Distributed Rendering - should I be transferring assets?

[tricky]

I've not really bothered with this feature before as I always ensure that everything is asset tracked correctly (UNC and all that). However, I'm wondering if, by setting the limit to 5GB or so, would I be getting an improvement in render completion speed after the first render as all the assets will already be stored in some sort of cache in each render node?

[Macker]

My understanding is that it only transfers missing assets. If you have 5GB of missing assets, then you have some problems.

[vlado]

However, I'm wondering if, by setting the limit to 5GB or so, would I be getting an improvement in render completion speed after the first render as all the assets will already be stored in some sort of cache in each render node?

Like with most things, you'll need to test it and see how it goes. The primary goal of the feature is convenience rather than render speed although in some cases caching things like proxies locally on each render server might improve render speed and make the loading of bitmaps at the start of a render go faster. But Windows is also pretty good at caching network assets in memory, so the effect on render speed might not be that large.

For me it is convenient as I can just pick textures off of my hard drive and not worry about copying them on a network location, then making sure that all servers can access that location etc. But if you have your render farm already set up with a good workflow, it may not matter that much.

Best regards,
Vlado

[tricky]

Thanks Vlado. I didn't realise Windows did network caching itself. All useful information to add to my 'VRay brain'.