>> I hardly call "module size blows up" to be no rationale for limiting the
>> number of CPUs.
> Well, if now is set to 512, the problem was not that big, I mean...
> solvable.
IIRC the Linux kernel handling of CONFIG_NR_CPUS has changed: it used to
allocate potentially large static data of size proportional to
CONFIG_NR_CPUS, but has been changed recently to do that dynamically
based on the actual number of CPUs. Which is why the default can now be
set to 512 without suffering from undue blow up.
>> Why 32? Well, it's a nice power of 2 and lots more than the number of
>> cores in 99.9999% of all computers.
> That seems not to be a very practical approach to base such decision.
On the contrary, it seems eminently practical.
Stefan
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive:
http://lists.debian.org/jwvei2pkqon.fsf-monnier+gmane.linux.debian.u...@gnu.org
