AWS Thinkbox Discussion Forums

Memory Leak? Deadline?

I have a scene when rendered locally consumes about 4 GB of Ram. When I send it to Deadline it errors out, having peaked all ram with an alloc error:

[code]=======================================================
Error

Error: RenderTask: Unexpected exception (MAXException caught in 3ds max: BuildMesh (Frost001): raw_byte_buffer.add_element: Failed to realloc 18253611023 bytes of memory.
2016/07/09 13:35:02 INF: [29712] [12516] SYSTEM: Production renderer is changed to Default Scanline Renderer. Previous messages are cleared.
2016/07/09 13:35:09 DBG: [29712] [12516] Starting network
[/code]

I just noticed

2016/07/09 13:35:02 INF: [29712] [12516] SYSTEM: Production renderer is changed to Default Scanline Renderer. Previous messages are cleared.

If not using instancing, it would certainly explain the memory issue. What would cause it to change to DSR? The submission accurately has the vray settings.

Job_2016-07-09_13-48-50_57814714024bf25fa8ade84b.txt (68.6 KB)

I don’t think the message means you are not rendering in V-Ray. But it appears that Frost in Slave mode is building a TriMesh and not performing instancing. This could happen in Frost does not load the Frost_VRay DLL correctly - I have experienced this even in workstation mode with some early builds.

We will try to reproduce this on Monday. However, we now have a newer internal build (which is probably gonna blow your mind :wink: ) and it might behave differently. I will test both builds and see if either one renders on Deadline.

Thank you–I look forward to having my mind blown.

Thank you for your report! This issue should be fixed in Frost 2.0.2.

Privacy | Site terms | Cookie preferences