snowdond@r2.cs.man.ac.uk (D.N.Snowdon (MSc PJ)) (02/20/91)
Can anyone tell me what the maximum rate of interrupts generated by the CIA timers is? I have a program that uses a CIA interrupt to cause a soft int - the soft int does the real processing as it can be quite lengthy. All my CIA int code does is to check a flag indicating if the soft int is still executing and if not it Cause's the soft int and exits. I can get up to about 544 interrupts/sec on a PAL A500 however this seems to be independant of the length of code in the CIA int & whether I use a soft in or not. I seem to remember someone saying he had acheived about 2000 CIA ints/sec (albeit with sever CPU loading) but I don't know whether that was using a 68000 or a faster processor.
dillon@overload.Berkeley.CA.US (Matthew Dillon) (02/22/91)
In article <2160@m1.cs.man.ac.uk> snowdond@r2.cs.man.ac.uk (D.N.Snowdon (MSc PJ)) writes: > > Can anyone tell me what the maximum rate of interrupts >generated by the CIA timers is? I have a program that uses >a CIA interrupt to cause a soft int - the soft int does the >real processing as it can be quite lengthy. All my CIA int >code does is to check a flag indicating if the soft int is >still executing and if not it Cause's the soft int and exits. > I can get up to about 544 interrupts/sec on a PAL A500 >however this seems to be independant of the length of code >in the CIA int & whether I use a soft in or not. I seem to >remember someone saying he had acheived about 2000 CIA ints/sec >(albeit with sever CPU loading) but I don't know whether that >was using a 68000 or a faster processor. I can run the FLAG input from the parallel port at aproximately 500uS intervals... any faster and there is no CPU left for any tasks (this is on a standard A500 -- an oscilliscope thingy I made). The actual parallel port interrupt handler doesn't take too much time, at least not looking at it with an oscilliscope (a real one) by doing a BSET at the beginning and a BCLR at the end of an output bit, so I can only assume a lot of overhead in the amiga's CIA interrupt handler code. -Matt -- Matthew Dillon dillon@Overload.Berkeley.CA.US 891 Regal Rd. uunet.uu.net!overload!dillon Berkeley, Ca. 94708 USA