by Peter Gordon on 08-Sep-2003 11:06 GMT
In reply to Comment 7 (Bill Hoggett):
Don't get me wrong, I didn't mean to insult the guy, I just wanted to find out if this was the "bad" port i'd heard about. If there is no bad port, thats great.
SDL 1.2.6 released : Comment 10 of 93ANN.lu
Posted by miksuh on 08-Sep-2003 11:19 GMT
Any idea when they will release the SDL 2.0 ? I have halted all of my SDL projects and I think I'll continue when SDL supports many windows.
SDL 1.2.6 released : Comment 11 of 93ANN.lu
Posted by miksuh on 08-Sep-2003 11:20 GMT
Good work though :) i hope you continue to support miga SDl :)
SDL 1.2.6 released : Comment 12 of 93ANN.lu
Posted by miksuh on 08-Sep-2003 11:20 GMT
In reply to Comment 11 (miksuh):
Amiga
SDL 1.2.6 released : Comment 13 of 93ANN.lu
Posted by miksuh on 08-Sep-2003 11:25 GMT
What's new in the 1.2.6 ? (I can't download and extract the archive right now.)
SDL 1.2.6 released : Comment 14 of 93ANN.lu
Posted by Gabriele Greco on 08-Sep-2003 12:13 GMT
In reply to Comment 10 (miksuh):
Actually SDL 2.0 is only an idea, the API is not yet finalized and there is not yet a single line of code, so, if you need multiple windows, check for another solution. :)
SDL 1.2.6 released : Comment 15 of 93ANN.lu
Posted by Gabriele Greco on 08-Sep-2003 12:16 GMT
In reply to Comment 4 (Rik Sweeney):
Send me the code and I'll look for the problem, it's almost impossible to debug something without being able to replicate it. And the few issues that may be still left in the SDL Amiga port are not well reported nor reproductable, at least in my enviroment...
SDL 1.2.6 released : Comment 16 of 93ANN.lu
Posted by Gabriele Greco on 08-Sep-2003 12:25 GMT
In reply to Comment 6 (Peter Gordon):
What you hear may be not false, the fact is that the Amiga hardware is SO slow compared to the modern one that you cannot make comparisons.

SDL runs on an Amiga with 040 like it runs on a 486/100 PC win Win95? Maybe it's true since the 040 machine is slower than the 486.

On my 060 Amiga the testsprite program does 22fps in software & windowed mode (the one most SDL games use), on my XP2100/GF4/Linux it does 350fps...

You cannot blame the software if the hardware is so obsolete.
SDL 1.2.6 released : Comment 17 of 93ANN.lu
Posted by Peter Gordon on 08-Sep-2003 13:00 GMT
In reply to Comment 16 (Gabriele Greco):
No, they didn't say it was slow compared to the PC, they said the actual implementation was badly done. They said that there were OTHER *AMIGA* ports which were better.

It is all second hand information, anyway.
SDL 1.2.6 released : Comment 18 of 93ANN.lu
Posted by Fabio Alemagna on 08-Sep-2003 13:28 GMT
In reply to Comment 15 (Gabriele Greco):
> Send me the code and I'll look for the problem, it's almost impossible to debug
> something without being able to replicate it. And the few issues that may be
> still left in the SDL Amiga port are not well reported nor reproductable, at
> least in my enviroment..

Is the "16 bits bug" fixed?
SDL 1.2.6 released : Comment 19 of 93ANN.lu
Posted by tonya on 08-Sep-2003 23:41 GMT
In reply to Comment 18 (Fabio Alemagna):
just tried it , i expected alot....and got nearly nothing ;(

warpsdl is much faster! and less buggy and the source tree is still weirdo.

sixk , next time it will rock i hope!!! , good luck..

but for now i will stick to warpsdl as with that i can actually play the games.
both on 68k and ppc , also if i was to compare my stuff with a pc all day then i would have laughed at pc everyday, so its a bad thing to blame loops , loops is something you should remove from the code as they are cpu intensive on any setup! , ie if u remove it on a pc like u have...u should get 900fps i guess..

a last comment is , this sdl version is better than the last...so small steps but its in the correct direction.

cheers

(i didnt intend to attack anyone with this post....but blaming on looped code + comparing with pc's triggerd me...)
SDL 1.2.6 released : Comment 20 of 93ANN.lu
Posted by hairy on 09-Sep-2003 06:21 GMT
In reply to Comment 19 (tonya):
> just tried it , i expected alot....and got nearly nothing ;(
> warpsdl is much faster! and less buggy and the source tree is still weirdo.
> but for now i will stick to warpsdl as with that i can actually play
> the games.
> both on 68k and ppc

Well, I still prefer it slower, but provided for a broader range of "amigoid" platforms from the same tree. WarpOS is a choice for you, but not for other users/developers (with no BPPC/CVPPC).
Speed can be improved later.

@All
Please, be a little bit more supportive.
Expecially on projects where source is available, and you can try to do better.

Ciao
Alessandro
SDL 1.2.6 released : Comment 21 of 93ANN.lu
Posted by Gabriele Greco on 09-Sep-2003 07:07 GMT
In reply to Comment 19 (tonya):
Ok, so use WarpSDL, WarpSDL is a totally new library not a port at all, when you port a software you must pay attention to compatibility, not only speed. WarpSDL it's only partly API compatible with SDL and can run, maybe, the 5% of the available SDL applications, while the real port it's compatible with the 100% of them (obviously you may need to modify some unix/win32 or mac specific code and many of them may be unusable for the lack of speed). Anyway if you use HWSURFACE, fullscreen rendering and double buffering the performance of SDL on 68k is the same as WarpSDL (that uses them by default). Please note that WarpOS version of SDL can't be optimized too much, since it's the application that calls the api and choose how many context switches to cause every game loop.
Anyway I really like to know where a busy or not busy loop can be in a library that has only a single blocking call, SDL_WaitEvent(), that uses Wait().Fortunately ANN users are not a significant portrait of the amiga user, otherwise I had stopped any amiga development long time ago: if you don't like how other people do things do them better, the only thing you can obtain with the "complaints-only" attitude is that you will lose also the "things-not-done-as-you-think-they-should".
SDL 1.2.6 released : Comment 22 of 93ANN.lu
Posted by tonya on 09-Sep-2003 08:29 GMT
In reply to Comment 20 (hairy):
@hairy:

did i say warpos ???? read it again..i said WARPSDL , i lib for 68k and PPC!

WARP is named warp because of beeing FAST! and its 68k + ppc compliant , soon mos version and so on...
SDL 1.2.6 released : Comment 23 of 93ANN.lu
Posted by tonya on 09-Sep-2003 08:37 GMT
In reply to Comment 21 (Gabriele Greco):
[quote] @All
Please, be a little bit more supportive.
Expecially on projects where source is available, and you can try to do better.[/quote]

funny that you are saying this , i am a developer and i support em all which is worth supporting , i try all the new software and i use what i think is good,
i dont use it because its new , i aint a pc user ... thats all.

warpsdl is an new lib , as in developed from scratch and has been around on amiga for a long time now, a new version should be out before you can say....plonk! , i advice all to loook at that piece of thing and try to understand why i love it.

bloated software is for the bloated computer , go buy a pc if you want that...oh i forgot you own a pc allready it seems ;(

tonya

cheers
SDL 1.2.6 released : Comment 24 of 93ANN.lu
Posted by Peter Gordon on 09-Sep-2003 08:57 GMT
In reply to Comment 21 (Gabriele Greco):
> Fortunately ANN users are not a significant portrait of the amiga user

Hey don't shoot the messenger! I was just asking if it was "the bad port" because I had been told there was a bad port. No offense was meant! I for one am very glad that there are people such as yourself porting useful things to the Amiga. Thanks for all your hard work!
SDL 1.2.6 released : Comment 25 of 93ANN.lu
Posted by Anonymous on 09-Sep-2003 10:05 GMT
> sixk , next time it will rock i hope!!! , good luck..

?
SDL 1.2.6 released : Comment 26 of 93ANN.lu
Posted by Gabriele Greco on 09-Sep-2003 10:50 GMT
In reply to Comment 24 (Peter Gordon):
Sorry, it's not a personal offense, I wrote it about the general mood of the ANN threads, almost any thread can became a flame, everyone has something to say but often their contribution to the Amiga community stop there.
The problem with WarpSDL and similar "self-made" projects is that it's very difficult for a single person to handle all the stuff it's needed in a flexible and complete multimedia library. Often the "bloat" of opensource crossplatform projects introduce a small overhead, but give a lot of new features. Let's talk about YUV overlays, clipping, alpha/colorkey blitting, run length surfaces, complete event handling, unicode, multithread support, endianness support macros, opengl wrapper... just to talk about some of the features warpsdl miss and require a lot of coding to be implemented.
BTW: Why there is not a dev archive in the WarpSDL site?
SDL 1.2.6 released : Comment 27 of 93ANN.lu
Posted by hairy on 09-Sep-2003 12:01 GMT
In reply to Comment 23 (tonya):
> did i say warpos ???? read it again..i said WARPSDL , i lib for 68k and PPC!
> WARP is named warp because of beeing FAST! and its 68k + ppc compliant ,
> soon mos version and so on...

Quote from ChaoZers homepage: "* PPC (WarpOS) support"
Put it as you like, PPC support is offered only for WarpOS kernel, and this is exactly my point: It may work on systems offering WarpOS emulation, or may not. SDL from Gabriele also has MOS native link library.


> warpsdl is an new lib , as in developed from scratch and has been around on
> amiga for a long time now, a new version should be out before you can
> say....plonk! , i advice all to loook at that piece of thing and try to
> understand why i love it.

Don't get me wrong, I think it's good to have more amiga-specific and maybe more efficient ports of libraries.
But it's also important to have library conformance with the main source tree, expecially for a multi-platform library like this. Conformance allows SDL games and apps to be converted to our operating system(s) at much more speedy rate, with minimal effort.
It's the same with MiniGL and OpenGL. Sure I would not complain about Mesa 5.x being slow, if ever ported!

So let's look first at the ADVANTAGES of having both an amiga-specific and a conformant library, and later at limitations and comparisons.
Limitations are best expressed in a detailed bug report, not like "it's slower, it's crap, it adds nothing, etc...".


> bloated software is for the bloated computer , go buy a pc if you want
> that...oh i forgot you own a pc allready it seems ;(

I used to think the same, but I have partially changed my mind about this.
Bloated things like Cubase, Mozilla or Maya, or even look-alikes, are badly needed on our platforms.
(With "platforms" I mean AmigaOS/MorphOS/AROS/Amithlon. Oh, and... YES, I also have *MANY* PCs... but never had to sign anything with my blood just for that ;-)

Peace, Yo!
Alessandro
SDL 1.2.6 released : Comment 28 of 93ANN.lu
Posted by tonya on 09-Sep-2003 12:37 GMT
In reply to Comment 27 (hairy):
gabrielle: warpsdl dev kit will be released very soon , mos and wos and 68k.

hairy: resistance is futile... i wasnt flaming i was just telling what i thought of the thing.
SDL 1.2.6 released : Comment 29 of 93ANN.lu
Posted by tonya on 09-Sep-2003 12:42 GMT
In reply to Comment 27 (hairy):
@hairy: ok , it seems it all was put to rest there , remeber chaozer does have a pegasos and yes it will come very soon! (warp sdl) for it , and thats when you will see what speed is about , maybe it was awesome on pegasos and mos1.4 , i dont know (SDL) as i havent checked it on that yet, all that i saw was endless or repeated code, but it was better than the last one so i am not screaming CRAP WORK etc, i am just saying the problems should have been fixed before it was released or!!!!! it might be forgotten very fast ;( and thats a shame , i am a developer myself and well i know jackshit about C++ atm but i am learning it now , all i have done till now is assembler.

tonya / push entertainment.
SDL 1.2.6 released : Comment 30 of 93ANN.lu
Posted by Gabriele Greco on 09-Sep-2003 13:08 GMT
In reply to Comment 29 (tonya):
I wanted only to make a simple test with the classic SDL test programs as benchmark, just to know how much bloated is the full port :) Hope to see a development kit soon :)
SDL 1.2.6 released : Comment 31 of 93ANN.lu
Posted by Gabriele Greco on 09-Sep-2003 13:55 GMT
In reply to Comment 30 (Gabriele Greco):
Anyway I really doubt that with a well written application the speedup of WarpSDL over SDL can be over the 30%.

I code for Amiga since 15 years (in 1992 I used to write commercial games for the italian software house Simulmondo for instance...) and I doubt the videodriver I wrote is so "bloated", it can have problems with certain resolutions/graphics card with P96, but this is mainly dued by the fact P96 returns wrong pixel format info in the graphics.library calls and the fact I don't own a few different graphic cards to test every possible mode / format combination... It's a free project from which I don't get a single dime, you cannot expect it works on every possible scenario, it would be different with a commercial interest behind like the linux or the OSX version... Anyway it's not money what I need, it's time, nowaday I work as senior software engineer in an italian software house and I get for a week of work more money than I got in a year from the royalities of ETW (a commercial amiga soccer game I wrote in '98)... I've so much work I have to often refuse it, well payed work of course, if I use part of my spare time to code for Amiga projects it's only since I'm still intrested in supporting this plaform, I've loved so much in the past.

Bye,
Gabry
SDL 1.2.6 released : Comment 32 of 93ANN.lu
Posted by tonya on 09-Sep-2003 14:38 GMT
In reply to Comment 31 (Gabriele Greco):
gabry: yeah ok , nice post there .. and i agree with u on many aspects of it.

why not just try the 68k version of warpsdl or run it on a wos based machine? then compare warpsdl with sdl, you might see some differences..

anyway i never wanted to attack you , keep up the good work and i will certainly give you feedback , i aint a troll but i am from norway (LOL) :)

cheers

pps:ETW works better in sdl version than in my orginal but i dont have mos1.4 nor pegasos so it aint exactly fast atm ;( ... i hope a warpsdl version will come soon :))))
SDL 1.2.6 released : Comment 33 of 93ANN.lu
Posted by hairy on 09-Sep-2003 14:52 GMT
In reply to Comment 29 (tonya):
> hairy: resistance is futile... i wasnt flaming i was just telling what i
> thought of the thing.

Nice idea... BorgSDL!!! ;-)
Ok. No flaming here, just... talking? :-)
I'll give WarpSDL a try if MOS version is available, or at least 68k for now.

Ciao
Alessandro
SDL 1.2.6 released : Comment 34 of 93ANN.lu
Posted by tonya on 09-Sep-2003 14:57 GMT
In reply to Comment 33 (hairy):
rofl! :)

sorry i am abit of a [b]trekkie :)
SDL 1.2.6 released : Comment 35 of 93ANN.lu
Posted by StAn on 09-Sep-2003 15:48 GMT
void SDL_Delay (Uint32 ms)
{
// Do a busy wait if time is less than 50ms

if(ms<50)
{
clock_t to_wait=clock();

#ifndef __SASC
ms*=(CLOCKS_PER_SEC/1000);
#endif
to_wait+=ms;

while(clock()<to_wait);
}
else
{
Delay(ms/20);
}
}
SDL 1.2.6 released : Comment 36 of 93ANN.lu
Posted by Adam Waldenberg on 09-Sep-2003 17:54 GMT
In reply to Comment 26 (Gabriele Greco):
dev archive and new version is on the way.. There is no dev archive.. because there is no point releasing one untill warpsdl has reached final stages. (or is getting therer) (And its there now basically...)
SDL 1.2.6 released : Comment 37 of 93ANN.lu
Posted by Adam Waldenberg on 09-Sep-2003 17:57 GMT
In reply to Comment 26 (Gabriele Greco):
And btw... WarpSDL has all the above mentioned features.. it implements basically 90% of the SDL api now.... And everything is done much better :)... Also... I have hand optimized (ppc/68k asm) several parts of the code.... And oh ... minimizing context switchs is not hard at all.. haven't you heard about fat libraries with separate ppc and 68k functions?
SDL 1.2.6 released : Comment 38 of 93ANN.lu
Posted by Adam Waldenberg on 09-Sep-2003 18:11 GMT
In reply to Comment 31 (Gabriele Greco):
Yes... I have to say it's great you did the SDL port :)... It was badly needed.... And it made me realize that we should have something less bloated and properly optimized.... But as long as WarpSDL is not done and/or not compatible enough ... Your SDL is badly needed :) ....

Over and out :)...

I have a request tho .... Scrap the linker libs .... If you would only have shared libraries only people would use that and I could just make a SDL.library->WarpSDL.library wrapper to get the other stuff "working"... It could be an idea :)
SDL 1.2.6 released : Comment 39 of 93ANN.lu
Posted by mark on 09-Sep-2003 23:30 GMT
Does anyone know if it's possible to compile SDL programs with StormC 3?

I get an error in SDL_video.h, "illegal use of bitfields", for the structs SDL_VideoInfo and SDL_Overlay.
SDL 1.2.6 released : Comment 40 of 93ANN.lu
Posted by hairy on 10-Sep-2003 06:31 GMT
In reply to Comment 35 (StAn):
> void SDL_Delay (Uint32 ms)
> {
> // Do a busy wait if time is less than 50ms
> if (ms<50) {
> clock_t to_wait=clock();
> to_wait+=ms;
> while(clock()<to_wait);
> } else {
> Delay(ms/20);
> }
> ...

And this is the nasty-nasty-evil-evil loop?!?!?

As you see, Delay() wants an argument with a granularity of 50ms. So you should blame the AmigaOS function for that.
Also, worst case latency could be even higher. Anyone measured it?

I admit my ignorance here, forgot if the exec got a better function to sleep a task (only used ixemul lately).

Suggestions?

Ciao
Alessandro
SDL 1.2.6 released : Comment 41 of 93ANN.lu
Posted by Adam Waldenberg on 10-Sep-2003 07:22 GMT
In reply to Comment 40 (hairy):
You don't know much about AmigaOS, do you? :).. Why would someone even consider using Delay for something like this?
SDL 1.2.6 released : Comment 42 of 93ANN.lu
Posted by Fabio Alemagna on 10-Sep-2003 07:31 GMT
In reply to Comment 35 (StAn):
> Do a busy wait if time is less than 50ms

50ms = 0.005 secs. Using the timer for that little period of time would result in wasting the same amout of cycles, if not more, and wouldn't give the same precision.
SDL 1.2.6 released : Comment 43 of 93ANN.lu
Posted by Fabio Alemagna on 10-Sep-2003 07:35 GMT
In reply to Comment 41 (Adam Waldenberg):
> You don't know much about AmigaOS, do you? :).. Why would someone even consider
> using Delay for something like this?

Even if you used timer.device directly, using it would involve task switches and all sorts of latencies.
SDL 1.2.6 released : Comment 44 of 93ANN.lu
Posted by Adam Waldenberg on 10-Sep-2003 08:08 GMT
In reply to Comment 42 (Fabio Alemagna):
No way.. It's way faster than that. If you'r theory was correct and it was so slow, nothing would be usable... And oh .. how do you think clock() measures time?? its either timer.device or the functions in lowlevel afaik. (or internal ppc timers if PPC . and thats what you use anyway..) Its stupid to call it again .. and again... and again :) ....

Ayway .. I get tired of speaking about the obvious here, enough of this.
SDL 1.2.6 released : Comment 45 of 93ANN.lu
Posted by Adam Waldenberg on 10-Sep-2003 08:14 GMT
In reply to Comment 43 (Fabio Alemagna):
> Even if you used timer.device directly, using it would involve task switches > and all sorts of latencies.

Sorry to say it.. But this is probably one of the most stupid things I have heard in ages :)
SDL 1.2.6 released : Comment 46 of 93ANN.lu
Posted by Gabriele Greco on 10-Sep-2003 10:44 GMT
In reply to Comment 43 (Fabio Alemagna):
Yep, the theoretical mimimal latency available in AmigaOS is 20ms, since the scheduler is run 50 times per second, as usual if you start waiting just AFTER a scheduler call you'll have 39ms of latency, this is why I used that 50ms.

About clock():

A few years ago I've benchmarked timer.device/GetSysTime() command and clock(),
the second one was faster on my setup so I choose to use it. The timer.device init code is anyway available in the SDL_systimer.c amiga code, just to make clear the fact that clock() is not there only to avoid writing some code...
SDL 1.2.6 released : Comment 47 of 93ANN.lu
Posted by hairy on 10-Sep-2003 11:13 GMT
In reply to Comment 45 (Adam Waldenberg):
> You don't know much about AmigaOS, do you? :).. Why would someone even
> consider using Delay for something like this?

Correct, my brain has limited storage capacity. Had to relocate the AGA register table also! :-)
POSIX systems have sleep()/usleep(). I guess Delay() is somewhat the AmigaOS equivalent. So it was the first choice while waiting for a better solution.

>> 50ms = 0.005 secs. Using the timer for that little period of time would
>> result in wasting the same amout of cycles, if not more, and wouldn't
>> give the same precision.
...
>> Even if you used timer.device directly, using it would involve task
>> switches and all sorts of latencies.

> No way.. It's way faster than that. If you'r theory was correct and it
> was so slow, nothing would be usable... And oh .. how do you think clock()
> measures time?? its either timer.device or the functions in lowlevel afaik.
> (or internal ppc timers if PPC . and thats what you use anyway..) Its stupid > to call it again .. and again... and again :) ....

Fabio is right. The task has to be awakened by a kernel event. Unless you're suggesting a bad hack (TM).
Busy waiting for a few cycles is a common technique (or maybe the ONLY technique), whenever you require precision acting on an edge signal.

> Ayway .. I get tired of speaking about the obvious here, enough of this.

Suppose you want to catch the train tomorrow, you set your alarm 30 minutes in advance, sleep all night, and only then busy wait for it to come.
No way the train conductor would come to wake you, and still arrive at the next station on schedule... oh, sorry for the explanation of the obvious!

Take it easy!!! We're not running for the "timer prize"! ;-)

Ciao
Alessandro
SDL 1.2.6 released : Comment 48 of 93ANN.lu
Posted by Fabio Alemagna on 10-Sep-2003 12:46 GMT
In reply to Comment 44 (Adam Waldenberg):
> No way.. It's way faster than that.

you never know. There are task switches involved there, your task might be put to sleep for too long. Indeed, the average latency should be detected at runtime and then use that information to calibrate that loop, but the loop itself is not evil.

> If you'r theory was correct and it was so slow, nothing would be usable... And
> oh .. how do you think clock() measures time??

By subtracting the current time from the time measured at program's startup, and this operation doesn't involve any task switches.


> by its either timer.device or the functions in lowlevel afaik. (or internal
> ppc timers if PPC . and thats what you use anyway..) Its stupid to call it
> again .. and again... and again :) ....

It's not stupid, it gives you less latency. It's not function calls time the problem, it's scheduling.

> Ayway .. I get tired of speaking about the obvious here, enough of this.

Be less presumptuous, as you evidently have no clue about all this.
SDL 1.2.6 released : Comment 49 of 93ANN.lu
Posted by Fabio Alemagna on 10-Sep-2003 12:47 GMT
In reply to Comment 45 (Adam Waldenberg):
> Sorry to say it.. But this is probably one of the most stupid things I have
> heard in ages :)

Prove me wrong. if you can. I'm waiting.
SDL 1.2.6 released : Comment 50 of 93ANN.lu
Posted by StAn on 10-Sep-2003 18:51 GMT
A task switch can happen at any time anyway. I don't think there is a point in busy waiting here to increase the precision.
Anonymous, there are 93 items in your selection [1 - 50] [51 - 93]
Back to Top