Re: Kernels dying on parallel code
- To: mathgroup at smc.vnet.net
- Subject: [mg111471] Re: Kernels dying on parallel code
- From: "Jon Harrop" <usenet at ffconsultancy.com>
- Date: Mon, 2 Aug 2010 07:06:49 -0400 (EDT)
- References: <i33ctt$85p$1@smc.vnet.net>
I only have a license for 4 kernels but I'd expect Mathematica to tell me that rather than spew out pages of errors every time I try to use parallelism. I just wanted to check that everyone else was seeing the same behaviour. Cheers, Jon. "Zach Bjornson" <bjornson at mit.edu> wrote in message news:i33ctt$85p$1 at smc.vnet.net... > Do you have a proper license for 8 kernels? The normal home(?) license is > only > for four kernels; this may be Mathematica's way of saying you can only use > four. Just a guess. > > Quoting Jon Harrop <usenet at ffconsultancy.com>: > >> I have an 8-core desktop composed of two quadcore Xeons. Whenever I try >> to >> run parallel Mathematica 7.0.1 code it tries to spawn 8 kernels and then >> four of them die, resulting in lots of communication errors. Is this >> normal? >> >> -- >> Dr Jon Harrop, Flying Frog Consultancy Ltd. >> http://www.ffconsultancy.com >> >> >> > > >