Menezes, Evandro - RE: [PATCH] Allow cmpxchg16b even with -march=core2 (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: "Menezes, Evandro"
• To: "Jakub Jelinek"
• Cc: "Jan Hubicka" , gcc-patches at gcc dot gnu dot org
• Date: Thu, 1 Feb 2007 09:54:38 -0600
• Subject: RE: [PATCH] Allow cmpxchg16b even with -march=core2
• References: 20070117154851.GA29911@devserv.devel.redhat.com 20070117160911.GJ21319@atrey.karlin.mff.cuni.cz 20070118213756.GH29911@devserv.devel.redhat.com 20070118214807.GL21319@atrey.karlin.mff.cuni.cz <1449F58C868D8D4E9C72945771150BDF0149A2B6@SAUSEXMB1.amd.com> <20070201152549.GI27294@devserv.devel.redhat.com>

Jakub,

Given that CMPXCHG16 presence is indicated by CPUID and no currently existing AMD processor moniker exists in GCC that could be used to make it implicit, I'd suggest adding an -m option to indicate this feature.

Not even amdfam10 (noticed in the latest patches it wasn't setting PTA_CX16)? I guess -mcx16 wouldn't hurt though.

Yes, it can be assumed in amdfam10, but I didn't count it as current yet. ;-)

But -mcx16 would be cool.

Thanks,

Evandro Menezes GNU Tools Team 512-602-9940 Advanced Micro Devices evandro.menezes@amd.com Austin, TX

• References:
  • RE: [PATCH] Allow cmpxchg16b even with -march=core2
    * From: Menezes, Evandro
  • Re: [PATCH] Allow cmpxchg16b even with -march=core2
    * From: Jakub Jelinek