Opened 15 years ago

Closed 15 years ago

#63 closed feature (fixed)

Generic wrappers for threading of 1d and 2d functions

Reported by: srkline Owned by: srkline
Priority: minor Milestone: Analysis_4.0_Beta
Component: Analysis Keywords:
Cc: Blocking:
Task: Speed

Description

Threading would be easier if the pertinent information could be sent to generic wrappers (for 1d or 2d calculations). The "test" calculations should be done threaded, but the basic functions that do the fitting should not be threaded, since Fitfunc operation is already multithreaded - and threading threads makes no sense.

so what this means for wrappers and declarations, I don't know yet.

Change History (3)

comment:1 Changed 15 years ago by srkline

  • Summary changed from Gerneric wrappers for threading of 1d and 2d functions to Generic wrappers for threading of 1d and 2d functions

comment:2 Changed 15 years ago by srkline

After thinking a bit about how threading is broken up, I've sort of come to the conclusion that the "internal" function in a threaded function really is best suited as a point calculation, rather than an AAO function. If it's pointwise, then [p1,p2] subranges can be used for each "chunk" of the calculation that is sent to a thread. If it's AAO, then the chunks need to be prepared beforehand (with make?). shelved for now.

comment:3 Changed 15 years ago by srkline

  • Resolution set to fixed
  • Status changed from new to closed

See Cylinder_2D.ipf for an example of how to thread a calculation. A generic wrapper for threading can be pushed off to the future, if at all.

Note: See TracTickets for help on using tickets.