1 2 Previous Next 42 Replies Latest reply on Nov 16, 2009 12:38 PM by Harm Millaard

    PPBM is Back--Benchmarking Premiere Pro CS4.1

    Bill Gehrke Most Valuable Participant

      Well after a long absence PPBM is back and is appropriately called PPBM4.  If you are not familiar with my previous Adobe Premiere Pro BenchMarking I have finally developed a new hardware benchmark specific to the current version of Premiere Pro.  As a matter of fact I do not advise using it on anything other than CS4 version 4.1.

       

      With this benchmark you can evaluate how well your hardware performs when actually using Premiere and as I get more feedback of results from forum users like you I will publish the results.  With that data you should be able to see how well your system is tuned, what hardware will improve your performance and/or help to specify a new system.

       

      There are three basic tests:

      1. I measure how long it takes to render the timeline with my "standardized" project.
      2. I measure how long the project takes to encode MPEG2-DVD
      3. I measure how long the project takes to encode to Microsoft DV AVI.

       

      Unfortunately it will not run on a Mac with the native OS.

       

      Even if you do not want to benchmark you should go out to this site to see how far Harm Millaard's super overclocked i7-920 system outperforms my previous generation dual Xeon 5400 series system.  Also, I have data on my system with identical hardware running on XP Pro 32, Vista 64 and Win 7 64 (RC).

        • 1. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
          Harm Millaard Level 7

          Bill,

           

          I want to be the first here to congratulate you on a job well done

           

          and to thank you for the enormous efforts and time spent you made to accomplish this.

           

          FABULOUS WORK.

          • 2. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
            Harm Millaard Level 7

            Bill,

             

            I think you are too modest. This is a very valuable tool to establish where strength's and weaknesses are in one's system.

             

            Everybody interested in improving the performance of his configuration should do the benchmark test.

             

            That is the sensible way to gather data about various setups and to derive what the success factors are to make one system faster/better than the other. Everybody can learn and profit from the data gathered and it only takes a couple of minutes.

             

            From my own perspective, I urge you to try it and to submit your results to Bill. I know he has spent so much time and effort on this and is willing to post your results on the site, so don't let this chance slip by you. The more participants, the more we can learn from the results and the better advise can be given for new or improved system setups.

             

            Do the benchmark!!!

            • 3. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
              nados Level 1

              The script bombs when i enter the path for the preview files.

               

              The script says not to enter: \Adobe Premiere Pro Preview Files\Benchmark.PRV

               

              So I tried:

               

              E:\test\Projects\Adobe Premiere Pro Preview Files

              E:\test\Projects\"Adobe Premiere Pro Preview Files"

              "E:\test\Projects\Adobe Premiere Pro Preview Files"

               

              So what can i enter that will work? Or are you implying that you need to enter ther full path "E:\test\Projects\Adobe Premiere Pro Preview Files\Benchmark.PRV" and not just the relative path "\Adobe Premiere Pro Preview Files\Benchmark.PRV"?

               

              In any case, nothing seems to work....

              • 4. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                nados Level 1

                Ok...

                 

                I had to open and read your script to figure out that you are appending "Adobe Premiere Pro Preview Files\Benchmark.PRV" to whatever path the user entered. Since the preview files folder is in the projects folder, i just had to enter "E:\test\Projects" again.

                 

                It would be a little clearer to modify the instruction to say something like "Enter the path to the preview files up to but not including "Adobe Premiere Pro Preview Files\Benchmark.PRV"" or  "Enter the path containing the preview files folder. "Adobe Premiere Pro Preview Files\Benchmark.PRV" will be appended to that path".

                 

                Also...

                 

                1. When entering numbers such as number of; cores, ram, cpu speed.... do you want us to enter the units (GHz, GB...) as well?

                2. When entering the cpu speed, do you want the manufacturers rated speed or the current Over Clocked speed?

                • 5. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                  Bill Gehrke Most Valuable Participant

                  Is this what you missed?  Maybe I have to make it bold faced.

                   

                  File-Path.jpg

                   

                  1.  No

                  2.  Actual (Overclocked speed)

                  • 6. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                    nados Level 1

                    Hi,

                     

                    I didn't miss it, i was confused by the way it's written.

                     

                    I just didn't get it because you wrote the instruction as two seemingly contradictory or disjointed statments. It needs a conjunction. A conjunction is a joining word - it links words or groups of words.

                     

                    For instince...You wrote:

                    "Enter the path to your benchmark preview files do not enter \Adobe Premiere Pro Preview Files\Benchmark.PRV"

                     

                    Some examples with a conjunction would be:

                    1. Enter the path to your benchmark preview files and do not enter \Adobe Premiere Pro Preview Files\Benchmark.PRV.

                    2. Enter the path to your benchmark preview files but do not enter \Adobe Premiere Pro Preview Files\Benchmark.PRV.

                    2. Enter the path to your benchmark preview files but leave out \Adobe Premiere Pro Preview Files\Benchmark.PRV.

                    3. Enter the path to your benchmark preview files but do not include \Adobe Premiere Pro Preview Files\Benchmark.PRV.

                    4. Enter the path to your benchmark preview files up to but not including "Adobe Premiere Pro Preview Files\Benchmark.PRV".

                    5. Enter the path containing the benchmark preview files folder. "Adobe Premiere Pro Preview Files\Benchmark.PRV" will be appended to that path.

                     

                    Just a thought.

                    • 7. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                      Bill Gehrke Most Valuable Participant

                      Sorry, I am not an English Major, just a stupid engineer.  Over a 100 people have successfully stumbled through this in the past.  I will put it on the list.

                      • 8. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                        nados Level 1
                        Sorry, I am not an English Major, just a stupid engineer.

                         

                        Neither am i. I'm not even an engineer - however, I was a systems and applications programmer. I just tend to take things litterally. So I figured it out the programmers way - i looked at your code!

                         

                        In fact, your code was more clear in the way it read (with the "&" ampersand read as "and") than displayed...

                         

                        strPath = InputBox("Enter the path to your benchmark preview files" & vbCrLf  & "do not enter \Adobe Premiere Pro Preview Files\Benchmark.PRV" ,"Path")

                         

                        also, this line - appending the path variable with the string - clinched it...

                         

                        Path = strPath  & "\Adobe Premiere Pro Preview Files\Benchmark.PRV"

                         

                        Anyway, I'm not trying to beat you up........... well maybe just a little bit - uppercut - uppercut - jab!   I just thought it could be more clear, but as you said, no one else seems to have gotten confused.

                        • 9. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                          the_wine_snob Level 9

                          Bill,

                           

                          Thank you for your continued work on this.

                           

                          I've already linked to this thread 3x today, as others want to find their bottlenecks and you provide them a means.

                           

                          Appreciated,

                           

                          Hunt

                          1 person found this helpful
                          • 10. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                            Harm Millaard Level 7

                            Bill Hunt,

                             

                            Time to upgrade to 4.1 and run the test yourself.

                             

                            With two Bill's here, I had to use your last name as well.

                            • 11. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                              fred tyler Level 1

                              Hi Bill G, thanks again but when will you be able to post more results?

                               

                              regards fred

                              • 12. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                Harm Millaard Level 7

                                Fred,

                                 

                                AFAIK Bill G. is currently checking the accuracy of test results submitted so far. Earliest results show some things that need to be checked before publishing, but knowing Bill he is working on that. I would expect them shortly. Just a little patience, so the published results are accurate. Just this morning I gave Bill some additional info that may help to get the results out ASAP.

                                 

                                Anyway, I did not yet see your results amongst the submitted data, or am I wrong on that account? If you haven't submitted your data yet, please do as it will help to make the benchmark more reliable and informational for all of us.

                                • 13. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                  Bill Gehrke Most Valuable Participant

                                  Hello, sorry but sleep and confusion with new data over some previous conclusions interupped my posting of the first results.  I will be making some major changes on my conclusions with this new data that I am now gathering.  Nuances of 64-bit OS's are the cause.  I will have to be making some hardware changes and rerunning a bunch tests.

                                   

                                  Also I will be providing data on my experience with SSD's

                                   

                                  I have posted the results that I had as of early last night, only 4 so far and I know that there are probably some changes to them I will be incorporating.

                                  • 14. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                    fred tyler Level 1

                                    No probs, have sent output text file again.

                                     

                                    regards fred

                                    • 15. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                      Ferdie2

                                      Well, I'm 4th on the list.

                                       

                                      Smilodon,   Personal or Computer ID
                                      Asus,  Computer Manufacturer
                                      P6T SE,   Computer Model
                                      60.6,  secs Total Benchmark Time
                                      3.6,  secs AVI  Encoding Time
                                      36,  secs MPEG Elapsed Time
                                      21,  secs Rendering Time
                                      Intel,    CPU Manufacturer
                                      Core i7,    CPU Model
                                      2.67,    GHz CPU speed
                                      1,    Number of CPU chips
                                      4,    Total Number of Cores
                                      12,    GB RAM
                                      4.1.0,    APP Version PPBM4 DV
                                      Windows 7 64-bit,    OSVersion
                                      SATA,    OS Disk Interface
                                      300,    GB OS Disk Capacity
                                      10,000,    OS Disk Speed
                                      SATA,    Project Disk Interface
                                      1000,    GB Project Disk Capacity
                                      7200,    Project Disk Speed
                                      SATA,    Preview Disk Interface
                                      1000,    GB Preview Disk Capacity
                                      7200,    Preview Disk Speed
                                      SATA,    Output Disk Interface
                                      1000,    GB Output Disk Capacity
                                      7200,    Output Disk Speed
                                      nVidia 260,    Graphics Board
                                      ,    Comment

                                      • 16. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                        Ferdie2 Level 1

                                        Also, I agree with Nados above regarding the confusing instructions with the path.  It took me several tries using trial and error.  It would be nice if this screen was a little clearer.

                                        • 17. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                          Bill Gehrke Most Valuable Participant

                                          Ferdie2

                                           

                                          Please follow the instructions to email me the results so I can easily enter your data.

                                          • 18. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                            Ferdie2 Level 1

                                            Sent yesterday.  Let me know if your spam filter ate it.

                                            • 19. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                              Ferdie2 Level 1

                                              So, if I overclocked, I would be in Harm's company?

                                              • 20. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                Bill Gehrke Most Valuable Participant

                                                YES, If you are careful.

                                                • 21. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                  Ferdie2 Level 1

                                                  You know, Windows Experience gave my WD Velociraptor 10k 300GB drive a 5.9.  Everything else got a 7.4 or above.  How much stock should I put into this score?  I'm really personally offended by it.

                                                  • 22. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                    Bill Gehrke Most Valuable Participant

                                                    I know that 5.9 for VelociRaptor score.  I finally got a 6.9 score by using a SSD OS drive, but I found insignificant performace increase in my PPBM benchmark.  And who in this forum is interested in 3D gaming performance?

                                                    • 23. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                      Harm Millaard Level 7

                                                      Ferdie,

                                                       

                                                      Why would you feel offended? The index number is rather meaningless, apart from showing you the weakest link in your chain. Your PPBM results are great, in the top 10.

                                                       

                                                      I have the same score 5.9 for my Velociraptor, and 7.6, 7.7, 7.5 and 7.5 for the other tests. So what?

                                                       

                                                      The question is really, are you happy with the responsiveness of your system with your source material? If you want to test how your system relates to other systems, there are two benchmarks that make sense, one is PPBM4 (that Jim Simon is apparently unwilling to submit his results to, possibly  because he may be ashamed of his results???) and the other one is PassMark, altough that is more generic and more tuned to gamers.

                                                      • 24. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                        Jim_Simon Level 9

                                                        Jim Simon is apparently unwilling to submit his results to, possibly  because he may be ashamed of his results?

                                                         

                                                        Oh, you do like to tease.

                                                         

                                                        No, not ashamed.  Bill's script threw up twice, all over my nice clean sweater.  I gave up and did some laundry.

                                                        • 25. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                          Bill Gehrke Most Valuable Participant

                                                          Send me the bill for the dry cleaning.

                                                           

                                                          I looked into scripting to bring up a window to browse to the location.  Now if you will go back to Windows XP I will provide a version that you can browse with Windows Explorer to the file locations.  Unfortunately Microsoft changed things and it will not work in Vista or Win7.

                                                          • 26. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                            Jim_Simon Level 9

                                                            Well...whatever your scripting guy did might not work.  But as the "Browse" operation is still pretty standard for Vista and 7, it clearly does work, however different the method of doing it may be.

                                                            • 27. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                              Bill Gehrke Most Valuable Participant

                                                              Jim, he is getting his first exposure to Win 7 tomorrow.  He has completely bypassed Vista.  He is retired and a volunteer sysadmin for a church 70+ computer network.  I will give him a good chance to get comfortable. I do imagine he will find a method.

                                                              • 29. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                Harm Millaard Level 7

                                                                Bill,

                                                                 

                                                                What fabulous improvements on your score, especially on the AVI results with the new Areca drivers. I hope the earlier problems are now history.

                                                                 

                                                                Jim Simon's results were extremely disappointing. He did not even make the cut, in golfer's terms. Probably around 350 seconds or above.

                                                                • 30. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                  Ferdie2 Level 1

                                                                  Hey, I edged our FredT.  MUHAHAHA!

                                                                  • 31. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                    fred tyler Level 1

                                                                    Fertie2 cheating *******!  'sarc"

                                                                     

                                                                    regards fredt

                                                                    • 32. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                      Bill Gehrke Most Valuable Participant

                                                                      If you do not like the first result you can always try again.

                                                                       

                                                                      A possible reason I can see from the data might be the 10k rpm OS disk, but I highly suspect it really may be in the tuning of Win7 because Fred's CPU score (MPEG encoding) and the Render Timeline (quite CPU dependent) are appreciably higher than Smilodon's (Ferdie2).   Notice Fred's RAID array does score appreciable better on the disk intensive AVI encoding.  Fred try reducing the number of processes you have running and retest.

                                                                       

                                                                      But really, as you can see from my linked multiple runs (for instance the Win7 data) there can easily be +0.5/-0.8 or more variance between my small test sample of 10 runs with Win7.  I have found much less varience that with

                                                                       

                                                                      Thank you both for providing a great PPBM baseline for for non-overclocked i7-920 performance.

                                                                       

                                                                      And neither of you had the script throw up all over your sweaters!!!!

                                                                      • 33. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                        iamjohnsscreenname

                                                                        Just emailed my pleasing results to Bill.

                                                                         

                                                                        Now I need to interpret them because I don't know which hardware is the slowest/weakest link.

                                                                         

                                                                        71.3 seconds total benchmark time

                                                                        2.9 seconds avi encode

                                                                        36.4 seconds mp4 encode

                                                                        32 seconds rendering time

                                                                         

                                                                        You can see the system specs when they appear online. Both my 74GB raptor system drive and my 2TB (R)AID 0 project drive are packed to 50% and 75% respectively, so maybe when I put in the empty 150GB raptor and clear all the junk from the RAID, the numbers will improve. Or maybe if I double the RAM. Or maybe it's pretty damn fine as it is...

                                                                        • 34. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                          Bill Gehrke Most Valuable Participant

                                                                          When I see and enter your email I will have a basis to comment back to you, without your complete system details shown (like CPU and overclocking?) it is impossible to intelligently comment.

                                                                          • 35. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                            skeeze

                                                                            Here are my results...great idea.  Please note that with Passmark, if you are running Windows 7, your 2D scores are goign to be much lower due to a change in the way Micosoft handles graphics now.  SSD's don't show reliable stat's in that program either...had to use another one, but great for everythign else.

                                                                             

                                                                            BZ800 HP,   Personal or Computer ID
                                                                            HP,  Computer Manufacturer
                                                                            HPZ800,   Computer Model
                                                                            46.6,  secs Total Benchmark Time
                                                                            1.6,  secs AVI  Encoding Time
                                                                            24,  secs MPEG Elapsed Time
                                                                            21,  secs Rendering Time
                                                                            Intel,    CPU Manufacturer
                                                                            Xeon W5590 3.33,    CPU Model
                                                                            3.33,    GHz CPU speed
                                                                            2,    Number of CPU chips
                                                                            8,    Total Number of Cores
                                                                            36,    GB RAM
                                                                            4.2,    APP Version PPBM4 DV
                                                                            Win 7 64,    OSVersion
                                                                            SATA,    OS Disk Interface
                                                                            128 SSD,    GB OS Disk Capacity
                                                                            15000,    OS Disk Speed
                                                                            SATA,    Project Disk Interface
                                                                            1024,    GB Project Disk Capacity
                                                                            7200,    Project Disk Speed
                                                                            SATA,    Preview Disk Interface
                                                                            160,    GB Preview Disk Capacity
                                                                            Same as Project Disk,    Preview Disk Speed
                                                                            SATA,    Output Disk Interface
                                                                            1024,    GB Output Disk Capacity
                                                                            Save as Project Disk,    Output Disk Speed
                                                                            NVidia Quadro FX3800,    Graphics Board
                                                                            SSD for C:, water cooled..not using hyperthreading,    Comment

                                                                            • 36. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                              Bill Gehrke Most Valuable Participant

                                                                              Well I have posted these twin W5590 results on the PPBM4 Benchmark Results web page along with Michael's W5580 results.  These results have taken away the CPU intensive MPEG encoding crown that Harm has been wearing. but not by any means his well tuned and overclocked total score crown.  Also posted at this ungodly hour of the morning (for me) are the results from Michael's twin W5580's where he has shown absolutely amazing disk intensive AVI encoding results with just two SAS 15,000 rpm disks!  Both of these systems possibly can yet be fine tuned.

                                                                               

                                                                              Thanks to all you friends for the 34 current results that cover a wide range of systems!

                                                                              • 37. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                                skeeze Level 1

                                                                                Many thanks...so, based on what you have seen and know, what disks make the most difference for overall speed, the project disk, scratch disk, or both?  I thought the disk really wasn't that important with rendering based on all I have read, but obviously that is not the case.

                                                                                 

                                                                                Thanks in advance, this was a great idea you had.

                                                                                • 38. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                                  Bill Gehrke Most Valuable Participant
                                                                                  1. If you are using a single fast RAID array then my tests show that all the files (project, media, scratch) should be on the RAID array, putting any one on a slower separate drive slows the benchmark.  I have all my project files on a 500 MB/s average read rate Array, if I put the Preview Files or the Output on even a ~200 MB/s RAID array you get slower performance.
                                                                                  2. If you prefer not to use RAID, then using the forum recommended three drives, Take your choice (actually both are essentially the same)::
                                                                                    1. Then if you have faster and slower drives depends on what you want to speedup.  My opinion is use your fastest single drive drive for the OS/Applications.  If you Render the Timeline and intend to use the Preview Files for encoding you may want to put them on another fast disk.  The when you are trying to optimize encoding from AME this is a CPU intensive operation and a slightly slower Output Files disk probably will work perfectly adequate.  I would guess that Media files (my benchmark does not have media files by design) would also benefit from fast disk access. 

                                                                                     

                                                                                    Skeeze, when you say "rendering" are referring to encoding/transcoding as done in AME or are you referring to Rendering the Timeline as in hitting "Enter".

                                                                                     

                                                                                    I am absolutely amazed at the Michael's W5580 disk intensive AVI encoding results with only two SAS drives, since it is also a brand new system I suspect that they are Seagate 15K.7 drives with sustained read transfer rates from 204 to 122 MB/sec.  This one drive is equivalent to a pair of Seagate 7200.12 rpm drives in RAID 0!

                                                                                     

                                                                                    I would appreciate comments on my opinions above.

                                                                                      .
                                                                                    • 39. Re: PPBM is Back--Benchmarking Premiere Pro CS4.1
                                                                                      Harm Millaard Level 7

                                                                                      I absolutely agree with Bill on this.

                                                                                       

                                                                                      Running the benchmark on my fastest array (average 850+ MB/s transfer) gave me the (still) highest score of 13 s on the rendering test. Dividing the preview files and project files amongst several raids (one the very fast one and the other a plain 2 disk raid0) decreased my performance significantly and my rendering times went down to around 21 s, like yours.

                                                                                      1 2 Previous Next