Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Capturing and Editing Video > Avisynth Development

Reply
 
Thread Tools Search this Thread Display Modes
Old 5th September 2009, 18:44   #21  |  Link
BigDid
Actually in reserve
 
Join Date: Oct 2004
Posts: 1,605
Hi,

Thanks for the update. I have tried it, but the colors problems are forbidding a regular usage.
That could be something to put on top of the stack

I have followed (but note always understood) the "main/single-threaded" 2.6 thread and have seen these exchanges.
I will not put pressure on either side but if I can be of any help for the MT side let me know.

Did
__________________
Having a problem with AutoGK? Read & use the FAQ & MORE FAQ first
Want to exchange on AutoGK? try doom10.org
In reserve (inactive) for an undefined period of time.
BigDid is offline   Reply With Quote
Old 4th October 2009, 15:44   #22  |  Link
levi
Registered User
 
Join Date: Mar 2003
Posts: 116
Hey guys, it seems my problem is related to my quad core CPU as I have exactly the same symptoms as mentioned here: http://forum.doom9.org/showthread.ph...83#post1319483

Switched to several different SetMTmodes, but it eventually stalls (CPU use & fps in vdub drop to zero) It starts out fast though.

So I was going to try the other modes...

Code:
MTSource(""" MPEG2Source("D:\my.d2v") """)
but i get:
Quote:
Avisynth open failure:
Script error: Invalid arguments to function "MTSource"
I checked the syntax a few times but I can't find the error
levi is offline   Reply With Quote
Old 4th October 2009, 16:01   #23  |  Link
Gavino
Avisynth language lover
 
Join Date: Dec 2007
Location: Spain
Posts: 3,431
Quote:
Originally Posted by levi View Post
I checked the syntax a few times but I can't find the error
The syntax looks right. Perhaps it can't find MPEG2Source.
Do you have DGDecode.dll in your plugins folder?
Gavino is offline   Reply With Quote
Old 4th October 2009, 17:30   #24  |  Link
levi
Registered User
 
Join Date: Mar 2003
Posts: 116
Yes, dgdecode is in plugins - it works fine with
Code:
MPEG2Source("D:\my.d2v")
Code:
 MTSource(""" AviSource("pass2.avi") """)
gives the same error, but works fine when used normally

MT 0.7 (includes modified avisynth 2.5.7.5) works fine

:<

Last edited by levi; 4th October 2009 at 19:15. Reason: tested older version
levi is offline   Reply With Quote
Old 5th October 2009, 12:24   #25  |  Link
SEt
Registered User
 
Join Date: Aug 2007
Posts: 374
There actually was one more threading fix i haven't bothered to post due to expected rewrite of cache/MTMode in 2.6 that should solve the problem with hanging. See the first post for the rebuilt dll.

Code:
--- cacheMT.cpp	2009-08-08 22:03:34.034750000 +0400
+++ cacheMT.cpp	2009-10-05 15:15:03.734375000 +0400
@@ -430,7 +430,9 @@
 		  InterlockedDecrement(&e_generated_refcount[i->e_generated_index]);	
           ++g_Cache_stats.vfb_found;
           LockVFB(i, env);  //  BuildVideoFrame expect the VFB to be locked
+		  EnterCriticalSection(&cs_cache);
           PVideoFrame retval=BuildVideoFrame(i, n);
+		  LeaveCriticalSection(&cs_cache);
           InterlockedDecrement((long*)&retval->refcount);
           return retval;
             }
SEt is offline   Reply With Quote
Old 9th October 2009, 18:41   #26  |  Link
a451guy451
Xbox Live: o 4lif o
 
a451guy451's Avatar
 
Join Date: Jun 2009
Location: Monrovia, CA
Posts: 64
Thanks for posting the new build. I'm looking forward to playing with it on a 4 and 8 core machine today...
a451guy451 is offline   Reply With Quote
Old 9th October 2009, 20:11   #27  |  Link
levi
Registered User
 
Join Date: Mar 2003
Posts: 116
I'm curious to hear your results. I did not see any improvements in the hangs / crashes.
levi is offline   Reply With Quote
Old 9th October 2009, 21:38   #28  |  Link
a451guy451
Xbox Live: o 4lif o
 
a451guy451's Avatar
 
Join Date: Jun 2009
Location: Monrovia, CA
Posts: 64
I'm still running my first test on my 8-core. It's a really complicated script that usually gives me 3-4fps, and it's running 12-14fps with this new DLL. It's been rock solid for the last 2h 30min, and I have another hour to go. BUT, when I was using the previous build it would hang itself in less than 5 minutes. I expect this will complete just fine. After I finish the 8-core test, I'll hop onto my quad core box and see if I can duplicate the results (I see no reason why it would give me any problems).

Sorry you're not seeing any improvement though. I'm only using Setmtmode(2,0), so perhaps you should give that a try if you haven't already.

VERY promising though. Thanks again SEt!
a451guy451 is offline   Reply With Quote
Old 9th October 2009, 22:47   #29  |  Link
a451guy451
Xbox Live: o 4lif o
 
a451guy451's Avatar
 
Join Date: Jun 2009
Location: Monrovia, CA
Posts: 64
8-core test complete, with no issues (yay). I'll try to post the results of my test on a quad core tomorrow.
a451guy451 is offline   Reply With Quote
Old 10th October 2009, 01:51   #30  |  Link
7ekno
Guest
 
Posts: n/a
Quote:
Originally Posted by a451guy451 View Post
8-core test complete, with no issues (yay). I'll try to post the results of my test on a quad core tomorrow.
Was that with the link called "Version 2009.09.19" ?

If that's the case, I will retry it later too ...

7ek
  Reply With Quote
Old 10th October 2009, 15:33   #31  |  Link
a451guy451
Xbox Live: o 4lif o
 
a451guy451's Avatar
 
Join Date: Jun 2009
Location: Monrovia, CA
Posts: 64
Yes it was.
a451guy451 is offline   Reply With Quote
Old 11th October 2009, 16:47   #32  |  Link
a451guy451
Xbox Live: o 4lif o
 
a451guy451's Avatar
 
Join Date: Jun 2009
Location: Monrovia, CA
Posts: 64
Quad core works as well. Really, almost as fast as the 8-core, but I think the explanation for that is that I'm hitting a bottleneck elsewhere in my system. Cool deal either way.
a451guy451 is offline   Reply With Quote
Old 20th October 2009, 21:15   #33  |  Link
levi
Registered User
 
Join Date: Mar 2003
Posts: 116
set, does it matter which version of devil.dll is used?
levi is offline   Reply With Quote
Old 20th October 2009, 22:04   #34  |  Link
BigDid
Actually in reserve
 
Join Date: Oct 2004
Posts: 1,605
Quote:
Originally Posted by SEt View Post
Color format conversions are currently broken (at least their dynamically compiled part). Also note that i haven't rebuilt the dll with posted bugfixes yet.
Hi,

The last 2.6-MT from SET is really stable for me, but I am still getting some horizontal colors bands; don't know if related to the above quote though?

I could try to pinpoint what script/filter does trigger these bands if interested...

Did
__________________
Having a problem with AutoGK? Read & use the FAQ & MORE FAQ first
Want to exchange on AutoGK? try doom10.org
In reserve (inactive) for an undefined period of time.
BigDid is offline   Reply With Quote
Old 21st October 2009, 11:44   #35  |  Link
SEt
Registered User
 
Join Date: Aug 2007
Posts: 374
I can't say anything about devil.dll. And of course if you want something fixed you need to post a way how to reproduce the bug.
SEt is offline   Reply With Quote
Old 25th October 2009, 21:46   #36  |  Link
BigDid
Actually in reserve
 
Join Date: Oct 2004
Posts: 1,605
Quote:
Originally Posted by SEt View Post
... And of course if you want something fixed you need to post a way how to reproduce the bug.
Hi,

Here are the infos I can give you:

Having a mpegsouce or a directshowsource,

Source being this test movie:
http://www.mediafire.com/file/r2rnvnnn2y1/bride1.m2v
from this thread:
http://forum.doom9.org/showthread.ph...15#post1328915

autocrop() gives some colors problems (patchs of blue and red plus a thin horizontal green line at the bottom) with a resulting resolution of 700*384

Crop(movie,0,46,0,-50) is OK

mt("DeGrainMedian(mode=5)") is Ok

lanczosresize(720,432) is ok (as is autocrop +lanczos)

BlindPP(cpu2="xxxxoo", quant=16) -deblocking- or
BlindPP(cpu2="ooooxx", quant=5) -deringing- gives a green horizontal thick line at the bottom quite similar to the one in the 2.6 thread:
http://forum.doom9.org/showthread.ph...38#post1337438
Blindpp() being a separate use of the cpu function from dgdecode, this could be expected.

Complex scripts sharpening and/or denoising with or whitout MT are Ok:
lsfmod()or
*mt("""lsfmod(strength=150, soothe=true,keep=10,edgemode=1)""")
MCtemporaldenoise() or
*mt("""MCTemporalDenoise(settings="medium",sigma=20,sharp=true,strength=200,protect=true,flat=true)""")
Softsharpen() or
*mt("""softsharpen(preset="film", net=14,debrT=1,debr=0, flou=16,derin=0)""",2,4,splitvertical=true)

* are examples only; parameters may not be appropriate or only if associated with other filters.

So I hope that the solution(s) from the main 2.6 thread can apply here. If not, someone (not me) will have to ask an update of dgdecode().
__________________
Having a problem with AutoGK? Read & use the FAQ & MORE FAQ first
Want to exchange on AutoGK? try doom10.org
In reserve (inactive) for an undefined period of time.

Last edited by BigDid; 26th October 2009 at 17:26.
BigDid is offline   Reply With Quote
Old 25th October 2009, 22:24   #37  |  Link
IanB
Avisynth Developer
 
Join Date: Jan 2003
Location: Melbourne, Australia
Posts: 3,167
@BigDid,

It ain'i rocket science put a SetPlanarLegacyAlignment (True) after the BlindPP() line and see if it hides the problem.

(Hint: Having looked at the code, just halving the pitch, as is done here, is going to cause this problem.)
IanB is offline   Reply With Quote
Old 26th October 2009, 03:09   #38  |  Link
levi
Registered User
 
Join Date: Mar 2003
Posts: 116
BigDid, how are you doing MT("...") syntax with 2.6? where did you get 2.6 MT.DLL? I only saw avisynth.dll in those 7zips set posted.
levi is offline   Reply With Quote
Old 26th October 2009, 17:27   #39  |  Link
BigDid
Actually in reserve
 
Join Date: Oct 2004
Posts: 1,605
Quote:
Originally Posted by IanB View Post
@BigDid,

It ain'i rocket science put a SetPlanarLegacyAlignment (True) after the BlindPP() line and see if it hides the problem...
Hi IanB,

Works fine. As I use the deblock part of blind PP before denoising/sharpening and the dering part after, I have tested that 1 statement of "SetPlanarLegacyAlignment (True)" at the end of the script will work fine also.
Thanks alot, good to have some constructive answers in these difficult times.

Quote:
Originally Posted by levi View Post
BigDid, how are you doing MT("...") syntax with 2.6?
Same as you are. Some syntax works some do not.
Basics are on the wiki page:
http://avisynth.org/mediawiki/MT_support_page
summarizing the basic MT 0.7 thread:
http://forum.doom9.org/showthread.php?t=94996

You have to juggle with triple quotes, overlap and/or splitvertical. I have updated the complex scripts in the above post as examples.

Quote:
where did you get 2.6 MT.DLL? I only saw avisynth.dll in those 7zips set posted.
Same as yours:
http://forum.doom9.org/showthread.ph...10#post1331510
http://www.avisynth.org/tsp/MT_07.zip
__________________
Having a problem with AutoGK? Read & use the FAQ & MORE FAQ first
Want to exchange on AutoGK? try doom10.org
In reserve (inactive) for an undefined period of time.
BigDid is offline   Reply With Quote
Old 26th October 2009, 22:06   #40  |  Link
IanB
Avisynth Developer
 
Join Date: Jan 2003
Location: Melbourne, Australia
Posts: 3,167
Quote:
Originally Posted by BigDid View Post
...
Works fine. As I use the deblock part of blind PP before denoising/sharpening and the dering part after, I have tested that 1 statement of "SetPlanarLegacyAlignment (True)" at the end of the script will work fine also.
It is not really fine to put SetPlanarLegacyAlignment(True) at the end of the script. Doing so may force any SSE2 code to use the slower unaligned memory access code path for YV12 chroma planes or poorly written filters may even crash. It also only works for YV12.

For filters that make illegal assumptions about pitch you should put SetPlanarLegacyAlignment(False) right before the problem filters statement and SetPlanarLegacyAlignment(True) right after it. i.e.
Code:
Mpeg2Source("fred.d2v, cpu=1) # Bad planar code!
SetPlanarLegacyAlignment(True) # Do slow legacy alignment
... 
SetPlanarLegacyAlignment(False) # Restore fast alignment
BlindPP(cpu2="xxxxoo", quant=16) # Bad planar code!
SetPlanarLegacyAlignment(True) # Do slow legacy alignment
...
IanB is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 10:09.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.