[CPMD-list] NCACHE size

[Axel Kohlmeyer]

>>> "IK" == I Kozin <Kozin> writes:

hello igor,

IK> Hi Axel,
IK> first of all thanks a lot for putting up such a good CPMD website!
IK> An excellent source of resources!

thanks. it is nice to know that that stuff is useful.

>> btw: the best way to optimize the fft is use it as little as 
>> possible (e.g. by using the REAL SPACE WFN KEEP keyword, provided
>> there is enough memory on your machine).

IK> I am only doing benchmarking. Therefore I don't think it is fare
IK> to use it although I'll keep it in mind.

in my opinion, everything that helps a job to run faster on a 
specific machine while producing a correct result is fair. (just look
at how the SPEC benchmarks are run!). if it does not help on a specific
architecture, then this is the problem of the architecture.
for something as cpu-time consuming as cpmd, even small 
improvements add up. but then again, you can always gain more by
smarter planning of your calculations.

>> not that i know of. feel free to try. the default fft is quite
>> competitive, considering that it is portable fortran code.
>> IMO it would be more interesting to see, how well a port 
>> to fftw3 performs.

IK> Could you please elaborate your last point? (what fftw3 is?)

this is a portable fft implementation, that can (kindof) tune itself
to the hardware at runtime. they are especially good at non powers-of-2
grids. which should be quite helpful for cpmd (less memory wasted). 
the goedecker fft also supports powers of 3 and 5, but some combinations
are less good than others. see fftchk.F.
since version 3 fftw also supports some SIMD instruction sets 
see http://www.fftw.org/
the benchmark numbers published on their webpage are quite impressive.

regards,
        axel.

IK> Regards,
IK> Igor



--

=======================================================================
Axel Kohlmeyer       e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.