Results 1 to 5 of 5

Thread: Tracing bottlenecks

  1. #1
    Join Date
    Dec 2017
    Posts
    116

    Default Tracing bottlenecks

    What should I look at in the log files--if it's even useful--to try to figure out where the bottlenecks are for the initial loading of a scene? I have a model where it can take around 50 seconds from activating the view mode, which seems excessive, though some of it has to do with the overall size of the file as opposed to just what I'm rendering. This is the log file up to where I first see something , and but I don't see any hints as to where it's taking up its time?

    1.0 IRAY rend info : Updating geometry.
    1.0 IRAY rend info : Using OptiX Prime ray tracing (5.0.1).
    1.0 IRAY rend info : Importing scene graph.
    1.0 IRAY rend info : Importing geometry for motion time 0
    1.0 IRAY rend info : Geometry import (8256 objects with 2719k triangles, 8277 instances yielding 2786k triangles) took 0.673587
    1.0 IRAY rend info : Updating materials.
    1.0 MATCNV rend info : found 34 textures, 0 lambdas (0 unique)
    1.0 IRAY rend info : Updating environment.
    1.0 IRAY rend info : Updating lights.
    1.0 IRAY rend info : Updating object flags.
    1.0 IRAY rend info : Updating caustic portals.
    1.0 IRAY rend info : Updating decals.
    1.0 IRAY rend info : Allocating 1-layer frame buffer
    1.0 IRAY rend info : Using interactive scheduling, architectural sampler unavailable, caustic sampler disabled
    1.0 IRAY rend info : Initializing local rendering.
    1.0 IRAY rend info : Rendering with 4 device(s):
    1.0 IRAY rend info : CUDA device 3 (GeForce GTX 1080 Ti)
    1.0 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti)
    1.0 IRAY rend info : CUDA device 1 (GeForce GTX 1080 Ti)
    1.0 IRAY rend info : CUDA device 2 (GeForce GTX 1080 Ti)
    1.0 IRAY rend info : Rendering...
    1.3 IRAY rend progr: CUDA device 0 (GeForce GTX 1080 Ti): Processing scene...
    1.2 IRAY rend progr: CUDA device 3 (GeForce GTX 1080 Ti): Processing scene...
    1.13 IRAY rend progr: CUDA device 2 (GeForce GTX 1080 Ti): Processing scene...
    1.4 IRAY rend progr: CUDA device 1 (GeForce GTX 1080 Ti): Processing scene...
    1.6 IRAY rend stat : Geometry memory consumption: 94.7226 MiB (device 3), 0 B (host)
    1.8 IRAY rend stat : Geometry memory consumption: 94.7226 MiB (device 0), 0 B (host)
    1.11 IRAY rend stat : Geometry memory consumption: 94.7226 MiB (device 1), 0 B (host)
    1.9 IRAY rend stat : Geometry memory consumption: 94.7226 MiB (device 2), 0 B (host)
    1.6 IRAY rend stat : Texture memory consumption: 203.25 MiB for 15 bitmaps (device 3)
    1.6 IRAY rend info : Importing lights for motion time 0
    1.6 IRAY rend info : Initializing light hierarchy.
    1.6 IRAY rend info : Light hierarchy initialization took 0.00s
    1.6 IRAY rend stat : Lights memory consumption: 196 B (device 3)
    1.6 IRAY rend stat : Material measurement memory consumption: 0 B (GPU)
    1.6 IRAY rend stat : Materials memory consumption: 227.758 KiB (GPU)
    1.8 IRAY rend stat : Texture memory consumption: 203.25 MiB for 15 bitmaps (device 0)
    1.8 IRAY rend stat : Lights memory consumption: 196 B (device 0)
    1.3 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti): Scene processed in 49.837s
    1.8 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti): Allocated 4.07866 MiB for frame buffer
    1.3 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti): Allocated 1.6875 GiB of work space (2048k active samples in 0.000s)
    1.2 IRAY rend info : CUDA device 3 (GeForce GTX 1080 Ti): Scene processed in 49.843s
    1.6 IRAY rend info : CUDA device 3 (GeForce GTX 1080 Ti): Allocated 4.07866 MiB for frame buffer
    1.2 IRAY rend info : CUDA device 3 (GeForce GTX 1080 Ti): Allocated 1.6875 GiB of work space (2048k active samples in 0.000s)
    1.2 IRAY rend info : CUDA device 3 (GeForce GTX 1080 Ti): Used for display, optimizing for interactive usage (performance could be sacrificed)
    1.11 IRAY rend stat : Texture memory consumption: 203.25 MiB for 15 bitmaps (device 1)
    1.11 IRAY rend stat : Lights memory consumption: 196 B (device 1)
    1.4 IRAY rend info : CUDA device 1 (GeForce GTX 1080 Ti): Scene processed in 50.070s
    1.8 IRAY rend info : CUDA device 1 (GeForce GTX 1080 Ti): Allocated 4.07866 MiB for frame buffer
    1.4 IRAY rend info : CUDA device 1 (GeForce GTX 1080 Ti): Allocated 1.6875 GiB of work space (2048k active samples in 0.000s)
    1.9 IRAY rend stat : Texture memory consumption: 203.25 MiB for 15 bitmaps (device 2)
    1.9 IRAY rend stat : Lights memory consumption: 196 B (device 2)
    1.13 IRAY rend info : CUDA device 2 (GeForce GTX 1080 Ti): Scene processed in 50.658s
    1.8 IRAY rend info : CUDA device 2 (GeForce GTX 1080 Ti): Allocated 4.07866 MiB for frame buffer
    1.13 IRAY rend info : CUDA device 2 (GeForce GTX 1080 Ti): Allocated 1.6875 GiB of work space (2048k active samples in 0.000s)
    1.0 IRAY rend info : Received update to 00001 iterations after 50.726s.
    1.0 IRAY rend info : Rendering...

  2. #2
    Join Date
    Dec 2017
    Location
    Melbourne, Australia
    Posts
    302

    Default

    So you can see where almost all the time is going in this line:

    Code:
    1.3 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti): Scene processed in 49.837s
    When I see this the first place I look is the organisation of the geometry. If you look closer to the start you have this line:

    Code:
    1.0 IRAY rend info : Geometry import (8256 objects with 2719k triangles, 8277 instances yielding 2786k triangles) took 0.673587
    So you have only slightly less objects than instances, so basically everything is it's own instance (so most objects are unique). Given this you may want to ensure instancing is disabled in the options, otherwise each object will potentially need its own acceleration structure, slowing things down. Aside from there the scene appears to contain 2.8M triangles. While not a huge amount in itself, since that is split among a large number of objects it is going to increase processing time for sure. If it's evenly distributed that's only a few hundred triangles per object in this scene.

    In any case, the first thing I would suggest trying is disabling instancing and seeing if that drops the pre-processing time. This scene makes little to no use of instancing anyway. You can also try with and without OptiX Prime to see if that results in a difference in processing time.

  3. #3
    Join Date
    Dec 2017
    Posts
    116

    Default

    Wow, disabling instancing sped it up massively. 0.369s! Disabling Optix Prime was a few seconds.

    Thanks!

  4. #4

    Default

    We have the same problem.

    The "pre-processing time" for scenes with large number of instances is very slow.

    I attach a test with a GRID of 1000 trees (1.2 million polygons each). It takes 6 minutes to start rendering in Iray and 5 seconds in FStrom (similar in Octane render).

    This problem prevents using Iray for exterior scenes with large instantiated geometry content.

    Our company has been a user of Iray in the past and we are currently studying this option again, since in general the behavior of the engine seems perfect, but due to this problem it is impossible to return to Iray.

    Is there any solution?

    Click image for larger version. 

Name:	IRAY_SlowTranslatingInstance_low.jpg 
Views:	3 
Size:	305.3 KB 
ID:	10656
    Click image for larger version. 

Name:	FSTORM_SlowTranslatingInstance_low.jpg 
Views:	3 
Size:	220.5 KB 
ID:	10657

  5. #5
    Join Date
    Dec 2017
    Location
    Melbourne, Australia
    Posts
    302

    Default

    Iím a bit confused, your screenshots are all of Iray for 3ds max where as this is the Iray for Rhino section of the forum. If your issues are in Iray for 3ds max perhaps move the post over to that forum.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •