.Other Versions of the Mobsync.dll FileThe latest version of the Mobsync.dll file is 5.2.3790.3959 version. Before this, there were 3 versions released. All versions released up till now are listed below from newest to oldest. 5.2.3790.3959 - 32 Bit (x86). 5.1.2600.5512 - 32 Bit (x86). 5.1.2600.2180 - 32 Bit (x86) ( 2012-06-30). 5.1.2600.2180 - 32 Bit (x86)How to Download Mobsync.dll File?.
![]()
Mpeg2enc.dll download. The Mpeg2enc.dll file is a dynamic link library for Windows 10, 8.1, 8, 7, Vista and XP. You can fix 'The file Mpeg2enc.dll is missing.'
First, click on the green-colored ' Download' button in the top left section of this page (The button that is marked in the picture). Step 1:Start downloading the Mobsync.dll file. When you click the ' Download' button, the ' Downloading' window will open. Don't close this window until the download process begins. The download process will begin in a few seconds based on your Internet speed and computer.Methods to Solve the Mobsync.dll Errors ATTENTION! Before starting the installation, the Mobsync.dll file needs to be downloaded. If you have not downloaded it, download the file before continuing with the installation steps.
If you don't know how to download it, you can immediately browse the dll download guide above. Method 1: Copying the Mobsync.dll File to the Windows System Folder. The file you downloaded is a compressed file with the '.zip' extension. In order to install it, first, double-click the '.zip' file and open the file.
You will see the file named ' Mobsync.dll' in the window that opens up. This is the file you need to install. Drag this file to the desktop with your mouse's left button.
Step 1:Extracting the Mobsync.dll file. Copy the ' Mobsync.dll' file and paste it into the ' C:WindowsSystem32' folder. Step 2:Copying the Mobsync.dll file into the Windows/System32 folder. If you are using a 64 Bit operating system, copy the ' Mobsync.dll' file and paste it into the ' C:WindowssysWOW64' as well.
On Windows operating systems with 64 Bit architecture, the dll file must be in both the ' sysWOW64' folder as well as the ' System32' folder. In other words, you must copy the ' Mobsync.dll' file into both folders.
Step 3:Pasting the Mobsync.dll file into the Windows/sysWOW64 folder. First, we must run the Windows Command Prompt as an administrator. We ran the Command Prompt on Windows 10. If you are using Windows 8.1, Windows 8, Windows 7, Windows Vista or Windows XP, you can use the same methods to run the Command Prompt as an administrator.
Open the Start Menu and type in ' cmd', but don't press Enter. Doing this, you will have run a search of your computer through the Start Menu.
In other words, typing in ' cmd' we did a search for the Command Prompt. When you see the ' Command Prompt' option among the search results, push the ' CTRL' + ' SHIFT' + 'ENTER' keys on your keyboard. A verification window will pop up asking, ' Do you want to run the Command Prompt as with administrative permission?' Approve this action by saying, ' Yes'.Step 4:Running the Command Prompt with administrative permission. Let's copy the command below and paste it in the Command Line that comes up, then let's press Enter. This command deletes the Mobsync.dll file's problematic registry in the Windows Registry Editor (The file that we copied to the System32 folder does not perform any action with the file, it just deletes the registry in the Windows Registry Editor. The file that we pasted into the System32 folder will not be damaged).%windir%System32regsvr32.exe /u Mobsync.dll Step 5:Deleting the Mobsync.dll file's problematic registry in the Windows Registry Editor.
If the Windows you use has 64 Bit architecture, after running the command above, you must run the command below. This command will clean the broken registry of the Mobsync.dll file from the 64 Bit architecture (The Cleaning process is only with registries in the Windows Registry Editor. In other words, the dll file that we pasted into the SysWoW64 folder will stay as it is).%windir%SysWoW64regsvr32.exe /u Mobsync.dll Step 6:Uninstalling the broken registry of the Mobsync.dll file from the Windows Registry Editor (for 64 Bit).
You must create a new registry for the dll file that you deleted from the registry editor. In order to do this, copy the command below and paste it into the Command Line and hit Enter.%windir%System32regsvr32.exe /i Mobsync.dll Step 7:Creating a new registry for the Mobsync.dll file. If the Windows version you use has 64 Bit architecture, after running the command above, you must run the command below.
With this command, you will create a clean registry for the problematic registry of the Mobsync.dll file that we deleted.%windir%SysWoW64regsvr32.exe /i Mobsync.dll Step 8:Creating a clean registry for the Mobsync.dll file (for 64 Bit). If you did all the processes correctly, the missing dll file will have been installed. You may have made some mistakes when running the Command Line processes. Generally, these errors will not prevent the Mobsync.dll file from being installed. In other words, the installation will be completed, but it may give an error due to some incompatibility issues. You can try running the program that was giving you this dll file error after restarting your computer. If you are still getting the dll file error when running the program, please try the 2nd method.
Method 2: Copying The Mobsync.dll File Into The Software File Folder. In order to install the dll file, you need to find the file folder for the software that was giving you errors such as ' Mobsync.dll is missing', ' Mobsync.dll not found' or similar error messages. In order to do that, Right-click the software's shortcut and click the Properties item in the right-click menu that appears.
Step 1:Opening the software shortcut properties window. Click on the Open File Location button that is found in the Properties window that opens up and choose the folder where the application is installed. Step 2:Opening the file folder of the software. Copy the Mobsync.dll file into the folder we opened. Step 3:Copying the Mobsync.dll file into the file folder of the software. The installation is complete.
Run the software that is giving you the error. If the error is continuing, you may benefit from trying the 3rd Method as an alternative. Method 3: Doing a Clean Install of the software That Is Giving the Mobsync.dll Error. Open the Run window by pressing the ' Windows' + ' R' keys on your keyboard at the same time.
Type in the command below into the Run window and push Enter to run it. This command will open the ' Programs and Features' window. Appwiz.cpl Step 1:Opening the Programs and Features window using the appwiz.cpl command. The Programs and Features screen will come up. You can see all the softwares installed on your computer in the list on this screen. Find the software giving you the dll error in the list and right-click it. Click the ' Uninstall' item in the right-click menu that appears and begin the uninstall process.
Step 2:Starting the uninstall process for the software that is giving the error. A window will open up asking whether to confirm or deny the uninstall process for the software. Confirm the process and wait for the uninstall process to finish. Restart your computer after the software has been uninstalled from your computer. Step 3:Confirming the removal of the software. After restarting your computer, reinstall the software that was giving the error. This method may provide the solution to the dll error you're experiencing.
If the dll error is continuing, the problem is most likely deriving from the Windows operating system. In order to fix dll errors deriving from the Windows operating system, complete the 4th Method and the 5th Method. Method 4: Solving the Mobsync.dll Error using the Windows System File Checker (sfc /scannow). First, we must run the Windows Command Prompt as an administrator. We ran the Command Prompt on Windows 10.
If you are using Windows 8.1, Windows 8, Windows 7, Windows Vista or Windows XP, you can use the same methods to run the Command Prompt as an administrator. Open the Start Menu and type in ' cmd', but don't press Enter. Doing this, you will have run a search of your computer through the Start Menu. In other words, typing in ' cmd' we did a search for the Command Prompt. When you see the ' Command Prompt' option among the search results, push the ' CTRL' + ' SHIFT' + 'ENTER' keys on your keyboard. A verification window will pop up asking, ' Do you want to run the Command Prompt as with administrative permission?' Approve this action by saying, ' Yes'.Step 1:Running the Command Prompt with administrative permission.
After typing the command below into the Command Line, push Enter. Sfc /scannow Step 2:Getting rid of dll errors using Windows's sfc /scannow command. Depending on your computer's performance and the amount of errors on your system, this process can take some time. You can see the progress on the Command Line. Wait for this process to end.
After the scan and repair processes are finished, try running the software giving you errors again. Method 5: Fixing the Mobsync.dll Error by Manually Updating WindowsSome softwares need updated dll files.
When your operating system is not updated, it cannot fulfill this need. In some situations, updating your operating system can solve the dll errors you are experiencing.In order to check the update status of your operating system and, if available, to install the latest update packs, we need to begin this process manually.Depending on which Windows version you use, manual update processes are different. Because of this, we have prepared a special article for each Windows version. You can get our articles relating to the manual update of the Windows version you use from the links below. Windows Update Guides.The Most Seen Mobsync.dll ErrorsThe Mobsync.dll file being damaged or for any reason being deleted can cause softwares or Windows system tools (Windows Media Player, Paint, etc.) that use this file to produce an error. Below you can find a list of errors that can be received when the Mobsync.dll file is missing.If you have come across one of these errors, you can download the Mobsync.dll file by clicking on the ' Download' button on the top-left of this page. We explained to you how to use the file you'll download in the above sections of this writing.
You can see the suggestions we gave on how to solve your problem by scrolling up on the page. ' Mobsync.dll not found.' Error.
' The file Mobsync.dll is missing.' Error. ' Mobsync.dll access violation.' Error. ' Cannot register Mobsync.dll.'
Error. ' Cannot find Mobsync.dll.' Error. ' This application failed to start because Mobsync.dll was not found. Re-installing the application may fix this problem.' ErrorOther Dll Files Used with Mobsync.dll.
Hi Steven (& Andrew),On Fri, 2003-10-31 at 17:03, Steven M. Schultz wrote: I really think a better method than relying on non-portable behaviour needs to be found.pkg-config solves this. Make every package 'depend' on other packagesand - when calling -cflags on the given package, also give the -cflagsfor all libs it depends on (recursively). Obviously, pkg-config is thefreedesktop.org (and thus GNOME) standard way of package dependencytracking.Anyway, some of this stuff is going over my head (I should have studiedCS if I wanted to understand all this, I guess).Oh, btw, the libtool/C issue is supposedly solved by upgrading fromlibtool = 1.5.
I've had several C linking issueswith libtool 1.4, and 1.5 solves them.Ronald-Ronald Bultje Linux Video/Multimedia developer. On Fri, 31 Oct 2003, Andrew Stevens wrote: This turns out to be a real Rat's nest of portability issues. Basically, at least some BSD mandate PIC (position indepdent code) for shared libs. libtool on such platforms is quite picky and just records any static libsIt's apparently not a 'cast in stone' requirement but workingaround it does mean manually editing the generated libtoolscript.The problem I see (at least in the future) is that there probablyis a platform on which linux runs that does have a (hardwareenforced) rule against nonshared code going into a shared segment.
a executable is linked against the.la. Also RH9 (for some reason) also appears to behave in this way. Under OS/X something similar appears toI may have simply been dazed momentarily by all the permutationsbeing tried;) Adding libmjpegutils to the list appears to haveresolved the issues of undefined symbols on RH9 - sorry for theconfusion. happen.and. a minor libtool C issue makes life even more complicated. mentioned when linking a.so into the.la and links these statically wheneverThe OS/X C vs libtool issue is one that is real but i know howto deal with - it's the same issue I have to workaround when./bootstrap'ing the smilutils on OS/X.
Not a big deal oncethe underlying problem was discovered (but the journey leading upto that was, uh, interesting;)). To quote Steven: 'AAAAARRRGHH' For now I intend to work around the problem by making everything static and building the gstreamer plugin as shared lib from the static stuff.The most general/portable method would be to defer the linkingof the static/non-pic libraries (some of which have or will havehand written assembly code which is definitely NOT PIC aware)until the final application is linked. This, alas, does havethe requirement that libmjpegutils, libmotion and perhapslibmmxsse.a be present on the system. Then a -config script(similar to mjpegtools-config - perhaps it could be calledmpeg2enc-config) would be used with 'mpeg2enc -libs' to dothe right thing for a given platform - it would refer to theshared libmpeg2encpp, and then whatever static libs were neededto resolve symbols).
That's the approach SDL, curl, fltk,libfame, libold2, libvideogfx, libquicktime, movtar, mpeg4ip,and others use. No one's revolted against those yet;-)Cheers,Steven Schultz. On Fri, 31 Oct 2003, Ronald Bultje wrote: back out some of the earlier deplib changes, add mjpegutils in one spot. NOTE: forcing static (.a) code into a shared object (.so or.dylib) is NOT a good idea and libtool will validly complain about doing so! Why not? I know the warning, but afaik, the warning is void.
It doesn't mean a single real thing. Is there anything important to know about it?It's a warning for a reason.Basically at that point your forcing non-PIC compiled codeinto a shared object where the rest of the code is PICcompiled. Hi Ronald, On Thu, 2003-10-30 at 23:46, Steven Schultz wrote: Modified Files: Makefile.am Log Message: back out some of the earlier deplib changes, add mjpegutils in one spot. NOTE: forcing static (.a) code into a shared object (.so or.dylib) is NOT a good idea and libtool will validly complain about doing so! Why not? I know the warning, but afaik, the warning is void.
It doesn't mean a single real thing. Is there anything important to know about it?This turns out to be a real Rat's nest of portability issues.
Basically, atleast some BSD mandate PIC (position indepdent code) for shared libs. So thelibtool on such platforms is quite picky and just records any static libsmentioned when linking a.so into the.la and links these statically whenevera executable is linked against the.la. Also RH9 (for some reason) alsoappears to behave in this way. Under OS/X something similar appears tohappen.and. a minor libtool C issue makes life even more complicated.To quote Steven: 'AAAAARRRGHH'For now I intend to work around the problem by making everything static andbuilding the gstreamer plugin as shared lib from the static stuff.Andrew Ronald.
Hi,On Thu, 2003-10-30 at 23:46, Steven Schultz wrote: Modified Files: Makefile.am Log Message: back out some of the earlier deplib changes, add mjpegutils in one spot. NOTE: forcing static (.a) code into a shared object (.so or.dylib) is NOT a good idea and libtool will validly complain about doing so!Why not?
I know the warning, but afaik, the warning is void. It doesn'tmean a single real thing. Is there anything important to know about it?Ronald-Ronald Bultje Linux Video/Multimedia developer.
Hi STefan, PS: after that I should rework the upscaling-code, as sinc isn't really a good idea there but all the other methods I 'invented' don't work reliably enough to activate them.Just out of interest, what scaling filters.are. you implementing?The rule of thumb amoungst Digital TV pro's I've talked to is that you reallywant at least a 7-tap polyphase filter to do a decent job horizontallyscaling TV and at least 5-taps vertically. Plus of course you want to ensurea decent kernel to suit your preferred aliasing/passband balance, andeliminate any sampling frequency ('DC') ripple. Remember, bicubic kernelsand similar are more or less hand-hacked solutions to the problem of scalingGraphics (where you're not band-limited ).For a TV signal (which is band-limited, though less so once JPEG etc have donetheir dirty deeds) classical reconstruction / pre-filter kernels are the wayto go. By all accounts a Lanczos kernel (sinc windowed sinc) or similar'Textbook' kernels work well. Especially if you intend to use some kind ofDCT based coding afterwards where significant aliasing will hurt yourcompression.Scaling for.display. (rather than for encoding) is a different issue again.There all kinds of non-linear high-pass tweaks post-reconstruction to boosttransients make things look visibly better.Andrew.
Matto:),as the deinterlacer-core works nice, now, I'll have to clean up the codeand implement autodetection(x-flags) of the different y4m-interlacingmodes currently in use:YCrCb 4:2:0 only luma interlaced (I recall this to be our 'standard'version, with chroma plane of the first field???)YCrCb 4:2:0 PAL chroma and luma interlaced (my DV-camcorder + mplayer-vo yuv4mpeg )YCrCb 4:1:1 NTSC chroma and luma interlaced (Can anyone send me a 1-2 MBfile NTSC-DV, to see what mplayer does do with it?! I guessit'll be doing something like 4:1:1 to 4:2:0i conversion?!)Any missing? The first two are 'weird' as my videograbber does do thefirst variant and the DV-camcorder does the second.
I should implementcommandline options to override the x-flags, should I?StefanPS: after that I should rework the upscaling-code, as sinc isn't reallya good idea there but all the other methods I 'invented' don't workreliably enough to activate them. Hi James,I just took a closer look at the changes needed and it turns out to be goodnews. By sheer good fortune it would seem there is no need to carry anysetting or input stream dependent parameters into the low-leve routinesother than the quantizer.Soooo. Nastiness requiring maintain a per-instantiation workspace is notrequired. From the perspective of the interface to the SIMD routines thingsare stable. All that will happen is that the call's to their 'init' routineswill be seperated out from the stuff that has to happen for each newinvocation of the encoder.The only other thing I do want to do is to do a sensible run-time detection ofthe correct alignment for SIMD data. Currently it is just a hard-wired 16!I'll put a small routine in utils that detects CPU type and comes up with asensible suggestion.
For Athlon's 64-byte alignment is good.I'll set it up for x86 with a default to the current value which should allowyou to cleanly hook up PowerPC later.Andrew. On Sun, 19 Oct 2003, Ben Mackin wrote: On 10/19/03 6:48 PM, 'Steven M. Schultz' wrote: I'm not sure how many folks other than James and I are running mpeg2enc on a Altivec enabled machine. I think though we're I run it as well (I am here mostly to learn).
As long as it isn't scrapped, whatever is easiest should be done.Ah, ok - glad to 'meet' you.Oh, it's not being scrapped by any stretch of the imagination.The altivec portions may be unbuildable periodically as theencoder is restructured. Might be a few days until James getsre-sync'd with the changes - that's all.Cheers,Steven Schultz. On Sun, 19 Oct 2003, James Klicman wrote: I just recently updated the AltiVec code for compatibility with the previous changes. I'd hate to see CVS-HEAD broken again for AltiVec after all of those updates. What do you think about creating a branch for the refactoring? That will give me a chance to synchronize with your refactoring without breaking CVS-HEAD.I don't know about others but I'm somewhat allergic tobranches in CVS.
Used to use them at work and it wasa constant source of trouble - using the wrong tag (not all thathard to do unfortunately) at the wrong moment creates a mess thatcan be difficult to clean up when the eventual merge is done.I'm not sure how many folks other than James and I are runningmpeg2enc on a Altivec enabled machine. I think though we'regreatly outnumbered by the IA32 crowd:-)Personally I could live with short lived breakage in the Altivecarea. It'd get fixed fairly quickly (right, James?:-)) and I'mnot sure it's worth the (to me) greater inconvenience of dealingwith CVS branches.Steven Schultz. Hi Andrew,I just recently updated the AltiVec code for compatibility with theprevious changes. I'd hate to see CVS-HEAD broken again for AltiVecafter all of those updates.What do you think about creating a branch for the refactoring?
That willgive me a chance to synchronize with your refactoring without breakingCVS-HEAD.-JamesOn Sun, Oct 19, 2003 at 11:43:58PM +0200, Andrew Stevens wrote: Hi all, Just to let any PowerPC users out there know I'm currently in the middle of the refactoring needed to make mpeg2enc a nicely behaved re-entract library routine that can be hooked into gstreamer (and the like) cleanly. A big side effect of this is that the routines used to set up CPU specific low-level calls are changed. Currently on the the x86 stuff compiles. James: you probably want to hold off updating for a while until I've completed the modification to the x86 SIMD stuff. Its in a bit of a transitional state at present.
![]()
Andrew. Hi all,Just to let any PowerPC users out there know I'm currently in the middle ofthe refactoring needed to make mpeg2enc a nicely behaved re-entract libraryroutine that can be hooked into gstreamer (and the like) cleanly.A big side effect of this is that the routines used to set up CPU specificlow-level calls are changed. Currently on the the x86 stuff compiles.James: you probably want to hold off updating for a while until I'vecompleted the modification to the x86 SIMD stuff. Its in a bit of atransitional state at present.Andrew.
version of mplex makes mplayer complain that the resulting stream is encrypted, e.g. When I dump the video part with 'mplayer -dumpvideo lionz8.mpg' Playing lionz8.mpg Cache fill: 0.00% (0 bytes) Detected MPEG-PS file format! Encrypted stream but you did not request authentication!! MPEG: No Audio stream found. no sound core dumped:) What stream ID are you muxing the Z/Alpha streams under. Is it one that has a 'normal' meaning for DVD?I'm using 0x1f8 - could that be a.feeling guilty.;)Servus,Gernot-T-W-E-L-C-O-M-E-T The Austria Sweden connection. HO Homepage: E-F-U-T-U-R-E-/.
I am still quite confused that multiplexing MPEG Z/Alpha with the latest version of mplex makes mplayer complain that the resulting stream is encrypted, e.g. When I dump the video part with 'mplayer -dumpvideo lionz8.mpg' Playing lionz8.mpg Cache fill: 0.00% (0 bytes) Detected MPEG-PS file format! Encrypted stream but you did not request authentication!! MPEG: No Audio stream found. no sound core dumped:)What stream ID are you muxing the Z/Alpha streams under.
Is it one that has a'normal' meaning for DVD?Andrew. Hallo Added: zeroing of unit coefficient blocks. Implementation borrowed from ffmpeg (can't recall who proposed the idea originally).E sets the aggressiveness with which such blocks are zero-ed. I have found setting -E -20 to -10 works well. Set it too high and you get flickery pictures. Used well it can get you around 5% increased compression.
By default that option is disabled? Yep it is off by default.I have a weird question, in the manpage you write that the -E optionbehaves a litte bit differnent for positive and negativ values. Whatdoes the option do when you have as argument 0? The differnenc between the output size of mpeg2enc (version a few weeks ago) and the currrent CVS is alreaddy impressive. I'm not soure if that is because of the changes in the black size. That would be one part of it. There were some other subtler improvements too.Fine.:) older cvs cvs: mpeg2enc mpeg2enc -E 10 -E -10 -E -20 -K kvcd filesize: 57915402 435949 3718951 avg bitr: 2560000 192 16600 max bitr: 4410000 396 36400 You have to bear in mind though that -E and -K kvcd are (in effect) selectively increasing '-q' for selected DCT coefficients.
The really big question would be: how big are the files when they have the same visual quality.I have updated the howto (Section: Optimizing the stream) and includedthat. I have also included now the -K option description.Do you think it would be enough if the mpeg stream would be decompressedand afterwards the images compared? And the difference be recorded? Iguess it is not that easy.auf hoffentlich bald,Berni the Chaos of WoodquarterEmail: [email protected].
Hi Steven,I'd be interested to know what the bitrate would be with using the oldAltiVec DCT/IDCT code. In my observation, the accuracy of the DCT/IDCTcan have an increasing or decreasing effect on the bitrate depending onthe input stream.-JamesOn Thu, Oct 09, 2003 at 09:59:51PM -0700, Steven M.
Schultz wrote: Ok - I just finished encoding a 12962 frame.y4m file using the new Altivec DCT/IDCT routines. Pentium3: 6160000 bits/sec Altivec: 6380400 bits/sec Between the P3 and Altivec it's only 3.4% difference. That's not all that much but I wonder where the diff is coming from if it is not in the DCT/IDCT. The last time I ran the encoding on a dual Athlon system it came out a bit high: Athlon/3d: 7182400 bits/sec Actually I should run the Athlon test with the same identical.y4m data just to make sure that the DV decoding isn't tripping me up again.
'night. Steven Schultz. Hi James -On Thu, 9 Oct 2003, James Klicman wrote: The new DCT function is ready and has been imported into CVS. This DCT is very accurate, though slightly slower. Overall it seems that a I've encoded with the old 16bit DCT/IDCT and the new floating point DCT/IDCT and the improvement is visible with non-bitrate constrained output.
So the status of the AltiVec port is all AltiVec functions produce identical results as the C code except for the DCT/IDCT which producesOk - I just finished encoding a 12962 frame.y4m file usingthe new Altivec DCT/IDCT routines.Pentium3: 6160000 bits/secAltivec: 6380400 bits/secBetween the P3 and Altivec it's only 3.4% difference. That'snot all that much but I wonder where the diff is coming from ifit is not in the DCT/IDCT.The last time I ran the encoding on a dual Athlon system it came outa bit high:Athlon/3d: 7182400 bits/secActually I should run the Athlon test with the same identical.y4m data just to make sure that the DV decoding isn't trippingme up again.' Night.Steven Schultz.
On Thu, 9 Oct 2003, James Klicman wrote: So the status of the AltiVec port is all AltiVec functions produce identical results as the C code except for the DCT/IDCT which produces more accurate results than the C code. I believe that the AltiVec port may now potentially provide the highest quality output.I'm trying it now.The file fielddctbest.c was missing from utils/altivec/Makefile.amso I took the liberty of adding it and checking in a newMakefile.amNow to let the 1GHz G4 grind away for a while.;)Cheers,Steven Schultz. On Thu, 9 Oct 2003, James Klicman wrote: The new DCT function is ready and has been imported into CVS. This DCT is very accurate, though slightly slower.
Overall it seems that a 1 hour encoding might take an extra 12 seconds. That extra 12 seconds in12 sec out of 3600 or so probably isn't statistically significant(it wouldn't be noticed either in most cases;)). I've encoded with the old 16bit DCT/IDCT and the new floating point DCT/IDCT and the improvement is visible with non-bitrate constrained So the status of the AltiVec port is all AltiVec functions produce identical results as the C code except for the DCT/IDCT which produces more accurate results than the C code.Super! Thank you very much - I'll check it out tonight (viathe developer's cvs server since anon cvs is once againdeaf/dumb and several days out of step).Cheers,Steven Schultz. Hi Steven,The new DCT function is ready and has been imported into CVS.
This DCTis very accurate, though slightly slower. Overall it seems that a1 hour encoding might take an extra 12 seconds.
That extra 12 seconds inexchange for significantly improved accuracy is definitely worth it.I've encoded with the old 16bit DCT/IDCT and the new floating pointDCT/IDCT and the improvement is visible with non-bitrate constrainedoutput.So the status of the AltiVec port is all AltiVec functions produceidentical results as the C code except for the DCT/IDCT which producesmore accurate results than the C code. I believe that the AltiVecport may now potentially provide the highest quality output.-James. On Mon, 6 Oct 2003, James Klicman wrote: The AltiVec code has been synchronized with the changes to mpeg2enc. The code in CVS should compile now.I have good and interesting/puzzling/weird news.The good news is that mpeg2enc on an Altivec capable cpugives bitrate/filesize numbers that are quite close to thoseproduced on an Intel (Pentium3 in this case) based system.Good news - the new DCT/iDCT Altivec routines are noticeablyfaster than the old ones.Other news is that how the DV decoding is done can cause theoutput file to vary widely. On an Intel based system I wasusing either smilutils+avcodec(from ffmpeg) or ffmpeg directly.On the G4 based system I have to use smil2yuv+libdv (which is slowsince libdv is not Altivec optimized) because there are G4 OS/X/endian issues with avcodec at the moment.Earlier in the year (April 23) I performed tests that showed asizeable bitrate reduction when using ffmpeg's DV codec insteadof libdv and come to think of it the difference was 15%.What I did was take the 1.4GB DV file and process it on a P3/MMXsystem using ffmpeg1 to create a 6.3GB.y4m file which wasFTP'd over to the Powerbook.
Then on both the P3 and the G4systems the identical mpeg2enc command:mpeg2enc -f 8 -M 2 -q 7 -K default -N 0.5 -o foo.m2v.
![]() Comments are closed.
|
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |