GMP 6.0.0a on AMD A8-6600K
nisse at lysator.liu.se
Thu Apr 9 12:36:18 UTC 2015
Kevin Buckley <kevin.m.buckley at gmail.com> writes:
> Note, though I am sure you will have, that the snapshot's configure, inside
> a Xen environment, reports the Host type as
In general, it's good to repeat the important pieces of information.
There are surely readers on this list that didn't read and understand
all the information in your earlier messages.
> Host type: piledrivernoavx-unknown-linux-gnu
> whereas in a non-Xen environment, it reports
> Host type: piledriver-unknown-linux-gnu
> so is recognising (?) an AVX-based distinction.
I'm not familiar with the "noavx" logic, but from a quick look in
configure.in, I think the only difference is that gmp adds -mno-avx to
And, assuming config.guess is right, the above results mean that the cpu
seen by the xen guest doesn't support avx instructions, even though they
are supported on the host.
So my best guess at this point is that you get a GMP build where gcc
generates avx instructions, which your xen environment doesn't support.
Which would explain the illegal instruction exceptions.
Did the snapshot solve the problem or not?
Niels Möller. PGP-encrypted email is preferred. Keyid C0B98E26.
Internet email is subject to wholesale government surveillance.
More information about the gmp-bugs