site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com On 10.3.6 a process timing non-system using first "do": 400G4 129sec 350G3 118sec Thanks for listening and hope you can direct me Steve <tauvan@starpower.net> Please don't say "Apple" because I already know their response: "Buy a new computer". _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... Hi, I'm looking for the correct list for a processor-specific problem. This is code ran on MacOSX, 10.1.5 and 10.3.6, a PCI G4 computer. The reason it is processor- specific is that a G3 building the same code runs faster than the faster(Mhz) G4. I think it may be the way the OS sets up for simple math, then again it could be the compiler setting simple math for an alti-vec processor. The 10.3.6 base timings for a given system processes are: 400G4 178sec 350G3 176sec 400G4 100sec 350G3 99sec 400G4 188sec 350G3 186sec What caught my eye was: "do *--strPtr = ((val_ul & 7) + 48); while ((val_ul >>= 3));" ran 2sec slower than "do *--strPtr = _lowerhexstr[(val_ul & 7)]; while ((val_ul >>= 3));" on 400G4 10.1.5 but on the G3 it ran 5sec faster. Earlier system tests of 300G3(not typo) to 400G4 both under 10.1.5 showed times like 400G4 76sec 300G3 102sec 400G4 103sec 300G3 141sec This email sent to site_archiver@lists.apple.com