Closed valassi closed 2 months ago
The first puzzling thing is that the commit hash printed out in the log does not seem to be the correct one.
Anyway, what may be going on here is that the job used to succeed on Mac i386 github hosted runners.
Darwin Mac-1713169328310.local i386
machdep.cpu.brand_string: Intel(R) Core(TM) i7-8700B CPU @ 3.20GHz
It now seeme to be running on Mac ARM
Darwin Mac-1715670763535.local arm
machdep.cpu.brand_string: Apple M1 (Virtual)
This is weird, because from what I read github hosted runners are only supposed to use x64.
What I will try to do:
As a workaround I have disabled testmis on Mac. This allows the CI for PR #832 to succeed, so that this can be merged.
I have opened two new followup issues
Ok the CI for #832 is now succeeding after disabling this test. Closing this issue with a workaround (the real fix should come in #840).
The CI for PR #832 is now failing the tests on MAC with a segmentation fault, details below.
NB: the code of the test that is failing has NOT CHANGED with respect to when the same test had been executed on April 15, see also below. In other words, this is a problem with MAC or with the configuration of the MAC CI.
I do not have an interactive MAC, and even if I had I am not sure that I would be able to reproduce the issue (since one month ago there was no issue).
I will try to check if the compiler changed or anything similar.
--
Failing job May 15 (https://github.com/madgraph5/madgraph4gpu/pull/832/commits/7a25b556d959b0e92a7e6bf4cdfd99434122ff74) https://github.com/madgraph5/madgraph4gpu/actions/runs/9075595776/job/24936514897?pr=832
Successful job April 15 (https://github.com/madgraph5/madgraph4gpu/pull/832/commits/d6986688d70c786816b54b13f38111f556c23af8) https://github.com/madgraph5/madgraph4gpu/actions/runs/8685948894/job/23816454188