6 Replies Latest reply on Oct 14, 2010 9:57 AM by sldwaa

    Adobe Premier Pro CS5 Build Question


      I am in the process of building a PC with the primary purpose of video editing, not gaming. I will be using Adobe Premier Pro CS5. I have been researching for several weeks (I did not realize until recently I should be in this forum as it focuses on the software I will be using instead of all the other gaming focused forums), and have two questions about hardware selections.  Price is not an option, but do not want to be excessive if not necessary (jr. needs new shoes for kindergarten afterall).  I have a Panasonic HDC-TM700 camera for my raw video (records in 1080p) hence the need for BR burning due to space considerations for disc burning (video of the kids to grandma, etc - I guess I will need to buy grandma a BR player - I am still not sure she mastered using the VHS player yet), and also have tons of mini-DV tapes I need to capture and edit.  Here is what I have selected to date with questions embedded into the selections:

      OS: Windows 7 64 bit ultimate (overkill but I am purchasing for $50 - yes, legal copy)

      Processor:  Here is one area I am struggling with which has the biggest financial impact on my build.  I have read hundreds of posts on this issue in other forums and am confused on speed vs number of cores and the actual benefits which will result using Premier CS5 (the posts have primarily been on gaming benefits hence not applicable to my needs).  The processors in scope are:

      I can afford the 980X as well as the 970 Gulftowns, however the 950 Bloomfield is much more cost effective. I guess what I need to know is will the leap from quad core to hex core be worth the extra $600?  If that is the case, is the extra $100 for the 980x worth it (I suspect not).  Any guidance on this item will be greatly appreciated.  I have read the 4 page posting in this forum on PC building, but am still somewhat confused on this item.  I know what the gamers recommend, but need to understand from a video editing perspective.


      Here is the other item I am struggling with - video card.  I reviewed what is compatible with CS5 from this Adobe website (mercury engine, CUDA, etc...).  I do not want to spring for the Quadro series (quite spendy), and believe I am going to settle for the NVIDIA GeForce GTX 470.  I understand some of its shortcomings such as running hot, power, etc., but believe it is the most cost effective for running the CS5 and is poised for future apps (correct me if I am wrong).  Any guidance on this will be appreciated. I have spent several hours and weeks on this and am at a standstill.  If you agree on the NVIDIA GTX 470, which should I buy? I like the MSI N470GTX Twin Frozr II GeForce GTX 470 (Fermi) 1280MB 320-bit GDDR5 as it has good components and good cooling reviews, but I know Gigabyte, Palit, ASUA all have NVIDIA GeForce GTX 470 products (just to name a few) and EVGA also has lifetime warranties.  Which NVIDIA GeForce GTX 470 should I buy?


      The next is informational on my build thoughts for the rest of the components (not all inclusive) but only provided for info and comments if anything is obviously out of line....

      Motherboard:  MSI Big Bang-XPower LGA 1366 (IntelX58 SATA 6Gb/s USB 3.0 ATX Intel Motherboard).

      Minimum needs: 2 PCI express 2.0 16 slots, 2 eSATA (I currently have multiple eSATA external storage drives), 2 USB 3.0 (for future), SATA 6 GB/S (for future - but is overkill in my opinion but buying just in case) and SATA RAID(0/1/5/10).  Board looks good for next few years expansion. ASUS Rampage Extreme only has one eSATA so I could not use, and the GIGABYTE-UD3/7/9 all look good so I could use this, however, the MSI has the overclocking genie which makes overclocking easy for folks like me if needed.  It was also rated very highly in many reviews.  They also say the capacitors, etc are military grade,etc. - it is marketing lingo in my opinion, but it cannot hurt.  Some posts say MSI is not as reliable as GIGABYTE and ASUS, but I will willing to take the risk.


      Other items:

      12 GB DDR3 G.Skill

      WD VelociRaptor OS drive, 10,000 RPM 600 GB

      Storage Drives:  4X 1TB each SATA 3 7200 RPM something (TBD) (RAID 10)

      BR burner (TBD)

      Case (TBD) most likely Lian Li PC-A77 but not sure

      PSU - 850 Watt

      CPU cooler TBD

      Monitor 2X of Dell 24" S2409W (recommended by both consumer reports and CNET)



      Help please with the two items above (CPU and video card). After I select these components, I move onto confirming the case, power supply, etc....  All comments, questions, constructive feedback appreciated.

        • 1. Re: Adobe Premier Pro CS5 Build Question
          Harm Millaard Level 7

          Check the configurations here: PPBM5 Benchmark


          The top positions are all 980X or X5680 machines, with a lonely 920 around position 11 or 12. All top performers have 24 GB or more. How crucial is the last second gain in your workflow? When do you want to invest in new technology? Are you willing to wait for a new X68 mobo, an octa-core Sandy Bridge and 32 GB memory in DDR4-4000 architecture?


          Your wallet is your own business, you have to decide whether the premium for top-of-the-bill components is worth it to you. Remember that what you buy today is outdated by tomorrow and a lot cheaper.

          • 2. Re: Adobe Premier Pro CS5 Build Question
            sldwaa Level 1

            I reviewed the benchmark data you directed me to - thank you for that.  Now, I am going to make a gross generalization based on the data, after all, one data point is a trend correct? (at least that is what I learned in Six Sigma school)...  .  Please advise if I am anywhere in the ballpark or if I am extremely far off base.


            If I assume 12 meg of ram and an i7-930 or 950 processor (similar results in benchmarks) vs. the i7-980X with the same ram (obviously many other factors have to be factored in but this is a gross analysis with no overclocking), it takes at least 2x time for most benchmark tests, of which the overall performance index is also approximately 2x different.  If the H.264 encoding time is ~50 seconds for a 45 meg test file used (size I believe used for testing) for i7-980X and the time is ~100 seconds for the i7930 or 950, and if I have a file 32 Gig in size (max size for me), then the difference for time for me would be (assuming my size is 711X greater than test sample) or ((32,000,000,000 divided by 45,000,000 = 711)), then if I purchase the i7-950 my total time would be approximately 35,550 seconds longer than if I have the i7-980x (50 second difference between the 2 processors multiplied by 711 size difference = 35,500).  This is approximately 10 hours more (9.88).  Is this correct?  I must be off here, as this seems considerable.


            Any input to my logic errors or math would be appreciated.  If it is really that much time difference, it is a no-brainer for me to spend the extra cash for the faster CPU.

            • 3. Re: Adobe Premier Pro CS5 Build Question
              Harm Millaard Level 7

              Please keep the following in mind:


              When we designed this test, we had to keep the download size of the test files as small as possible, had to make the test as strenuous as possible without people with less powerful systems think that the system gave up on them, without giving up a whole day of work while testing and still give a reasonable reflection on everyday editing. A matter of give and take. You can't meet every desire so as all tests are, this is a compromise between what is realistic, feasible, do-able, relevant and matter-of-fact.


              The timelines are unrealistically heavy in effects, transitions and comprise an unexpected variety of source materials for everyday editing. How often would you see multiple video tracks with a mixture of 29.97 NTSC AVCHD, 25 PAL XDCAM-EX, both 1080 and 720 and SD DV PAL formats intermixed with that load of effects and transitions and made all the more complex with large amounts of bezier curves?


              This was done to keep the download size and time to decent proportions, while at the same time stressing the components quite heavy.


              For comparison reasons this is quite acceptable, but for everyday editing this is far more strenuous that the average timeline. Therefore your argument may not be accurate in terms of expected times.


              BTW, a single point is never a trend, but an observation and from a number of observations one may derive a trend, see "How to lie with statistics".


              Example: Todays temperature at 12:00 was 10 degrees celsius. An observation.


              What is the trend? Will tomorrow be higher or lower and by how much? Care to make a bet that tomorrow it will be below 8 or higher than 12 degrees?


              Had you said that the last 90 days (each of them observations) have shown that the temperature at 12:00 was 0.05 degrees lower each day, that is a trend, even though one may still dispute the statistical relevancy of such a statement.


              Let me take that a step further:


              Statement 1:


              In the past 20 years we have seen that the average life expectancy of males has risen by 3.5 years....


              Statement 2:


              In the past 20 years we have seen that the average sea temperature has risen by 0.6 degrees...


              Can we derive from these statements that the influence of health care has an impact, that medical progress has an impact or that we would all be better off if we heated sea temperatures far more by boiling sea water?


              Sorry to make this into a rather ridiculous statement, but math is not everything, even though series like Numb3rs make you think differently.

              • 4. Re: Adobe Premier Pro CS5 Build Question
                sldwaa Level 1

                I understand completely what you are saying.  The statement about one data point being a trend was supposed to be a joke - I am actually a certified six sigma black belt / electrical engineer so I guess this was my weak attempt at humor.  I understand statistical thinking and data analysis fairly well and know you can twist the facts and numbers to show about anything you desire it to...


                So, all that being said, obviously my math was based on significant assumptions.  My last question is of a more practical nature then I guess....  Can I expect approximately a 2x time difference between the i7-950 and the i7-980x?  I realize the test does tweak the systems pretty hard, but can this logic and time be approximated between the two different CPUs?  I anticipate your response will be "it depends", but if you had to make a gross approximation, is this true?  I am just trying to determine if it is worth the extra cash for the hex vs. quad core.


                Sorry for all this, but I am obviously struggling with which CPU to purchase.

                • 5. Re: Adobe Premier Pro CS5 Build Question
                  ECBowen Most Valuable Participant

                  It is not normally a 2x difference between them. The results vary depending on the amount of layers and effects per layer. There is less difference in performance between the 4 and 6 cores when dealing with 3 layers and and AVCHD. When you add layer 4 though with effects then the difference grows and so on. The meaning behind this though is that the 6 Core chips give you about a 40% to 50% increase in performance at the ceiling and as little as 17 to 20 percent at the lower or average. So that is the performance range you can expect from the 6 cores if you get one. I think that is worth $600 when your editing. Obviously it's not if you are just gaming most of the time or doing something office related. However it can often mean editing non-stop over an hour and exporting for an upload 2 hours before broadcast. I have client's with that tight of a deadline. Then it's worth every dime.




                  1 person found this helpful
                  • 6. Re: Adobe Premier Pro CS5 Build Question
                    sldwaa Level 1

                    Thanks for the info...  If I may, just one more question.  I plan on doing some complicated timelines (several layers) on an occassional basis, however, most of the time I will be doing very simple timelines (from my video camera which outputs *.mts format (MPEG transport stream) when transferred to my hard drive on my PC), and I will be encoding for Blue-ray.  As I am new to Blu-ray, I will experiment with different settings, etc..  Assuming a full 25GB file size, from what you mentioned, can I assume a 20% reduction in time maximum between the two CPUs for these simple encodings?  Also, what is the expected timeline for encoding the entire 25GB?  Sorry for my basic questions, but this is all new to me.