what is the difference between ffdshow-20050822.exe and ffdshow-20050822MSVC71.exe

[Yama4050242]

thanx

[Sirber]

GCC compîled vs MS Visual Studio 7.1 compiled?

[neuron2]

@Yama4050242

Please observe forum rules, specifically, rule 9. Thank you.
:readrule:

[Yama4050242]

GCC compîled vs MS Visual Studio 7.1 compiled?
so any difference when i run 2 version? speed? or compatibility? i am not a programer, and i have nothing on the 2 difference compiler

[Shirokuu]

The MSVC project files from the older FFDSHOW sources don't have NASM enabled so it could be the reason the GCC compiled binary is slightly faster. I haven't done a compile for months now so I don't know what the current situation is though

[ExtraEye]

I'd like to know too...

[celtic_druid]

ffdshow-20050822 uses ICL9.
ffdshow-20050822MSVC71 uses MSVC 7.1.
We are talking just ffdshow.ax here.
ICL9 from all reports so far is faster, it however crashes on one particular sample. There could also be other issues.

gcc last time I tried produced a binary that just crashes.

[Yama4050242]

@celtic_druid
thank you for your reply:P

[Shirokuu]

@celtic_druid
thank you for your reply:P
Milan's Tikiwike site is borked, but i remember he was awaiting GCC 4.0 to be released before adjusting FFDSHOW to be compatible with GCC. ( ? )

[hellfred]

Milan's Tikiwike site is borked, but i remember he was awaiting GCC 4.0 to be released before adjusting FFDSHOW to be compatible with GCC. ( ? )
That is correct. I am wondering, why there is still no official MinGW gcc-4.0. When googling, you will find many reports of building gcc-4.0 successfully under MinGW and even binaries.

Hellfred

EDIT: Read here (http://sourceforge.net/mailarchive/message.php?msg_id=11826036)
about problems with gcc.4.0 under MinGW

[celtic_druid]

There were also some cvs updates some time back in regards to getting gcc to work.

"ffdshow can be compiled by GCC: image filters are unoptimized, crashes immediatelly at start support for cross-compilation by Gianluigi Tiesi"

[Shirokuu]

I finally had a chance to play around with both new binaries, both the ICL and MSVC compile. The ICL version chokes on a very regular 900 kbit XVID encode with SPP enabled on notch 2. The MSVC version ran it smoothly without any problems. CPU load spiked on the ICL compile every few frames to 100%, the MSVC compile didn't have this problem.
So on top of celtic_druid's mentioned one-case-crash I would recommend the MSVC build.

[Sharktooth]

I confirm SPP on the ICL build is unuseable on athlon CPUs.

[Shirokuu]

@Sharktooth, i have an Intel Pentium IV HT. ( ? )

[clsid]

That one clip that crashes is a special case. It only crashes with the combo Haali + ffdshow ICL9. Weird thing is though that the clip doesn't crash if I first play some other file. It only crashes if it is the first thing that is played after opening the player.

[ExtraEye]

doesn't happen to me and i use halli splitter for avi's too

[Shirokuu]

For what it's worth i use the Haali splitter (latest version) too. (Sorry if this post is a bit redundant.)

[midiboy]

Hi!

I have opened a new thread for this here (http://forum.doom9.org/showthread.php?t=99837) but my experience on an Intel P4 system is different. At least for the 20050909 version, the MSVC71 compile crashes on 2 files which the ICL build plays fine ...

Maybe I am the only Inteluser here :rolleyes:

Bye,
Alex