[cad-linux-dev] Re: uber-converter
- From: "cr88192" <cr88192@xxxxxxxxxxx>
- To: <cad-linux-dev@xxxxxxxxxxxxx>
- Date: Thu, 10 Jun 2004 12:14:47 +1000
----- Original Message -----
From: "Bruno Postle" <bruno@xxxxxxxxxx>
To: <cad-linux-dev@xxxxxxxxxxxxx>
Sent: Thursday, June 10, 2004 11:36 AM
Subject: [cad-linux-dev] Re: uber-converter
> On Tue 08-Jun-2004 at 10:11 +1000, cr88192 wrote:
>
> > an issue I had wondered about was how to handle compression, basically
if I
> > should compress enough data to fill a certain buffer (possibly
problematic)
> > or if I should compress blocks of data and grab enough archive blocks to
> > hold the compressed block (eg: the file is compressed in 16 or 64 kB
> > chunks), with the end being padded.
> > this could be problematic as it would be waseful of space due to
internal
> > fragmentation, and wouldn't be succeptible to end packing.
>
> These problems of packing small files are kernel/filesystem issues,
> you shouldn't have to solve them in a userspace application.
>
> It doesn't do compression, but ReiserFS does a lot of this already
> for free:
>
> http://www.namesys.com/
>
they shouldn't, but things like reiserfs are not ubiquitus...
I, for one, am using windows, which has a fairly crappy fs (ok, ntfs is
decent but most of my drives are fat32 for linux compatibility...).
so, as a result, I have to do this kind of thing in userspace...
(just like I have to do image scaling because gl only accepts power of 2
sized textures...).
now I have beating at this one for a while, and I have come up an algo which
can scale images almost as fast as I can use memcpy to copy the spans, yet
this is still an expensive operation...
then I am stuck in the fact that I can't tell whether the c version or the
assembler version is faster...
this is mostly all because I want to use videos as textures, and don't want
a crappy framerate (eg: 12 fps, though a lot of time is going into things
like: glTexImage2d, the code to mix audio samples, the divx/mp3 decoder,
..).
actually, just dropping teximage boosts it up to about 40fps or such, but
then I can't see the video on my polys...
probably things would be faster if I didn't decide to write a lot of this
myself, but this is just my way (doing it myself means at least I can port
to linux fairly easily if needed...).
hmm, I could just use glWritePixels and at least have probably a general
video player that doesn't seem to reboot my computer when I use it...
all for now.
Other related posts:
