On Fri, Dec 14, 2012 at 5:20 PM, Barry Smith wrote:
>
> It won't fix any complaints because people will just live with their
> out-of-date code until we remove the legacy support and THEN they will
> complain. (And by then we'll have half-forgotten what we did so it will be
> harder to help p
As a point of comparison. I've been running a PETSc CG algorithm on an
Nvidia K20. The simulation has 1.4e7 elements.
The PETSc AXPY takes .001 seconds in single precision. That's 26 GFlops.
In another simulation using a double complex BiCG algorithm with 1.e6
unknowns, the Petsc MatMult on the
We could have a two tar balls, the petsc release with backward compatibility
and one without and randomly parcel then out to downloaders. Then wait two
releases and collect all the petsc-maint data from both sets of users and see
which set are more of a pain in the ass.
Barry
On Dec 14,
Alright, then let's have it this way. :-)
>>It won't fix any complaints because people will just live with
their out-of-date code until we remove the legacy support and THEN they
will complain. (And by then we'll have half-forgotten what we did so it
will be harder to help people with t
/lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20121214/458ba7f1/attachment-0001.html>
-- next part --
A non-text attachment was scrubbed...
Name: ml_memory.c
Type: text/x-csrc
Size: 37505 bytes
Desc: not available
URL:
<http://lists.mcs.anl.gov/pipermail/petsc-dev/attac
It won't fix any complaints because people will just live with their
out-of-date code until we remove the legacy support and THEN they will
complain. (And by then we'll have half-forgotten what we did so it will be
harder to help people with their complaints). Far better to force them to
cha
Hey,
On 12/13/2012 03:46 PM, Jed Brown wrote:
> I'm sure that users would appreciate one release of deprecation. It's
> not hard to implement when deprecating a routine entirely, but it's
> trickier when changing the interface to an existing routine. It can be
> achieved through a "feature test ma
Hi guys,
today I got a gentle introduction into our testing machine equipped with
two Intel MICs. They are still beta, yet I could run some simple kernels
in native mode. As an example, without any modification of existing
OpenMP code for vector addition in double precision of 3e6 elements, I
niversity of Chicago
5735 S. Ellis Avenue
Chicago, IL 60637
---
Phone: 630-252-1229
Fax: 630-252-5986
-- next part --
An HTML attachment was scrubbed...
URL:
<http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20121214/984c6d91/attachment.html>