5 Replies Latest reply on Apr 23, 2010 12:56 PM by shooternz

    CS5 Benchmark with PPBM4

    Bill Gehrke Most Valuable Participant

      This is a double post as many readers here do not necessarily read the Hardware forum

       

      Harm's Beta CS5 benchmark with PPBM4 is now posted!  WOW!  And all that improvement without a CUDA graphics card so when we see the Mercury Playback Engine additional improvement is sure to appear.  Adobe must have hired a great multithreading expert/team.  Congratulations Adobe.  It must have hurt the disk performance slightly, but that is small peanuts.  It reduced his MPEG encoding time from 24.6 seconds under CS4 to 6.4 seconds in CS5, almost 200% faster!

       

      Yes there probably will be a new PPBM5 and if and when and it will include H.264 Blu-ray encoding and it will also have an AVCHD segment.  Since I do not have CS5, you may have to wait awhile.

        • 1. Re: CS5 Benchmark with PPBM4
          Harm Millaard Level 7

          That was fast Bill!


          • 2. Re: CS5 Benchmark with PPBM4
            shooternz Level 6

            Any concerns about what appears to have slowed down in Harms Benchmarks?  eg timeline exprt and avi rendering

            • 3. Re: CS5 Benchmark with PPBM4
              Bill Gehrke Most Valuable Participant

              Not really, they are both disk sensitive benchmarks.  AVI Encoding is very disk intensive and Render Timeline is somewhat disk intensive.  Let me preface this that first of all I am hardware oriented and know absolutely nothing about software, but if graphics performance or CPU performance have higher priority it logically could affect the disk performance.

               

              There are a lot of unexplained anomalies in the data as I sort through it, like looking at the Render Timeline results.  There are the lowest group that run 11-14, and the next grouping is 21-22.  Why with over a 100 results is their such a gap in the data?

               

              I rarely look at the total score except to catalog the scores.  Then there is my experimental board with the hex core CPU and the AVI encoding scores.  I accidentally found a way to lower the scores that I was getting, I have to have another program running in the background to get the scores you see posted.  Notice that the other P6T7 WS Supercomputer boards do not do well in AVI encoding.  Having another program running while you benchmark something is contrary to all logic

              • 4. Re: CS5 Benchmark with PPBM4
                Harm Millaard Level 7

                Craig,

                 

                I have been wondering about this as well and have no explanation for the differences. I have tested it multiple times and these results were pretty consistent. While prepraring for a PPBM5 benchmark, there were other strange things I encountered that we have to figure out before going public.

                 

                One thing that may have influenced the results in a negative way, is that the project file that I tested from does include additional tests and actual footage for the PPBM5 benchmark with two other sequences with multi-track material from AVCHD, HDV and XDCAM-EX material and different frame rates, loaded with effects and transitions for MPEG testing and H.264 testing, although this should not impact the test results in any significant way.

                 

                The doubling of my AVI results are disappointing and I simply do not know why. However, since the bulk of the time consuming work will be encoding to either MPEG or H.264 and no longer to AVI in todays environments, it does not really worry me.

                 

                As testing progresses and we are close to publishing, we will give you extensive information about the changes we made and the reasons behind them.

                • 5. Re: CS5 Benchmark with PPBM4
                  shooternz Level 6

                  Worries (or dissapoints)  me slightly because I use uncompressed avi and mov in my broadcast workflow.

                   

                  I dont expect it to be a big deal though because I am expecting performance gains at other levels in my new 64-bit system.

                   

                  Interestingly - I have no issues with the avi /mov export times in my CS4 system.  They are real fast  from complex timelines eg 1- 2 minutes (note: my movies are usually 15 secs to 90 secs long - TV commercials).