This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: hyperthreading fix, try #1

Volker Bandke wrote:
Hash: SHA1


a)  	Your test case fails on my machine as well, right at the

b)	I seem to remember that there was/is a separate problem with "make
- -j", even on non-hyperthreasd machines.  Unfortunately I cannot
search the mailing list right now due to some error (that possibly
has to do with's earlier breakdown.  I do remember a
sentence somewhere like "We don't recommend that you use 'make -j' as
it seems to be broken"

c)	Running your testcase with '-j1' instead of '-j2' does not fail on
my machine

Thanks for you info Volker. Here's what I've figured out:

This test does fail (in the same way) on non-hyperthreaded machines (Win2000Pro on a PIII). But, this is a regression from 1.5.12 (that test runs fine on the non-HT machine with 1.5.12. There was (maybe still is) a problem with running "make -j" without the max task counter, but "make -j2" has always worked very well on non-ht machines.

I will try out some other snapshots and see if I can narrow down the time when this showed up later today.


-- Unsubscribe info: Problem reports: Documentation: FAQ:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]