OctaneRender™ 2020.1 XB3 [superseded by 2020.1 RC1]

[karl]

Hi Otoy Team,

Whenever a scene needs OOC meory, I am getting errors like
"CUDA error 2 on device 0: out of memory"
"Failed to allocate page-locked memory"
"No device pointer for out-of-core memory"
"Render engine failure occurred"

22:14:22.521: Failed to allocate page-locked memory
22:14:22.535: No device pointer for out-of-core memory 00000063359CB060
22:14:22.536: No device pointer for out-of-core memory 00000063399DA060
22:14:22.536: No device pointer for out-of-core memory 0000006104C31060
22:14:22.536: Render engine failure occured! (WARNING)
22:14:25.538: Render engine failure occured! (WARNING)
22:14:28.541: Render engine failure occured! (WARNING)
22:14:31.895: 3 render engine failures within a minute - giving up! (CRITICAL)

This is happening to me since 2019.1.x and continues to happen in 2020.1 XBx... I am using 4 RTX 2080 Ti. I have 128GB of system memory and out of which I have allocated around 18GB of system memory to OOC Memory. I still get the above errors. Scenes which used to load and render and use OOC memory in 2018.x (on the same workstation) will load but no longer render in 2019.1.x and 2020.1 XBx for me. I always get the above errors.

This has been also reported before.

In terms of system info and driver info, they are in my signature.

Regards,
Rajib

Hi Rajib,

It's hard to tell what's causing the issue; there could be a few different things. There are some minor improvements to the out-of-core feature in the next version and they may help. I've also improved the logging so we can see an error code with the "Failed to allocate page-locked memory" message which should help to narrow things down if you can still reproduce this issue. Please let us know if you still run into this issue with the next version.

[karl]

Crashes sometimes mostly during when the rendering starts. Plugin version just quits itself so no log is created. In Standalone following error is displayed:

Code: Select all

CUDA error 700 on device 0: an illegal memory access was encountered
  -> failed to load symbol data to the device(kernel data)

Hi J.C,

Do you have an example scene that exhibits this issue that you are able to share? We would need to reproduce this issue in order to gain more information about the cause.

[karl]

Auto-focus and camera focus picker don't in universal camera when astigmatism is turned on.

Hi galleon27, this issue has been fixed for the next release. Thanks for the report.

[rajib]

Hi Otoy Team,

Whenever a scene needs OOC meory, I am getting errors like
"CUDA error 2 on device 0: out of memory"
"Failed to allocate page-locked memory"
"No device pointer for out-of-core memory"
"Render engine failure occurred"

22:14:22.521: Failed to allocate page-locked memory
22:14:22.535: No device pointer for out-of-core memory 00000063359CB060
22:14:22.536: No device pointer for out-of-core memory 00000063399DA060
22:14:22.536: No device pointer for out-of-core memory 0000006104C31060
22:14:22.536: Render engine failure occured! (WARNING)
22:14:25.538: Render engine failure occured! (WARNING)
22:14:28.541: Render engine failure occured! (WARNING)
22:14:31.895: 3 render engine failures within a minute - giving up! (CRITICAL)

This is happening to me since 2019.1.x and continues to happen in 2020.1 XBx... I am using 4 RTX 2080 Ti. I have 128GB of system memory and out of which I have allocated around 18GB of system memory to OOC Memory. I still get the above errors. Scenes which used to load and render and use OOC memory in 2018.x (on the same workstation) will load but no longer render in 2019.1.x and 2020.1 XBx for me. I always get the above errors.

This has been also reported before.

In terms of system info and driver info, they are in my signature.

Regards,
Rajib

Hi Rajib,

It's hard to tell what's causing the issue; there could be a few different things. There are some minor improvements to the out-of-core feature in the next version and they may help. I've also improved the logging so we can see an error code with the "Failed to allocate page-locked memory" message which should help to narrow things down if you can still reproduce this issue. Please let us know if you still run into this issue with the next version.

Thank you Karu for the update. Yes, I can easily produce the error. Will wait for the next version and then provide you with the updated log. Hopefully the improved logging will help you all to narrow down the issue and hopefully the improvements being done to the OOC memory will also help to get rid of the error. Much appreciated.

[J.C]

Crashes sometimes mostly during when the rendering starts. Plugin version just quits itself so no log is created. In Standalone following error is displayed:

Code: Select all

CUDA error 700 on device 0: an illegal memory access was encountered
  -> failed to load symbol data to the device(kernel data)

Hi J.C,

Do you have an example scene that exhibits this issue that you are able to share? We would need to reproduce this issue in order to gain more information about the cause.

Unfortunately not. It happens randomly with most of the scenes I tested. It often happens when I switch from RTX to non RTX acceleration.

[GillesKontrol]

Hi Otoy Team,

Whenever a scene needs OOC meory, I am getting errors like
"CUDA error 2 on device 0: out of memory"
"Failed to allocate page-locked memory"
"No device pointer for out-of-core memory"
"Render engine failure occurred"

22:14:22.521: Failed to allocate page-locked memory
22:14:22.535: No device pointer for out-of-core memory 00000063359CB060
22:14:22.536: No device pointer for out-of-core memory 00000063399DA060
22:14:22.536: No device pointer for out-of-core memory 0000006104C31060
22:14:22.536: Render engine failure occured! (WARNING)
22:14:25.538: Render engine failure occured! (WARNING)
22:14:28.541: Render engine failure occured! (WARNING)
22:14:31.895: 3 render engine failures within a minute - giving up! (CRITICAL)

This is happening to me since 2019.1.x and continues to happen in 2020.1 XBx... I am using 4 RTX 2080 Ti. I have 128GB of system memory and out of which I have allocated around 18GB of system memory to OOC Memory. I still get the above errors. Scenes which used to load and render and use OOC memory in 2018.x (on the same workstation) will load but no longer render in 2019.1.x and 2020.1 XBx for me. I always get the above errors.

This has been also reported before.

In terms of system info and driver info, they are in my signature.

Regards,
Rajib

Hi Rajib,

It's hard to tell what's causing the issue; there could be a few different things. There are some minor improvements to the out-of-core feature in the next version and they may help. I've also improved the logging so we can see an error code with the "Failed to allocate page-locked memory" message which should help to narrow things down if you can still reproduce this issue. Please let us know if you still run into this issue with the next version.

I've been getting this in almost all scenes which skirt with the edges of in core and out of core memory. That and semi transparant geometry or things just not loading in (or too late).

2019 as well as 2020. 2080ti & 1080ti (and 2070's on the renderrig but they can't even load in most of it because Out of Core just gives the page locked error all the time).

[enricocerica]

Hi,
Just noticed that after using the refresh scatters script, I cannot start any render anymore, no error but when selecting a render target or use the restart current render button, nothing happens.

[funk]

I've been trying to use Random walk on a volume and it just hides the volume.

Random walk does have "volume step length" and "ray step length" pins, so I assume it's meant to wok on volumes.

1. Load the orbx and start rendering. You'll see a cube volume.
2. Unplug the volume medium, and connect the random walk to the "cube.vdb"

The volume stops rendering

[enricocerica]

Hi,
Just noticed that after using the refresh scatters script, I cannot start any render anymore, no error but when selecting a render target or use the restart current render button, nothing happens.

Well, apparently nothing to do with the refresh scatter script. It occurs randomly, at least as I may observe. At one moment, for instance if I add a mesh, or I connect a mesh to the render target I not able anymore to start a rendering. I should stop and restart Octane.
Any idea how to avoid this ?
Thanks

Edit : I have exactly the same issue with the 2019.1.5 version ...

Edit : Ok, did some more scene analysis and finally discovered that the problem seems to come from an instantiated object. I have a grass scattering with about 2.500.000 instances. If I remove it from the scene, the problem does not occur. As soon as I add it again, the UI doesn't respond anymore. The rendering seems to continue where it freezes but the UI is not updated and I couldn't do anything else then restart Octane. I tried to reduce the amount of instances to 1.700.000 and got the same issue and again after reducing to 1.000.000.

[funk]

BUG: Displacement autobump seems to break on scaled items

1. Load the orbx and select the rendertarget to start rendering
2. Select the DISP_MAT nodegraph and toggle displacement > autobump off/on to see the difference