28 Replies Latest reply on Oct 25, 2010 7:48 AM by cmeira

    CS5 Direct Export 3X Faster Than Media Encoder-Why?

    Dailey Pike

      I created a video that runs about three minutes. I exported it from 1440 X 1080 to 1280 X 720 with the Media Encoder and it took about 11 minutes.

       

      I encoded the exact same video/settings with Direct Export and it took less that 3 minutes.

       

      A pleasant surprise but how come??

        • 1. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
          Dailey Pike Level 1

          Wow, I can't believe I stumped everybody.

          • 2. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
            Harm Millaard Level 7

            Maybe because it can't be replicated here.

            • 4. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
              RDA972 Level 3

              On my machine, exporting a 10-minute XDCAM EX HQ file with no effects to Blu-ray MPEG2 (High Quality) took 2 minutes and 22 seconds using Queue compared to 2 minutes and 19 seconds using Export. Hardly 3x faster I might add.

              • 5. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                shooternz Level 6
                Wow, I can't believe I stumped everybody..

                Didn't stump me.  ......I believe in magic.

                • 6. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                  Dailey Pike Level 1

                  Apparently I'm not the only magical one.

                   

                  http://forums.adobe.com/message/2805313#2805313

                   

                  "Media Encoder CS5 Faster Within PPRO Than Queued"

                   

                  So the results I got HAVE been replicated.

                  • 7. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                    shooternz Level 6

                    First time I have ever read that the issue is "things got faster"

                     

                     

                     

                    Now you have  choices :  Faster way or Slower way

                    • 8. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                      ChrissaSwenson Level 1

                      Harm, I can't figure your response out.  Why would it not be valuable to the poster to hear that others could not replicate the issue?  Indeed, it would be good information.

                      • 9. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                        ChrissaSwenson Level 1

                        Shooternz, do you think people will take this response as friendly forum-community humor (as I did) or possibly a glib and valueless contribution?

                        • 10. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                          ChrissaSwenson Level 1

                          Dailey, see my thread if you wish.  A kind response regarding task priority was posted recently.  Possibly AME standalone is sharing available resources to a higher degree than when requested directly within PPro.  We'll see.

                          • 11. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                            Harm Millaard Level 7

                            Chrissa,

                             

                            It is worthwhile to mention that only a very few machines are claimed to be influenced by some mysterious magic, and the majority  of the systems do not suffer from this magic and can not be induced to show that behavior.

                             

                            In the second place, the priority settings posted in the other thread are not accurate. The regular priority settings for AME are Normal at 8, and the maximum reliable settings are High at 13. But the effect is negligible. The argument incorrectly assumes a priority setting of Below Normal at 6, which is not the case. Even with long queues and long encoding times, there is no discernable difference in encoding times with priority settings at high versus normal. The only situation where there may be a slight difference is on systems with limited memory and a lot of superfluous processes/services running, but then these systems never perform very good to start with.

                            • 12. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                              ChrissaSwenson Level 1

                              Now Harm, THAT is good and appreciated information.

                              • 13. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                Dailey Pike Level 1

                                Harm said, "It is worthwhile to mention that only a very few machines are claimed to be influenced by some mysterious magic, and the majority of the systems do not suffer from this magic and can not be induced to show that behavior."

                                 

                                Use of the word "magic" trivializes a sincere and honest question.

                                 

                                Use of the word "claimed" implies it's not happening and we're making it up for some reason.

                                 

                                Since Harm is truly the forum's Wizard of Oz, I now don't know if I can believe my own lying eyes.

                                 

                                Harm, you take it much too seriously when you get stumped by a question.

                                • 14. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                  Ann Bens Adobe Community Professional & MVP

                                  I find Direct Export way faster than the Queue.

                                  • 15. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                    Steve C2 Level 2

                                    I had suggested that experimentation with process priorities within the Windows Task Manager might be useful.  I do not know the internal architecture of any Adobe product, but Microsoft provides users with the ability to either raise or lower process priorities so that we have greater control over how and when processes are completed.

                                     

                                    I don't have enough data, yet, with the PrPro to claim that raising the process priority will make a difference.  I can say, however, beyond a doubt, that the the encoding process within Premiere Elements benefits from raising the process priority.

                                     

                                    It is also a reasonable guess that the "Queue" encoding option could take longer, as some people have reported.  In general, one wants batch processes to use resources that are not used by interactive processes.  If this is not what Adobe intended by providing "Queue" encoding, then it would be valuable for someone from the Premiere architecture team to comment.  A thought...for some of the people who have noticed, on their machines, the discrepancy in completion times, would you be willing to contact Adobe and see if it will comment on this and then post your responses?

                                     

                                    Regarding the comment about whether there are unnecessary processes running...Yes, it can be valuable to end those processes.  My experience is, with 8GB RAM, that if there is no faulting (one can open the Windows Resource Manager) to determine this, work will complete more quickly.

                                     

                                    One more thought...How about the people who notice a discrepancy share their systems and those that do not notice a discrepancey also share their systems. in particular the amount of memory, whether or not they have a CUDA-based video adapter, and the number of cores in their cpu.  Could it be that the more powerful pcs have more cores and are able to process more work simultanelously?

                                    1 person found this helpful
                                    • 16. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                      shooternz Level 6
                                      Shooternz, do you think people will take this response as friendly forum-community humor (as I did) or possibly a glib and valueless contribution?

                                       

                                      OUCH!

                                       

                                      I would expect those with a sense of humour to get  it for what its worth.  Glib and valuless as you so rightly identified.

                                       

                                      Its tough being an editor locked in a semi darkened cave by oneself all day.  I apologise for  my weak attempts to bring some light in.

                                       

                                      Sorry ...It wont happen again ma'am.

                                       

                                        (serious face)

                                      • 17. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                        Harm Millaard Level 7

                                        In testing the new PPBM CS5 benchmark, I have tried various approaches, queueing single files, multiple files, exporting and in various formats, AVI, MPEG2-DVD and H.264, with and without metadata, with and without preview files, with and without audio, with and without MRQ, in short everything one can think of and never have I seen a discernible difference in encoding times.

                                         

                                        Sure, I sometimes had an export taking 98.3333 seconds, while in the queue is took 98.6667 seconds, but that is really negligent. And I have seen it the other way around. Ann has a valid point in that the queue may take some time to start when one has a long queue, say 30+ files, but not in the encoding.

                                         

                                        Someone claiming that a file that takes let's say 98.6667 seconds in the queue, only takes 32.8889 seconds on export (that is 3 x faster) is something I have never seen, and I have run this new benchmark hundreds of times. Those claims are not substantiated in any testing I have done and my reaction is:

                                         

                                        "I've gotta see it before I believe it, said the blind man."

                                        • 18. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                          Ann Bens Adobe Community Professional & MVP

                                          No i am talking of exporting 1 file.

                                          • 19. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                            Dailey Pike Level 1

                                            Harm said, "I gotta see it before I believe it."

                                             

                                            You can't see the air you breathe can you Harm... oops, my bad, you live in Finland where that volcano just blew up.

                                             

                                            You remind me of the guy who refused to issue any more patents.

                                             

                                            "Everything that can be invented HAS been invented."

                                            Charles H. Duell

                                            U.S. Patent Office 1899

                                             

                                            No correct answer points for you!!

                                            • 20. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                              KaiserBoy Level 1

                                              I see the same 3X-ish difference in my queued exports, too.  Very odd indeed, but I thought about it and want to put something out there that might be worth a think.

                                               

                                              In PPro, one has CUDA cards that meet the requirements and are in the list of such that PPro uses to configure its performance.  When PPro knows there is a good CUDA in there, it exports faster, right?  Mine sure does, and I've tested this by taking it off the list and seeing the slowdown.

                                               

                                              I have added my non-accepted but CUDA-compliant card to the list in the workaround process described recently and known by most of us.

                                               

                                              So, what if Media Encoder needs to know this list, too, and the "patch job" that we've done has somehow not given Media Encoder the go-ahead for our non-standard card?

                                               

                                              OR, what if some of us have experienced an installation that has not given Media Encoder the CUDA info correctly, and it trundles along like it doesn't know of any acceleration capability in the video card?

                                              • 21. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                Harm Millaard Level 7
                                                you live in Finland where that volcano just blew up.

                                                 

                                                You are only 1,600 miles off!. Did the same happen with your timings of 11 and 30 minutes, or was it 3 and was it measured in nanoseconds?

                                                • 22. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                  Dailey Pike Level 1

                                                  Harm, they're casting Grumpy Old Men 3.

                                                   

                                                  I think you have a good shot.

                                                   

                                                  Sorry you couldn't answer my question.

                                                   

                                                  I know that vexes you greatly.

                                                   

                                                  However, I have no doubt you will respond to this post in your Don Quixote-like quest to get in the last word.

                                                   

                                                  Prove me wrong.

                                                  • 23. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                    Jeff Bellune Adobe Community Professional

                                                    To Harm and Dailey,

                                                     

                                                    Gentlemen:

                                                     

                                                    Take it outside.  It doesn't belong in here.

                                                     

                                                    -Jeff

                                                    • 24. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                      shooternz Level 6

                                                      I did a few quick tests on a short project (30 second TVC  720P to QT, Uncompressed 8 bit, 10bit, DVCPROHD, avi)

                                                       

                                                      I found the Direct Export to be approximately 20% faster than going thru the Queue Encoder in each case.

                                                       

                                                      NOte: Number of bits makes a difference in speed and filesize.

                                                       

                                                      The Exports were all faster than realtime anyway and a significant degree faster than my CS4 system.

                                                       

                                                      FWIW ; MPE / FX3800 / I7-930 /12GBs/DDR3-1600

                                                      • 25. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                        firetrak Level 1

                                                        One possible reason for the direct encoder to work faster, than the media encoder queue, could be down to the fact that you are supposed to be able to edit whilst exporting via the queue, but not when using the direct encoder.

                                                         

                                                        In other words the direct encoder probably has more resources to work with vs the queue, which may be sharing resources.

                                                         

                                                        For those of you that have found the speed quicker on direct export, try closing ppro cs5 before starting the queue and see if the same sluggishness still occurs.

                                                         

                                                        ps. i found the magical comments rather fun.... didn't really think harm or shooternz where being in any way rude or whatever.... lighten up people.

                                                        • 26. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                          cmeira Level 1

                                                          But there is a bigger problem than this speed issue:

                                                          I exported a small video as XVid (CS5, Win7/64) -- directly within the export dialog window of Premiere and via the Encoder. Exactly the same settings, but two different results!

                                                           

                                                          The video exported directly was fine. But the video exported with the Encoder is jolting (in German: "ruckelnd").

                                                           

                                                          I tested this and played the videos several times -- each time the same result!

                                                           

                                                          Only the file sizes and the kbps value (and the meta data) of this small and simple video (21 seconds) differs slightly:

                                                           

                                                          # exported directly:
                                                          shot_2010-10-21 011.png

                                                           

                                                          # exported via Encoder:

                                                          shot_2010-10-21 009.png

                                                           

                                                          Unbelievably!

                                                          • 27. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                            davidbeisner2010 Level 3

                                                            Interesting thread here... and I'd agree, people need to lighten up! The humor was not lost on me, and it was quite enjoyable :-)

                                                             

                                                            FWIW, I just ran a test of exports to see if I could reproduce this strange phenomena.

                                                             

                                                            Ran a 3:22 timeline at 1280x720 23.98fps, with a mix of 1920x1080 23.98fps h.264 files with a mov wrapper, exported from REDCINE-X 4k files and 1920x1080 29.97 fps AVCHD files from a cheap Sony handycam. Checked the box each time I exported to have it use the settings that most closely matched the sequence, and here are my results:

                                                             

                                                            1st export, direct from PPro, took 2:04.

                                                            2nd export, from AME with PPro still open, took 2:03 + 0:09 for the project to load the encode to actually begin

                                                            3rd export, from AME with PPro closed, took 2:07 + 0:03 for the project to load and the encode to actually begin.

                                                             

                                                            So the only difference I saw in export times that could give direct export a leg up, is the fact that you don't have to wait for AME to load the PPro project and prepare the encode settings. All exports were still better than real time, and SIGNIFICANTLY better than CS4.

                                                             

                                                            I've got a 3DBOXX system with Win7x64/i7-920/12GB-DDR31600/7200RPM SATA RAID-0 for data and 7200RPM SATA for Projects and 7200RPM SATA for OS/Program Files, and a GTX285 CUDA enabled card.

                                                            • 28. Re: CS5 Direct Export 3X Faster Than Media Encoder-Why?
                                                              cmeira Level 1

                                                              ok, maybe you're right

                                                               

                                                              But do you see any differences between the two films?