[gmx-users] Re : Gromacs-3.3: release candidate 1 available!
lindahl at sbc.su.se
Thu Sep 8 18:23:35 CEST 2005
The retuned loops are for x86-64; I've rescheduled them to use all
available registers (16, iso 8 for ia32).
There are no improvements to the ia64 loops that I know of; the
limitation on host ia64 hosts seems to be the cache and the compilers
ability to optimize the C neighborsearching code.
For small systems (neighborlists fit in cache) the ia64 nonbonded
loops execute very close to machine peak. The only reason why x86-64
and ia32 are faster is that they provide SIMD instructions.
On Sep 8, 2005, at 4:03 PM, Michel Cuendet wrote:
> Hi Erik,
>> And, just to give you some incitement for testing: Release 3.3 has
>> ... x86-64 and EM64T 64-bit
>> assembly loops (recently re-tuned to use all register), ...
> Have these "recently re-tuned loops" been added after 27.05.2005 ?
> If yes, would they improve substantially the speed on itaniums ?
> Michel Cuendet, Ph.D. student
> Laboratory of Computational Biochemistry and Chemistry
> Swiss Federal Institute of Technology in Lausanne (EPFL)
> CH-1015 Lausanne
> Switzerland Phone : +41 21 693 0324
> lcbcpc21.epfl.ch/~mitch +41 1 633 4194
> gmx-users mailing list
> gmx-users at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
More information about the gromacs.org_gmx-users