Hi guys – we’re pretty stoked to tell you that we have branch-based multi-threading working in the Maya graph. Our expectation is that we will also be able to do this in Softimage, so we're pretty excited - we have to get through Siggraph before we can start on it, but it should only be a week or two of work.
https://vimeo.com/69000004 “This video demonstrates a feature of our Creation Core technology: performing multi-threading across branches of graphs. Creation: Splice can use this technology inside of Maya and perform multi-threading across branches in the Maya dependency graph. This is extremely useful when the computation per node is rather small, like in a character rig. Parts of the rig can be computed in parallel using this approach, and the overall performance of the graph is improved.” Very cool to see this working (in my unbiased opinion). Paul On 22 June 2013 16:14, Paul Doyle <[email protected]> wrote: > > On 22 June 2013 15:08, Tim Leydecker <[email protected]> wrote: > >> I guess the reason why the general plattform is less easy to sell >> than the modules is that it needs a bit more abstraction to see >> the benefit of investing into it in comparison to a module that >> either already does solve a specific problem or seems like a good >> basis to start modifications off from. >> >> It requires a lot of faith to start from scratch. >> > > Agreed - the problem with a platform is you have to search for > problems/justifications to start using it. What we found was the complexity > of getting started was a bit much for people that initially take a look in > their spare time - it's rare to get a studio that allocates time to > investigating new technology, and when they do, you really need to get them > somewhere valuable quickly. The challenge we had was that the investment > required to get somewhere useful with Creation kind of meant that the value > of the solutions had to be much more than just a simple tool, which meant > more time was required, etc etc ad infinitum > > Our aim with Splice is to get people to see immediate value with our core > - then they can either continue to exploit it within their existing tools, > or they can move to Creation Platform and start building more complete > applications. I think R&D teams are more likely to get sign off on > something like this if they have already demonstrated the core performance > capability elsewhere. It also helps establish trust in the company with > relatively low risk - "let's try and write a KL deformer in Maya and see > what we think" is a lot less involved than "let's implement our custom data > type in their real time renderer". > > There are a few companies that immediately 'got' Creation Platform - > generally because we either knew them well already, or they had a similar > design philosophy with their existing tools. The cool thing is they are > starting to push on Creation in interesting ways, so I'm looking forward to > some case studies from them later this year. It's pretty awesome to see > people building things we didn't even think of yet! > > If things continue to go well I promise to give Fabric dunce caps to those > that want them :) > > >
