Victory? (was Re: Mac Intel)

Mikael Pettersson <>
Thu Aug 17 14:39:24 CEST 2006


Fredrik Thulin writes:
 > On Thursday 17 August 2006 14:01, Mikael Pettersson wrote:
 > ...
 > > suffix), but the only thing missing is a safety check to see if the
 > > CPU has SSE2 or not before SSE2 FP exceptions are enabled.
 > 
 > Hi
 > 
 > Is that going to be a build-time or run-time check? At Stockholm 
 > university, we regularly build Erlang on one computer, and then use the 
 > resulting binaries on many. Distributions with binary pre-compiled 
 > packages like Debian would also be concerned I guess.

It will be a run-time check on x86-32. Binaries will still run on
older non-SSE2 processors. The only new requirement is that the
build environment must have an assembler that recognises the
stmxcsr and ldmxcsr instructions. (If it doesn't, then the test
in erts/configure will fail, FP exceptions will be disabled, and
HiPE won't work for FP code unless you compile with no_inline_fp.)



More information about the erlang-questions mailing list