If readers only read, then I would avoid any copying.
How is the data partitioned? Does each reader get a unique piece to work on? Or do all readers read and process all the same data?
gg
If readers only read, then I would avoid any copying.
How is the data partitioned? Does each reader get a unique piece to work on? Or do all readers read and process all the same data?
gg
The problem with working directly on the data would be that the writer process would be beholden to whatever's going on in the sockets code.
It's all the same stuff (it's the display of an SDL application, I am trying to remote a virtual screen).
I really should get round to running some tests on the mechanics of this, just need to get back to my desk