Has the release of DVD2AVI 1.77 any impact on SF.net project?

[tiki4]

Hi,

Jackei seems to be back in the coding scene and had released a new version ov DVD2AVI (of course you know that!). I know that there is something going on they call SAVE-OE, a SF.net project based on DVD2AVI. So has the release of Jackei's version any impact on this project? I think most of us believed that Jackei doesn't maintain his program any longer...

I'm just curious, but I guess everybody is watching soccer right now. Not too good that we now have only 10 guys left...


tiki3

[Nic]

:) Tom & the gang have been far too quiet recently haven't they.....

....I wonder if Jackei is answering emails then now...Worth a shot I guess.

I guess i'd better put that (or other) P4 support into my version....

-Nic

[trbarry]

I haven't been doing anything on DVD2AVI (Save-oe) recently but I haven't totally abandoned it.

(but I'm waiting for you to post your source Nic ;) )

And the Sourceforge save-oe version already had P4 optimized SSE2 code (reportedly now broken and turned off) so I'm not sure Jackei's new version affects save-oe that much. But it's nice to know Jackei is interested again.

My own working save-oe version of DVD2AVI with P4 speed-up's is still posted at www.trbarry.com/DVD2AVIT3.zip and www.trbarry.com/MPEG2DEC2.zip . Eventually I'll probably go in and fix the P4 stuff in Sourceforge if another P4 user hasn't done it first. So it would be nice if someone (with a P4) tested the relative P4 speed of 1.77 vs the one I linked above (I haven't). If Jackei's is faster then I'll pilfer his code when I go back to fix Save-oe. ;)

But I haven't checked out Jackei's new version yet so maybe it has other new features I haven't heard about.

- Tom

[Nic]

(but I'm waiting for you to post your source Nic )

:) & im waiting for me to finish it :D
(but you can have it in its unfinished state at present if you want Tom, just shout me....but excuse the mess :) )

(I was compiling your MPEG2DEC2 yesterday, its weird, its about the only thing I cant get the Intel Compiler 6.0 to optimise better than vc6.... :confused: )

Cheers,
-Nic

[vlad59]

Hi,

Sorry, I've been very very busy (And now I'm only busy ;) )

@Tom

I haven't commited any of my changes on CVS so your P4 motion compensation code is still valid and active. My brother has just bought a P4 so I'll be soon able to debug my crappy SSE2 stuff myself.

The only thing I made on CVS is to add mpeg2dec (your T2 or t3 zip).

From what I have seen of Jackei'code, he added a P4 IDCT (from Dmitry) and some minor SSE2 optimization in GetPic.c (Add_block, clear_block).

I think it could be a good idea to contact jackei to have his feelings about that.

Bye, time to get back to work........ :scared:

[trbarry]

From what I have seen of Jackei'code, he added a P4 IDCT (from Dmitry) and some minor SSE2 optimization in GetPic.c (Add_block, clear_block).

Vlad -

Those P4 functions in save-or already are (my versions of) DmitryR's code, plus some others contributed by Dr. Frankenstein.

But I'm glad to hear that the save-oe P4 functions are still intact. I think it might be easier in the future to have a totally separate GetPic_SSE2 member for the P4 stuff, and just keep the existing P4 code there with the run time checks for P4 stripped out. We could just check the CPU type before calling GetPic or GetPic_SSE2.

Once a member diverges past a certain point it's often simpler just to keep a separate member name. It makes some enhancement harder but some maintenance easier. Then folks without a P4 could optimze K6-II & stuff more simply. And there is starting to be quite a bit of code difference so removing the extra run time checks might even help performance a bit.

- Tom

[tiki4]

Funny, isn't it!

I just wanted to see what's happening at SF.net, now there's a new release from the original author and what I've got is a discussion of the programmers.
I thought the SF project was created because everybody believed Jackei won't maintain DVD2AVI any longer. Now it looks like there will be independent lines of the same program like there was before(OGO and so on).

Don't get me wrong, I'm very interested in your discussion, so please feel free to continue.

Sorry, as I stated elsewhere, I don't have a P4 (just Athlon XP), so I can't test the new version speedwise.

CU,

tiki4

[vlad59]

@Tom

Your SSE2 motion compensation code is still here, somebody told me that mine is faulty so I canceled my update.

Once a member diverges past a certain point it's often simpler just to keep a separate member name. It makes some enhancement harder but some maintenance easier. Then folks without a P4 could optimze K6-II & stuff more simply. And there is starting to be quite a bit of code difference so removing the extra run time checks might even help performance a bit.

I totally agree with you. That's why I've used some pointer function (to avoid some check) when I optimized the motion compensation code, it helped a lot. I also planned to apply the same trick to the IDCT call (later when I got some spare time).

@tiki4

Yes, Save-OE is a strange project. I personnaly think it is important to talk with Jackei to check his todo list and to know his feelings about Save-oe. Otherwise we'll have too many dvd2avi and it will be full of confusion (is it good english, I don't know ???).

[Nic]

function pointers are a very good idea...(thats the way xvid does it)

im hoping my version of DVD2AVI wont cause too much confusion. It's quite different to what Save-OE & Jackei's DVD2AVI is trying to achieve....I think :)

-Nic

[trbarry]

Yes, Save-OE is a strange project. I personnaly think it is important to talk with Jackei to check his todo list and to know his feelings about Save-oe. Otherwise we'll have too many dvd2avi and it will be full of confusion (is it good english, I don't know ???).

Yes, by all means if anyone knows Jackei or is in correspondence with him see if you can get him to comment here. I got interested in DVD2AVI mostly because it seemed to be an orphan (from both Jackei & Ogo) and I wanted a bit more robustness along with some HDTV & P4 support.

And eventually I'd like to be able to pass along syncronized ac3 streams directly into encoding. But I'm not actively working on that, mostly just lobbying (begging) for it. ;)

- Tom