Some optimization opportunities

[Marcos Dione]

On Sat, Jun 19, 2004 at 11:02:07AM +0200, Daniël Mantione wrote:
> Another observation:
> 
> 32761 15:36:42.449984 gettimeofday({1087565802, 450171}, NULL) = 0
> 32761 15:36:42.450714 gettimeofday({1087565802, 450905}, NULL) = 0
> 32761 15:36:42.451667 gettimeofday({1087565802, 451861}, NULL) = 0
> 32761 15:36:42.452777 gettimeofday({1087565802, 452980}, NULL) = 0
> 32761 15:36:42.453542 gettimeofday({1087565802, 453736}, NULL) = 0
> 32761 15:36:42.454275 gettimeofday({1087565802, 454467}, NULL) = 0
> 32761 15:36:42.455007 gettimeofday({1087565802, 455198}, NULL) = 0
> 32761 15:36:42.455732 gettimeofday({1087565802, 455924}, NULL) = 0
> 32761 15:36:42.456456 gettimeofday({1087565802, 456650}, NULL) = 0
> 32761 15:36:42.457184 gettimeofday({1087565802, 457373}, NULL) = 0
> 32761 15:36:42.457913 gettimeofday({1087565802, 458105}, NULL) = 0
> 32761 15:36:42.458727 gettimeofday({1087565802, 458925}, NULL) = 0
> 32761 15:36:42.459436 ioctl(5, FIONREAD, [0]) = 0
> 32761 15:36:42.459841 write(6, "\0", 1) = 1
> 32761 15:36:42.460624 gettimeofday({1087565802, 460817}, NULL) = 0
> 32761 15:36:42.461707 gettimeofday({1087565802, 461913}, NULL) = 0
> 32761 15:36:42.462460 gettimeofday({1087565802, 462654}, NULL) = 0
> 32761 15:36:42.463461 gettimeofday({1087565802, 463667}, NULL) = 0
> 32761 15:36:42.464204 gettimeofday({1087565802, 464393}, NULL) = 0
> 32761 15:36:42.465155 gettimeofday({1087565802, 465353}, NULL) = 0
> 32761 15:36:42.466002 gettimeofday({1087565802, 466196}, NULL) = 0
> 32761 15:36:42.466913 gettimeofday({1087565802, 467113}, NULL) = 0
> 32761 15:36:42.467762 gettimeofday({1087565802, 467961}, NULL) = 0
> 32761 15:36:42.468747 gettimeofday({1087565802, 468946}, NULL) = 0
> 32761 15:36:42.469698 gettimeofday({1087565802, 469896}, NULL) = 0
> 32761 15:36:42.470562 gettimeofday({1087565802, 470756}, NULL) = 0
> 32761 15:36:42.475218 gettimeofday({1087565802, 475440}, NULL) = 0
> 32761 15:36:42.475899 gettimeofday({1087565802, 476089}, NULL) = 0
> 32761 15:36:42.476416 gettimeofday({1087565802, 476590}, NULL) = 0
> 32761 15:36:42.477673 gettimeofday({1087565802, 478056}, NULL) = 0
> 32761 15:36:42.479127 gettimeofday({1087565802, 479343}, NULL) = 0
> 32761 15:36:42.479735 gettimeofday({1087565802, 479921}, NULL) = 0
> 32761 15:36:42.480452 gettimeofday({1087565802, 480642}, NULL) = 0
> 32761 15:36:42.482380 gettimeofday({1087565802, 482612}, NULL) = 0
> 32761 15:36:42.482945 gettimeofday({1087565802, 483115}, NULL) = 0
> 32761 15:36:42.483548 gettimeofday({1087565802, 483726}, NULL) = 0
> 32761 15:36:42.484345 gettimeofday({1087565802, 484540}, NULL) = 0
> 
> Calling gettimeofday this often seems a little strange to me and certainly
> an important part of the 35 ms here is wasted due to system call overhead.

  well, here you're assuming that *nothing* is being done between gettimeofday()
calls. and as Maks pointed, strace *really* slows things a lot.

-- 
(Not so) Random fortune:
http://www.grulic.org.ar/~mdione/