[CP2K-user] [CP2K:21060] Re: Use libint2 compiled with explicit SIMD instructions?

[Ralf Frischmann]

Maybe I wasn't overly specific in the above question:

Has anyone tried to use e.g. libint-2.9.0 in connection with CP2K with the 
libint cmake compile flag set to

*LIBINT2_REALTYPE= AVX or SSE*

The default for this variable is "double", which I have been using 
successfully over the years in connection with CP2K.

And, in case it is working properly, does an SIMD-enhanced libint2 library 
lead to a performance increase when it comes to 
CP2K ERI computation in a HFX-type calculation?

Thanks in advance. Cheers,

Ralf





On Friday, January 17, 2025 at 6:15:08 AM UTC+1 Ralf Frischmann wrote:

> Dear all,
>
> I was wondering whether CP2K is already able to support the libint2 
> library compiled with explicit SIMD acceleration (AVX, SSE: see e.g. 
> libint2 wiki), and if not, whether there are plans already to allow for 
> this potential performance-enhancing feature in a future CP2K release?  
> Thanks!
>
> Regards,
>
> Ralf
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/4ce25717-8aa2-405b-a2b9-c152c3660a09n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250117/5a042ca3/attachment.htm>