h5part AT lists.psi.ch
Subject: H5Part development and discussion
List archive
- From: "Matthew Bettencourt" <mbetten AT sandia.gov>
- To: "Achim Gsell" <achim.gsell AT psi.ch>
- Cc: h5part AT lists.psi.ch
- Subject: Re: [EXTERNAL] Re: [H5part] H5PartSetCanonicalView
- Date: Mon, 5 Nov 2012 08:39:22 -0700
- List-archive: <https://lists.web.psi.ch/pipermail/h5part/>
- List-id: H5Part development and discussion <h5part.lists.psi.ch>
OK, So i tried that, however, there is still an issue.
So, i did this code where I set a start and end, and check, are there no parts. If there aren't any parts I call H5PartSetViewIndices
npart = e-s+1;
if ( npart > 0 ) {
if (H5PartSetView(file_, s, e) != 0 )
error_out("Cannot set file view for h5part file\n");
}else {
if (H5PartSetViewIndices (file_, NULL, 0) != 0 )
error_out("Cannot set file view for h5part file\n");
}
///Here I ask what does the code thing the number of parts are. All my ranks think it has all the particles, and the assert fails.
h5part_int64_t npart2 = H5PartGetNumParticles(file_); // view didn't take, bug in h5part
assert(npart == npart2);
On 11/05/2012 07:30 AM, Achim Gsell wrote:
On Oct 31, 2012, at 7:28 PM, Matthew Bettencourt wrote:
In the doc you say both start and end need to be -1,Yes, there are more ways to reset the view:
"unset" by calling \c H5PartSetView(file,-1,-1);
H5PartSetView (file, -1, -1);
H5PartSetView (file, 0, -1);
H5PartResetView (file);
What I want is to be able to have one set of code which works in parallel forMy suggestion is to use
both if we have more particles than processors and if we have less particles
than processor. Right now I tried both the H5PartSetCanonicalView and
setting by hand, it is impractical to set all the indices that I want to read
and read them in for a very large run.
H5PartSetViewIndices (file, NULL, 0);
only in the special case where you want to select zero particles:
if (start <= end)
H5PartSetView (file, start, end);
else
H5PartSetViewIndices (file, NULL, 0);
Ideally one should have to have both start and end to -1, (which would mostYes, but it may break some existing codes. So we shouldn't change this in
likely fix the other bug)
Version 1.
Achim
- Re: [EXTERNAL] Re: [H5part] H5PartSetCanonicalView, Achim Gsell, 11/05/2012
- Re: [EXTERNAL] Re: [H5part] H5PartSetCanonicalView, Matthew Bettencourt, 11/05/2012
Archive powered by MHonArc 2.6.19.