Vantage unable to render, it just stalls....

[alex321]

HI

Just installed Chaos Vantage. I have a 64 gig RAM 2 X NVidia Quadro RTX 4000 ICore 7 Dell Presicion

For some reason whenever I open Chaos Vantage, it is excruciatingly slow.

Here s a screenshot of the default scene. It doesn t render it properly at all so forget my own vrscene.

I know that my Nvidia drivers ( v 472) are more recent than the ones recommended for Vantage . I was wondering if that was the culprit ?

Thanks

Alexandre Galin

[npg]

This is strange indeed. Could you please share the file %APPDATA%/Chaos Group/Vantage/debug_log.txt ? Are you using NVLINK?

[alex321]

I m totally new with Vantage, I do not believe I used NVLink since I opened it as is. I did uncheck Safe Render in the Preferences but it didn’t t change a thing. Once I m back home I ll see that I can share the file with you.
thank you

[alex321]

Here s what the LOG says

Loading scene took 2.83 seconds !

Using 0 lights !

Updating GPU lights took 1 milliseconds to execute.

Copying geometry data to GPU took 40 ms, copy scheduling alone took 10 ms!

Scene deleted in 0.34 seconds.

build schedule create buffers : 11 ms, build schedule command (CPU) : 0 ms, build schedule total : 11 ms

Build schedule : 299 ms; Build command (GPU) : 33 ms

BLAS schedule and build took 333 ms !

Compacting BLAS buffers...

Original BLAS buffers | Dedicated : 124.13 M ( 123.90 M alloced) |

Compacted BLAS buffers | Dedicated : 63.38 M ( 63.14 M alloced) |

Difference | Dedicated : 60.75 M ( 60.76 M alloced) |

Compacting BLAS buffers took 6 ms and saved 60.75 Mbytes on the GPU ( 48.94 % ), required additional 63.38 Mbytes during compaction.

--------------------------------------------------------------------------------------------------------

BLAS tmp mem | Dedicated : 56.00 M ( 54.46 M alloced) | Shared : 640.00 K ( 960.00 alloced) | Total : 56.63 M ( 54.46 M alloced) |

TLAS building took 0 ms !

Serialized shading graph size : 1132 bytes (Tree_A_Bark_Mtl@mtl_brdf_0_bump)

Serialized shading graph size : 436 bytes (Tree_A_Bark_Mtl@mtl_brdf_0)

Serialized shading graph size : 464 bytes (Tree_A_Leaf_01_Mtl@mtl_brdf_2)

Serialized shading graph size : 464 bytes (Tree_A_Leaf_02_Mtl@mtl_brdf_4)

Serialized shading graph size : 464 bytes (Tree_A_Leaf_03_Mtl@mtl_brdf_6)

Serialized shading graph size : 464 bytes (Tree_A_Leaf_04_Mtl@mtl_brdf_

Serialized shading graph size : 340 bytes (DefaultSceneGroundMtl@mtl_brdf_11)

Memory for all shading graphs : 3.68 KBytes.

Shading graph local result slots per thread changed to 15.

Shading graph local result slots per thread changed to 15.

Textures load time : 1.94 seconds, wait time : 0.28 seconds. Loaded 3 8-bit textures, 0 16-bit textures and 0 32-bit textures !

Geometry mem | Dedicated : 84.25 M ( 83.79 M alloced) | Shared : 92.25 M ( 91.33 M alloced) | Total : 176.50 M ( 175.12 M alloced) |

Instance mem | Dedicated : 64.00 K ( 784.00 alloced) | Shared : 64.00 K ( 784.00 alloced) | Total : 128.00 K ( 1.53 K alloced) |

BLAS mem | Dedicated : 63.38 M ( 63.14 M alloced) | Shared : 0.00 ( 0.00 alloced) | Total : 63.38 M ( 63.14 M alloced) |

TLAS mem | Dedicated : 128.00 K ( 3.00 K alloced) | Shared : 64.00 K ( 256.00 alloced) | Total : 192.00 K ( 3.25 K alloced) |

Total geom mem | Dedicated : 147.81 M ( 146.93 M alloced) | Shared : 92.38 M ( 91.33 M alloced) | Total : 240.19 M ( 238.26 M alloced) |

Total gizmo mem| Dedicated : 2.56 M ( 636.61 K alloced) | Shared : 4.25 M ( 354.39 K alloced) | Total : 6.81 M ( 991.00 K alloced) |

Texture mem | Dedicated : 12.50 M ( 12.02 M alloced) | Shared : 0.00 ( 0.00 alloced) | Total : 12.50 M ( 12.02 M alloced) |

Process Dedicated | Budget : 6.99 G, Usage : 1.30 G, Available for reservation : 3.62 G, Current reservation : 0.00

Process Shared | Budget : 31.17 G, Usage : 396.62 M, Available for reservation : 15.71 G, Current reservation : 0.00

--------------------------------------------------------------------------------------------------------

------------------------ Memory Manager State

0.00bytes in 0 free heaps.

Tier ' Forever' | heaps : 1 | free ranges : 2 | total : 64.00 Mbytes | free : 54.13 Mbytes | largest range : 45.19 Mbytes

Tier ' Frame' | heaps : 1 | free ranges : 1 | total : 64.00 Mbytes | free : 63.75 Mbytes | largest range : 63.75 Mbytes

Tier ' Temp0' | heaps : 0 | free ranges : 0 | total : 0.00bytes | free : 0.00bytes | largest range : 0.00bytes

Tier ' Temp1' | heaps : 0 | free ranges : 0 | total : 0.00bytes | free : 0.00bytes | largest range : 0.00bytes

Total managed memory : 128.00 Mbytes !

--------------------------------------------------------------------------------------------------------

Initializing lights' UI took 0.00 seconds.

Textures load time : 4.41 seconds, wait time : 2.49 seconds. Loaded 3 8-bit textures, 0 16-bit textures and 1 32-bit textures !

createImportanceMap took 370 milliseconds to execute.

Startup scene loaded in 8.71s

Starting local HTTP command server on port 20702.

Warning: Disabled path space filtering because it produces artifacts in multi-GPU setups.

Light tree build took 0 milliseconds.

Using 0 lights !

Using 84.00 Mbytes for shadow rays buffer on device 0.

Using 84.00 Mbytes for shadow rays buffer on device 1.

Updating GPU lights took 15 milliseconds to execute.

[npg]

Thanks, but I was also interested in the information at the top of the log which is missing here. It would be simpler to attach the txt file as it's a bit long for posting directly.

Could you try disabling NVLink from the Nvidia control panel and see if it makes any difference? If it's still bad, you can go to "Edit->Render Devices" in Vantage and select only one of the GPUs for rendering (requires Vantage restart). There is some issue with the dual-GPU communication or synchronization, but I don't know what it is.

[alex321]

I went on the Render Devices menu under Edit and selected my twin GPUs and it didn't change a thing. I ll see what can be done on the NVidia control Panel now .

Here's the TXT file in the meantime

thanks for your help

[npg]

OK, now I can see you have "Intel(R) UHD Graphics 630" as well. Is this the GPU your display is connected to? If so, it's could be the reason for the issue.

I went on the Render Devices menu under Edit and selected my twin GPUs and it didn't change a thing.

I meant something else. There are several possible combinations you could use:
* The two RTX linked together - this is what you're using now and it's not working
* The two RTX but without NVLINK, as separate devices
* Only one of the RTX cards (also requires disabling NVLINK first)

So I was suggesting you try the second and then the third option. But first check if the Intel GPU is the culprit.

[alex321]

Yep, I disabled the SLI in the NVidia control panel and now it works just fine

Many thanks Nikola

Alex

[npg]

OK, glad to hear this. We do support NVLINK and it works properly on machines that we have, so maybe there is something specific that breaks it in your case, but we won't be able to debug it until we can reproduce it on our end. Disabling NVLINK will only reduce your maximum possible framerate while using two GPUs, so it's not too bad. If the scene is heavy or the resolution is not very big it won't make much difference.