On Sat, Nov 9, 2013 at 7:35 AM, qwertgfdsa78 <qwertgfdsa78@gmail.com> wrote:

However, when I tried my data, I always got

"terminate called after throwing an instance of 'std::bad_alloc'"

Google suggests maybe memory is not enough.
I used VirutalBox and assign 8gb memory to it.
My protein has ~3000 residues. I guess this will use more memory but how much RAM I need?

The size of the protein alone doesn't necessarily give us the answer - the resolution and size of the unit cell are also important. If you're willing to send me your input files (off-list!) I can try running it on one of our machines with huge memory and see what the performance is like (or whether it crashes anyway).

As a follow-up to the original bug report: this appears to be a problem with the compiler on Ubuntu 12.04 (and on 13 too). Disabling optimization of the Rosetta build prevents the crash, which suggests that it's a bug in g++ rather than Rosetta. This kind of problem is notoriously difficult to fix, but we will keep digging. I do not particularly recommend the "debug" build for production use because it will be much slower, but the next build of Phenix will enable this (and it can be forced by editing $PHENIX/phenix/phenix/command_line/rosetta_build_phenix_interface.py).

-Nat