GNU Pth support

Werner Koch wk at
Fri Aug 10 15:31:38 CEST 2007

On Fri, 10 Aug 2007 14:11, andrewbass at said:

> this is just a security concern,
> and Pth was a safe choice because performance is not of big essence, right?


Actually the performance of Pth applications is really good for all
kinds of I/O bounded servers.  Obviously it does only use one CPU while
not in kernel model.  That is in many cases not a problem as there is
not much computation to do.  For RSA and other crypto computations this
might be different but there is also a tradeoff between distributing
such task between CPUs and the advantages of cache locality.



More information about the Gnupg-devel mailing list