You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Although the copy itself (or write to BSA, as in #73) is going to be strictly I/O bound, I believe that for files in BSAs, there is some non-trivial overhead and that parallel reads could lead to some CPU efficiency gains.
It has to be tested, but one user mentioned preferring loose files for a different tool because the build ran 20-30% faster that way. I think BSAs can probably be just as fast - there's really no reason they shouldn't be since it's just reading the same number of bytes or even fewer bytes (if compressed) from an I/O stream, and there's less overhead opening and closing the files. So any additional overhead coming from BSA extraction must be in things like decompression, which can be done concurrently on most modern rigs.
The text was updated successfully, but these errors were encountered:
Intent is to have a loosely-coupled arrangement of tasks, with a pipeline system to handle all the scheduling and dependencies. This will allow us to take some of the extremely complex tasks (like the former `MutagenMergedPluginBuilder` and `MergedFolder`) and turn them into manageable units of work, and maybe improve parallelism.
Currently not tested at all, there are probably several bugs in here, but as yet none of this is linked to the main app. The initial graph does assemble with Autofac.
#26#37#64#74
Although the copy itself (or write to BSA, as in #73) is going to be strictly I/O bound, I believe that for files in BSAs, there is some non-trivial overhead and that parallel reads could lead to some CPU efficiency gains.
It has to be tested, but one user mentioned preferring loose files for a different tool because the build ran 20-30% faster that way. I think BSAs can probably be just as fast - there's really no reason they shouldn't be since it's just reading the same number of bytes or even fewer bytes (if compressed) from an I/O stream, and there's less overhead opening and closing the files. So any additional overhead coming from BSA extraction must be in things like decompression, which can be done concurrently on most modern rigs.
The text was updated successfully, but these errors were encountered: