The RED decoder would spin up multiple parallel processes and would use 16, 18, 24 cores as needed to decode/debayer RED r3d media. RED workflows are a great example of this. Some workflows started with high-core-count optimization in Premiere Pro, and moved to the GPU for better performance. Do you send different frames to different cores? How do you ensure they're recombined in the correct order? How much time is wasted splitting the frames up to different cores? There are some tools that devs can use to automate some of this, but then there's a danger of making high core counts mandatory for optimal performance. It may sound silly, but these are real problems when you split up a task like video rendering.
#Adobe cc 2019 multicore how to#
(ever tried to push two carts through a supermarket?) How are the groceries recombined into one cart after the checker is done? How to manage paying for the groceries when 4 checkers are ringing them up simultaneously? Imagine my above example - it complicates how to manage the groceries - more carts means managing that. How that split happens, and recombining the data afterwards, is something that coders need to tackle on a task-by-task basis. In my above analogy, it's like splitting up that big cart of groceries into 2 or more carts, so that more checkers can tackle ringing up the groceries. To do optimization for high core counts, a task has to be split in an intelligent way so that multiple cores can tackle it.
#Adobe cc 2019 multicore full#
However, if one customer has a full cart, still only one checker is used, and that lane slows down while other checkers sit empty. If there are a lot of customers, all the checkers are busy and lines form for each checker. The best analogy I can provide is to picture the checkout counters at a supermarket. Some of these threads may use a core more than other cores, and right now, an "optimal" number of cores is somewhere around the 8-12 mark. Premiere breaks this down with tasks for decoding codecs, applying CPU-driven effects, encoding to specific codecs, etc. Cores are independent CPU processors that can be assigned tasks. What you're asking about is optimization for high core counts, which is different. People who say that Premiere Pro does not support multi-cores don't understand how multi-core support works.