Skip navigation
shuttie27
shuttie27
Currently Being Moderated

FrameMaker 11 very slow to create PDF

Aug 30, 2012 8:15 AM

Hello All

 

Using FrameMaker 11 (TCS4) on Windows Vista SP2 (32-bit), dual-core 2.8GHz, 4GB RAM.

 

I have a book file with nine chapters, title, TOC and Index, adding up to about 550 pages when printed. It used to take about 10 minutes to create PDF using FrameMaker 9. Now that we have upgraded to FM11 it takes over 90 minutes. The chapters contain screenshots as well as text, and I am printing the whole book as one file.

 

I have tried this using Save Book As (then choosing PDF), printing to Adobe PDF as .ps and then distilling, and Save as PDF, all with similar results.

 

Then, because I was working across our network, I copied all the files to my local drive and tried again. Same results. I have also set the printer to not start printing until all pages are spooled.

 

Watching the activity in the printer dialog, I see that pages are spooled quickly (about 1-2 sec/page) if the chapter file is small (up to 40 pages). But in the case of large files (up to 140 pages) each page takes over 10 seconds to spool, and some take 20 seconds. I wondered if this indicates insufficient RAM?

 

I've read lots of useful information about PDF creation from FM, and tried most of the recommendations (including Dov's recommendations for printer/Distiller settings from years back), but nothing makes a difference.

 

Any recommendations are welcome. Thanks.

 

Roger Shuttleworth

London, Canada

940 Views   6 Replies   Latest reply: shuttie27, Sep 4, 2012 1:46 PM
 
Replies
  • Error7103
    Currently Being Moderated
    1. Error7103,
    Aug 30, 2012 8:46 AM   in reply to shuttie27
    Report

    > I wondered if this indicates insufficient RAM?

     

    That would be visible in Task Manager as swapping (paging) activity.

     

    And if you are swapping, the Windoze default is to use a variable size swap file, which can get fragmented, particularly if the HDD is near full, or it's been some time since it was defragged.

     

    Your "4GB RAM", btw, is actually more like 3.5, due to memory allocations in 32-bit Windows kernels. But it "should" be enough, because FM .ps and temp files rarely exceed 2GB (and generally crash older FM versions if they do - I haven't tested FM11 on Win64).

     
    |
    Mark as:
  • shuttie27
    Currently Being Moderated
    2. shuttie27,
    Aug 30, 2012 9:22 AM   in reply to Error7103
    Report

    Hi, thanks for your reply.

     

    I had a look with my IT person at the paging/performance monitors. There is no paging activity going on when generating the PDF, but FrameMaker maxes out the CPU. My HDD has 180 GB of space available, so that is not the problem. A colleague with the same setup has the same experience.

     

    It's worth mentioning that since I am using TCS4 the Acrobat version is now Acrobat X Pro.

     

    Roger

     
    |
    Mark as:
  • Error7103
    Currently Being Moderated
    3. Error7103,
    Aug 30, 2012 10:25 AM   in reply to shuttie27
    Report

    Based on the number of recent complaints about PDF rendering performance, it rather looks like there might be a problem with the newer FM releases. I can't personally say, because we're still on FM7/Unix, and don't use our FM9/Win764 versions for production.

     

    Printing to .ps and Distilling separately should help at least get your FM GUI prompt back sooner, and unless Adobe has broken it, the full Acrobat Pro product can be set to have Distiller "watch" directories for new .ps files, and generate PDFs without any user attention.

     

    It would be interesting to know if the rendering process is multi-threaded to any significant degree. if so, having more cores in your CPU would help.

     

    I'm also wondering if FM11 gets any benefit from installation on Win64.

     

    And at the risk of stimulating conspiracy theories, would using Framemaker Server print any faster? I at least presume that it would free up the workstations during print.

     

    Per-core CPU clocks hit the wall 10 years ago. Per core perf has been creeping up since then, with most of the per-chip gain in the form of more cores. Application coders, however, seem to be assuming that per-core performance is still ramping up at 2x@2yr as a side effect of Moore's Conjecture.

     
    |
    Mark as:
  • Harpreet seehra
    Currently Being Moderated
    4. Harpreet seehra,
    Aug 30, 2012 12:04 PM   in reply to Error7103
    Report

    Hi Shuttie27,

     

    Being with the Framemaker Support  team,i have not seen that happening a lot ,except few cases .

     

    I know this will not make any sense however close outlook and if possible IE or any other browser that's open and print,see if that helps or send your book to me and i will try at my end.

     

    Regards

    Harpreet

     
    |
    Mark as:
  • Error7103
    Currently Being Moderated
    5. Error7103,
    Aug 30, 2012 1:26 PM   in reply to Error7103
    Report

    > ... if so, having more cores in your CPU would help.

     

    In that regard, since single-thread performance probably matters here, if you have an Intel CPU, turn off hyperthreading, if present.

     

    There's a reason why some people call it "hypethreading". It only improves net system throughput in some multi-threading scenarios, and can easily impair single-threaded apps.

     

    I can't test the FM printing scenario for this, because all of my FM machines have only real cores.

     
    |
    Mark as:
  • shuttie27
    Currently Being Moderated
    6. shuttie27,
    Sep 4, 2012 1:46 PM   in reply to Harpreet seehra
    Report

    Hello Harpreet

     

    Thanks for your reply. I don't have Outlook. I turned off Firefox and some other processes, but there was no difference in performance.

    If you send me your email address I will pass the file to you. Thanks for that offer.

     

    Roger

    rshuttleworth -at- avbasesystems -dot- com

     
    |
    Mark as:
Actions

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