[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] 030-version of gcc



...

Frank Naumann wrote:
The aim is still to be concise, easy and elegant.

Oh, so you don't need the sparemint packages and don't want that other
people port existing unix applications to sparemint?

Well, I think that our philosophy are rather opposed. Keep in mind that
I am speaking of ATARI hardware to run those applications. Did you
forgot it ? If freeMiNT is designed to be the free kernel of ARAnyM,
why not ? I personally have an unmodified Falcon 030 (1993), and Hades
060 (1996) that are not powerful enough to run "Unix" applications.
Running ARAnyM on the bi-G4 1.25 GHz I use, starts to be realistic.
You may want that ATARI users use their ATARI as a text console. You
may, as I do, wish another new ATARI platform. I don't see it clearly
coming sooner or later ... Do you ?

I could run NetBSD on my ATARI hardware. Would it worth it ? It already
has been done ... Why should I choose SpareMiNT to run a Unix ? I'm not
convinced ... Running 1.16.0beta kernel caused me a complete disk
crash because you added a cache to the "TOSFS" filesystem. No, I'm
not ready for it ...

This is not a specifical ATARI problem. It's been a long time since
GCC 3.x was first published. You can't ignore it. Things have
progressed.

I don't ignore anything.

That's not true. You have already decided that GCC 3.x is bad, and
you have condemned it without trying it. I personally experimented
it quite a lot, thanks to the work of Patrice Mandin. It took him
a few weeks to adapt the whole GCC 3.3 package, running ARAnyM
nights and days 24h/24h ... Give it a try, please !

I can't do it myself.

And you also can't help to recompile rpms due to a lack of time, right?

That's right. Do you want me to speak about projects that I maintain,
and my ATARI domain activity ?

I badly reacted? I just wanted to clear out that gcc 3.x not only have
advantages and that the advantages must be compared against the
disadvantages. I know this 'guy' personally, he is happy that the cc1 now
compile 5% faster on his Falcon030. Do you really think he like to upgrade
to a cc1 that compile much slower with the win of 1% or 2% in the produced
executables?

First of all, GCC 3.x can be optimized. Patrice made a 68000 version.
The ATARI developer that we are speaking of, made a 68030 version, that
is a very similar work that can be done on GCC 3.x. That's why I first
reacted to his message ...

Second hand, the applications that I'm interested in are quite complex.
You spoke about 1% or 2% speedup. With Persistence Of Vision or OpenGL
it could be quite satisfactory. For the latest OpenGL library that I
maintain (TinyGL 0.45 that will soon be delivered) it is much more !
The gain is around 5% of execution speed. Do we matter if the
compilation time is worse ? And OpenGL is real time ... Think about
execution time of POV, that I also maintain, that is much much more
complex. 5% represent quite a lot of spared time.

Please speak about what you really know. Change your opinion about GCC
3.x. Give it a try !

That's your opinion. And you are heard by a lot of people. Your opinion
is quite important. Let other opinions be explained here. Your interest
may not be shared by everyone.

Yes, I hope other tell their oppinion here too.

I wish it too. This question must be debated !

Bye,

-- Dr. François LE COAT
Author of Eureka 2.12 (2D Graph Describer, 3D Modeller)
http://eureka.atari.org
mailto:lecoat@atari.org