Answering my own question: it seems part of the issue is that the input image needs BlackOutside - I think I knew that from ages ago? We are doing a per pixel filtering operation (like Convolve), so that makes some sense. Also, I know there were some nice BlinkScript bug fixes in 10.0v4 so maybe there is a vast performance difference in certain situations.
On 26 October 2016 at 11:28, Michael Garrett <michaeld...@gmail.com> wrote: > Hi, just wondering if anyone has any suggestions as to why a BlinkScript > may be working fine on a Quadro K1200 on a cheap Linux box, but take hours > on a Quadro K600 and 600 (mixed environment) on Windows. Trying to remotely > troubleshoot this. > > I'm getting a draw speed of about a second or two for a 2K frame in Nuke > 10.0v4 on Linux with the K1200 whereas the Windows setup in Nuke 10.0v1 > the other guys are using is exponentially slower. It seems the GPU they are > using should "only" be about 4 times slower which is still very workable. > > I'll send a ticket to the Foundry with an example, but it's going to take > them a while to get back to me. > > Thanks, > Michael >
_______________________________________________ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users