Le 18 mars 2011 à 19:02, Christopher Faylor a écrit :
Or, are you saying that you still see failures when setting the buffer
size down to 64K?
With a buffer 64k all is fine.
With a buffer 31*1024*1024, that's different. It seems that
all pipes open at the same time add their buffers up to an upper limit
(the nonpaged value)
which cannot be exceeded.
This also explains a failure i had this afternoon, on the CPU which
was normally "non-failing":
i launched 5 xz-compressions at the same time (not big ones, but you
must know that xz-compression
can be rather slow even on small files, so it is very likely that the
5 were concurrent); and 3
of them failed unexpectedly. I understand now that 5 open pipes at the
same time would use
5*31*1024*1024 bytes, which is too much for a CPU that has
nonpaged=98Mb as i said in my previous
message.
Now, i suppose that with a 64k buffer, you would not be able to exceed
98Mb/64kb=1568 open pipes at the
same time.
Denis Excoffier.
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
