FFmpeg 2.0 (released)

	Bottom Previous Topic Next Topic
Register To Post

MickJT

Posted on: 2013/7/20 16:47 #1

Just can't stay away

Now released

Hello,

I'd like some people to test my builds of FFmpeg 2.0 before I put them on OS4Depot. These may not end up being the final builds, but will be close to it.

There's nothing specific to test, except to try a few different codecs and make sure it works OK.

Please state whether you are using the generic or altivec version.

You can grab it from: OS4Depot

Edited by MickJT on 2013/7/20 17:12:10
Edited by MickJT on 2013/7/21 7:24:53
Edited by MickJT on 2013/7/21 7:40:29
Edited by MickJT on 2013/7/21 13:55:05
Edited by MickJT on 2013/7/21 17:40:20
Edited by MickJT on 2013/7/22 13:20:29
Edited by MickJT on 2013/7/22 14:16:15
Edited by MickJT on 2013/7/24 10:48:02
Edited by MickJT on 2013/7/25 6:34:00

samo79

Re: FFmpeg 2.0 testing

Posted on: 2013/7/20 22:10 #2

Home away from home

I just converted a small MP4 video file to FLV and all goes well, it seems ok

samo79

Re: FFmpeg 2.0 testing

Posted on: 2013/7/20 22:51 #3

Home away from home

I had a couple of problems with FFPlay, sometimes it tend to crash at exit

Quote:

Crash log for task "ffplay"
Generated by GrimReaper 53.5
Crash occured in module newlib.library.kmod at address 0x01A48140
Type of crash: DSI (Data Storage Interrupt) exception

Register dump:
GPR (General Purpose Registers):
0: 5FFAA000 52395370 00000000 FFD05585 FFD05585 00000003 00000073 58D4C03C
8: 00000001 002FAA7B 00000091 01A4812C 00000000 00000000 00000020 0000FFFF
16: 00000007 00000000 53500570 5847155E 0000FFFF 52395378 52395568 58D7F9A0
24: 535004B8 00000012 5FF9A000 0000001C 00000000 00000000 00000009 FFD05585

FPR (Floating Point Registers, NaN = Not a Number):
0: nan 72 72 0
4: 0 663.8 24.8 0
8: 0 4.5036e+15 663.8 24.8
12: 736 0 0 -0
16: 0 0 0 0
20: 0 0 0 1.61895e-319
24: 0 0 0 -5.44175e+303
28: 0 1.35808e-312 8.6917e-311 -0.015625

FPSCR (Floating Point Status and Control Register): 0x82004000

SPRs (Special Purpose Registers):
Machine State (msr) : 0x0002F030
Condition (cr) : 0x48224048
Instruction Pointer (ip) : 0x01A48140
Xtended Exception (xer) : 0x00000002
Count (ctr) : 0x00000003
Link (lr) : 0x6FFF8388
DSI Status (dsisr) : 0x00000000
Data Address (dar) : 0xFFD05585

680x0 emulated registers:
DATA: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
ADDR: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
FPU0: 0 0 0 0
FPU4: 0 0 0 0

Symbol info:
Instruction pointer 0x01A48140 belongs to module "newlib.library.kmod" (HUNK/Kickstart)

Stack trace:
native kernel module newlib.library.kmod+0x00003800
module LIBS:locale.library at 0x6FFFB71C (section 5 @ 0x9704)
module CLASSES:requester.class at 0x6FB24B9C (section 5 @ 0x1B84)
module CLASSES:requester.class at 0x6FB25A80 (section 5 @ 0x2A68)
native kernel module intuition.library.kmod+0x000181fc
native kernel module intuition.library.kmod+0x00018420
native kernel module intuition.library.kmod+0x00008448
module IPrefs at 0x6FF37318 (section 5 @ 0xF300)
native kernel module dos.library.kmod+0x00030d6c
native kernel module shell.kmod+0x00019aec
native kernel module shell.kmod+0x000081d4
native kernel module shell.kmod+0x00008e40
native kernel module dos.library.kmod+0x00022d9c
native kernel module kernel+0x0003bd20
native kernel module kernel+0x0003bda0

PPC disassembly:
01a48138: 4182001c beq- 0x1A48154
01a4813c: 7ca903a6 mtctr r5
*01a48140: 7c8c18ae lbzx r4,r12,r3
01a48144: 398c0001 addi r12,r12,1
01a48148: 2c040000 cmpwi r4,0

System information:

CPU
Model: AMCC PPC440EP V1.3
CPU speed: 799 MHz
FSB speed: 133 MHz
Extensions:

Machine
Machine name: Sam440EP
Memory: 1048576 KB
Extensions: bus.pci

- Another strange problem i noted is that; i have some anime videos that is correctly play in my language (italian) if i play them on MPlayer or DvPlayer, however if i try to play them with FFPlay the audio turn in japanese !

WTF i never though that ! How can i set the correct language ??

MickJT

Re: FFmpeg 2.0 testing

Posted on: 2013/7/21 7:08 #4

Just can't stay away

@samo79

You can select the 2nd audio stream in ffplay by putting "-ast 1" in the command line before the video filename. 0 is the first audio stream.

You didn't mention whether you were using the generic or altivec version. The stack trace only shows native kernel modules and libraries, so that doesn't help me there.

"Crash occured in module newlib.library.kmod at address 0x01A48140"

Hmm.

Check the first post again for debug builds. Can you try setting a higher stack? It's already set for 1000000 which should be more than enough, but try with something that's overkill. If it doesn't crash with a stack size of 5000000, then lower it until it does. The built-in stack cookie won't override your stack setting if yours is higher.

Do you use IRC at all?

Edited by MickJT on 2013/7/21 7:26:21
Edited by MickJT on 2013/7/21 7:28:09
Edited by MickJT on 2013/7/21 7:29:58
Edited by MickJT on 2013/7/21 7:34:28

MickJT

Re: FFmpeg 2.0 testing

Posted on: 2013/7/21 12:30 #5

Just can't stay away

I've had reports of faulty encodes with the altivec version. I don't have an altivec machine here, but I'm messaging them with some things to try. I'm trying to determine if it's the cross-compiler too, or just a bug in the ffmpeg 2.0 altivec code.

Kicko

Re: FFmpeg 2.0 testing

Posted on: 2013/7/21 12:48 #6

Just can't stay away

Great when i oome back to Sweden ill have a bugtested version :)

samo79

Re: FFmpeg 2.0 testing

Posted on: 2013/7/21 13:46 #7

Home away from home

@MickJT

Quote:

You can select the 2nd audio stream in ffplay by putting "-ast 1" in the command line before the video filename. 0 is the first audio stream.

That solve the issue, thanks for the hint

Strange that i didn't found it on documentation

Quote:

You didn't mention whether you were using the generic or altivec version.

As Sam440 user i'm using the generic version only

Quote:

Check the first post again for debug builds. Can you try setting a higher stack? It's already set for 1000000 which should be more than enough, but try with something that's overkill. If it doesn't crash with a stack size of 5000000, then lower it until it does. The built-in stack cookie won't override your stack setting if yours is higher.

Ok i will check

Quote:

Do you use IRC at all?

You mean the chat ?
If you mean that, yes sometimes but more than IRC i'm often using MSN (SabreMSN)

MickJT

Re: FFmpeg 2.0 testing

Posted on: 2013/7/21 13:58 #8

Just can't stay away

It looks like there might be an issue with the cross-compile when enabling altivec instructions. It takes a lot lot longer to build it natively, so I might do a native build tonight when I go to sleep.

Edit: Native build is available to try.
Edit2: Cross-compiled gcc 4.5.4 build available to test.

Edited by MickJT on 2013/7/21 14:13:02
Edited by MickJT on 2013/7/21 17:53:04
Edited by MickJT on 2013/7/22 14:19:14

Thematic

Re: FFmpeg 2.0 testing

Posted on: 2013/7/22 15:05 #9

Quite a regular

Quote:

I just converted a small MP4 video file to FLV

Only muxing? That won't necessarily apply the codecs.

I'd try this if my system wasn't off Internet (probably bad line or router).

Raziel

Re: FFmpeg 2.0 testing

Posted on: 2013/7/22 15:44 #10

Home away from home

Altivec-cross-4.5.4 tested on X1000

The worst version upto now i'm afraid

Not even ffplay plays the test video with proper sound (glitched)
ffmpeg glitches the sound in all test cases (.mp3, .aac, .avi and .mp4) 1:1 to what it sounds through ffplay

--

game box/art scans
scummvm infrequent builds

MickJT

Re: FFmpeg 2.0 testing

Posted on: 2013/7/22 15:52 #11

Just can't stay away

Ahh damn. Well, I could release the working versions I have, but there's stuff I want to test :)

K-L

Re: FFmpeg 2.0 testing

Posted on: 2013/7/23 8:39 #12

Just can't stay away

MickJT : sorry, I forgot to answer.

Everything works OK with my AmigaOne G4 and the native builds (I could play some videos with ffplay and convert some others with ffmpeg).

--
AmigaONE X1000 and Radeon RX 560

Vulture

Re: FFmpeg 2.0 testing

Posted on: 2013/7/24 9:51 #13

Not too shy to talk

@MickJT

tried native altivec ffplay on my peg2 with a few clips and all was fine! :)

MickJT

Re: FFmpeg 2.0 (released)

Posted on: 2013/7/24 10:45 #14

Just can't stay away

Thanks. It'll be released ~~in 1 or 2 days~~ soon. The final builds will be re-compiled using zerohero's gcc 4.2.4 cross-compiler, which produces altivec binaries that work fine.

Edit: Released.

Edited by MickJT on 2013/7/24 20:52:48
Edited by MickJT on 2013/7/25 6:34:33

Register To Post
	Top Previous Topic Next Topic

Currently Active Users Viewing This Thread: 2 ( 0 members and 2 Anonymous Users )