Nick, I went ahead and created a JIRA story with the original e-mail of this thread: https://issues.apache.org/jira/browse/LUCY-216
On this ticket you should see 4 attachments: Inversion.sO0 19kb (Without optimization) Inversion.sO0 4kb (Ignore, this was a mistake unable to figure out how to delete from Jira) Inversion.sO1 16kb (With optimization) <https://issues.apache.org/jira/secure/attachment/12519523/Inversion.s-O0O1.diff>Inversion.s-O0O1.diff 4kb The diff has been modified to only include the assembly output of Inversion_invert. Let me know if this works for you. Thanks, Logan On Thu, Mar 22, 2012 at 2:11 AM, Nick Wellnhofer <[email protected]>wrote: > On 22/03/2012 00:15, Marvin Humphrey wrote: > >> The hypothesis is: >> >> * The sum is being accumulated into a register that is larger than 32 >> bits. >> * An extra op is necessary to determine that 32-bit overflow has >> occurred. >> * This op has been inappriately optimized away by the compiler. >> > > Or something with sign extension goes wrong. I can't test this on OS X > myself, but I'd love to have a look at the compiled assembler code. Can you > post it to the list? > > Nick >
