I have a particular scene (product rendering) where using a DR slave with 1x RTX 3090 GPU yields just 32% speedup (151 seconds vs 200 seconds) compared to using just the RTX 3090 in my workstation. Increasing the "Rays Per Pixel" and "Rays Bundle Size" didn't seem to help. Also, according to Hwinfo64, the second RTX 3090 was working at full power during rendering. I did a test with an interior scene and it was behaving more in line with my expectations of being around 2x faster - 6 minutes vs 11 minutes for that particular interior scene.
Why would the speed up be so low in one scene and as expected in another? Are such wide discrepancies normal with distributed GPU rendering? Both my workstation and the DR server have the recommended Nvidia driver installed - 466.47 and the latest versions of Vray and 3ds Max which are also matching between both machines.
Why would the speed up be so low in one scene and as expected in another? Are such wide discrepancies normal with distributed GPU rendering? Both my workstation and the DR server have the recommended Nvidia driver installed - 466.47 and the latest versions of Vray and 3ds Max which are also matching between both machines.
Comment