Display Performance with Reference Clip Masks? - z_[Archived] MicroStation V8 XM Edition Forum - MicroStation - Bentley Communities

Display Performance with Reference Clip Masks?


I love the Direct3D-based graphics display system in MicroStation, and I have several questions about it.  These questions are most applicable to to MicroStation XM, since that is the version I use most, but I have seen this behavior on MicroStation V8i also. 

We are in the process of moving to the NVIDIA GTX260 as the Graphics card of choice for our workstations, since it is a pretty quick D3D card and represents a good bang-for-the-buck.  However, even on the  PCs that have that card, we are noticing a slowdown on references with clip masks (clip boundaries don't seem as affected); the slowdown is all the more noticeable on older grapics cards. 

We can have pretty much as many references as we want without clipping and navigate (that is, pan and zoom within the file) quickly and smoothly.  Adding a rectangular clip mask or two adds a minimal but barely noticeable slowdown to navigating the file.  The more clip masks, the slower navigation becomes, bit by bit.  However, using a fence by circle and making a singlwe clip mask out of that adds a definite hit - it becomes as slow to pan and zoom as it was in V8, if not more so. 

When the move from V7 to V8 was made, the number of reference clip points available for use increased from 101 to 5000.  I am wondering if a really large number of clip points will reduce navigation performance, even on a card as powerful as the GTX 260.  A rectangular clip has... 2?  4?  Maybe 5 clipping points total?  How many points does a circular clip mask create?  Does a larger size circular clip mask mean it creates more clipping points?  For now, the workflow of one of our departments means circular clip masks are needed.  Might there be a workaround to this? 

Another question: I tried to upgrade the driver of one of the machnes from NVIDIA's Sept 2008 driver to the Driver released in Aug 2009.  Performance actually decreased when i made this driver switch - enough so that I reverted back to the older driver.  Has anyone heard of this or have an idea why that might be? 

Any ideas or information are appreciated!


5 Replies (Most Recent Reply)

  • jcallicott:

     I am wondering if a really large number of clip points will reduce navigation performance, even on a card as powerful as the GTX 260.... 

    Yes, this does occur in XM; however we've optimized this greatly in V8i.  I believe the best answer to this problem is upgrade.




  • Dear Jeff,

     We are using V8i and suspect a circular clip still has some performance issues;

    Her are some notes about clip regons;

    (1) Polygons (MOVE TO, lINE TO)  any number of segments is not a problem the maths for the intersection is easy.

    (2) Circles are more difficult hence more proccessing time for the maths

    (3) Clouds - do not use clouds as clip boundary these are bspline segments the maths is realy difficult and has a serious performance hit (you will really have to wait) - our users have been instructed not to do this any more...

    (4) Polygon clip with many circular or B-Spline elements in the drawing is likley to have a similar performane issue to (2) and (3)

    Buying a better graphics card will not help - it is pure maths for the element intersections that is causing the problem not the display speed (eventhough your drawing appears on the screen slowly) .

    Best Regards,

    Ian Emery.

  • In reply to Ian Emery:

    I am having the exact same problem except we are not even using circles for our clip mask.  We are only using polylines.  We can open the same drawing the 2004 V8 and it works fine.  Is there a solution other than an upgrade to v8i to solve this problem as upgrading is not a viable solution for the project I am working on. 



  • In reply to Josh Cook:

    V8i did not improve the performance in my case.

    V8i was slower.

    The v8i and XM systems were very different hardware and software wise but the v8i was much slower.

    I will load v8i and xm together and retest. 


^ Go to first post