Skip navigation
This discussion is locked
Currently Being Moderated

Performance discrepancies revealed by PPBM5

Mar 29, 2012 9:55 AM

Looking at Asus x79 based 3930K systems on PPBM5, there are substantial performance discrepancies between systems which look virtually identical, with respect to mobo, RAM, GPU and drives -- in some cases, one system offers half the speed of another.

 

It appears that MPE isn't working properly on some of the underperforming systems, with the performance gain is on the order of 2x, rather than 10-15x, but 

are there any obvious explanations for the differences in performance, since the hardware appears to be much same?

 
Replies
  • Currently Being Moderated
    Mar 29, 2012 12:20 PM   in reply to jamesp2

    Well we cannot of course verify all the results and yes I would guess #22 made some sort of mistake running the test  As you can see no one in that class of machines and Premiere version with a score of 25 is evidently doing something wrong or his computer has some process going on during that test.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 29, 2012 12:45 PM   in reply to jamesp2

    The case you are referring to may be hampered by the 2 disk raid0 SSD configuration, which disables the trim function, and the MPE test is about writing preview files. Writing has never been a strong point with SSD's but when trimming is turned off as well and the preview files are already compressed, it gives all the indicators of SSD limitations, coupled with several other factors that we don't know about.

     

    Obviously it does not bother the person in question, since he never followed up on his results. We do check all the data as submitted and occasionally ask for a rerun of the test and resubmission of the form, because we find inconsistent data or plain wrong results. Those faulty submissions are removed and not entered into the database. Not everybody resubmits his results, even if asked to do so and in that case, we exclude those data.

     

    You can probably imagine that with 800+ systems in the database, it is a rather time consuming task to check all new submissions, correct faulty data and enter them into the database. One of the most common mistakes is the entry about disk size.

     

    It clearly states: In GB for a SINGLE physical disk. Do not add GB to the size. Example: 2000

     

    but for whatever reason people often answer '1.5TB' or '9000' when they have a 6 disk raid5. We all know that there is no single disk with a capacity of 9000 GB and they should have answered 1500, or they answer '931' when it should have been 1000 and not the formatted capacity Windows shows.

     

    We are working on ways to improve the data submission for the hardware questions, so you can get better info.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 30, 2012 6:13 AM   in reply to jamesp2

    and there is a A LOT more to it than just hardware.

    you can have 2 exactly identical systems side by side, one corectly tweaked (included  bios settings, windows tweaks, adobe settings) and one completely incorrect

    and these 2 will benchmark vastly different.

     

    Scott

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 30, 2012 8:21 AM   in reply to Scott Chichelli

    Which is why PPBM5 might be good for tweaking an individual system, helping you find where things can be improved, but is not the best benchmark for determining true hardware performance.

     

    There's just too many variables.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 3, 2012 1:28 AM   in reply to Scott Chichelli

    Scott Chichelli wrote:

     

    ...corectly tweaked (included  bios settings, windows tweaks, adobe settings)...

    Scott, what are the adobe tweaks you're talking of? I'm just interested..

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 3, 2012 8:43 AM   in reply to spreeni

    sorry cant give that..

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 12:42 PM   in reply to spreeni

    Scott's priority is making money.  Ask Harm.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 12:50 PM   in reply to Moose_Breath

    what kind of crap answer is that? and what does that have to do with the post? you have 2 posts? so who are you really..

     

    Eric and i both give out a lot of info and help on these boards, at some point we have to hold back.

    so no i dont say what boards to use or give out certain tweaks.

     

    Scott

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 1:11 PM   in reply to Scott Chichelli

    It’s your know-it-all attitude Scott. You’ve been spoken to about it before by Eric and you need to reign it in. People come to these boards seeking knowledge, and when get this “I know but if I told you I’d have to kill you” nonsense it’s flaming annoying. Stop being so arrogant and join in the community spirit of giving. It will not affect your precious sales figures, it will actually help them. You’re turning a lot of people off.  If I were in a position to purchase a turnkey system, I would not consider ADK based solely on your arrogance. So your secretive tactics are not working. Try and take this onboard.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 1:13 PM   in reply to Moose_Breath

    Mods can we get rid of this guy!

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 1:40 PM   in reply to spreeni

    If tuning were so simple, there would be guides to follow, heck, Microsoft would have made that a DEFAULT on installation, but unfortunately that isn't the case. In the past I have written a guide that can can help you along and even though it says Vista, it also applies to Win7: Adobe Forums: Guide for installing and tuning a Vista... In the FAQ section there are some more articles that can be relevant.

     

    However, there is no 'Holy Grail' to tuning, because it depends to a large degree on your configuration, whether it is a stand-alone system, or part of a domain, how a printer is attached, whether you need to run Exchange Server over a VPN, whether you need TeamViewer for remote support, which Browser you use, etc. etc. so, while I understand the question, I have to agree with Scott, there is no clear-cut answer to your question.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 2:06 PM   in reply to Harm Millaard

    Harm, I didn't ask Scott about windows tuning. I understood his previous post as there are some special settings in Adobe's software which can give more performance. As I wrote, I was just interested to hear ...maybe of hidden switches or something like that. Not a big thing anyway.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 4, 2012 5:56 PM   in reply to jamesp2

    Okay, I'm locking this thread. Personally I think spreeni asked a reasonable question, Moose_Breath's response was not appropriate and Scott over-reacted. Harm gave a clear-cut description of the problem, so maybe that's enough on this.

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points