Hi @adamjstewart! I noticed that the following package(s) don’t yet have maintainers:…
data-vis-sdk-pr-build failed in https://gitlab.spack.io/spack/spack/-/jobs/2955038#L1721 …
> This is what I have in mind: https://github.com/jacobwilliams/hiop/blob/modern-fortran/src/Drivers/Dense/NlpDenseConsFEx1.f90 …
Isovolume also causes a problem, but slice works….
Looks great to me. I appreciate the inclusion of the new psm and hwloc variants….
MAINT: improve error for `from spack import *`
If you work with in-house spackages that use the old
from spack import * (which apparently was replaced in gh-30404), and update from an older
spack checkout, your
spack environment can become unusable for the addition of new packages and it will even suggest a command that doesn’t help to repair your environment like:
How to impose Dirichlet boundary conditions for saddle point problem
Hi, unfortunately, it is still now working well to impose the symmetry for the saddle point structure. I am sorry to create this issue again…. …View Comment
Include missing header file for the CUDA install (#710)
Co-authored-by: yzeng email@example.com</small>
@victorapm Thanks a lot for your reference output. I think I find my bug. The problem is the thread number. If I do not explicityly set
OMP_NUM_THREADS, the therad number will be the max of the available on the machine, which is 128 on my server. And according to my test, the threads cannot be more than 64. I am using the Hypre release v2.25.0. Here are the outputs in case they are helpful for you.
@psocratis Hi, I struggle to derive the weak form of advection (shown in Example 9). …
@njsyw1997, I cannot reproduce the behavior you reported. Here’s the output that I get with BoomerAMG-PCG: …
Mixed topology meshes fail in conduit::relay::io_blueprint::save
With v0.5 (63ef74f5387706801aec2ccb0db00cb3571f32d9), …View Comment
RestartManager path containing double slash
Hey we’ve noticed that if we provide a path container two forward slashes beside each other the following error occurs …View Comment
Clean up exercise files for 5pt stencil exercise and rework tutorial content.
Merge branch ‘feature/cuneo3/static-tensor-refs-and-indexes’ of github.com:LLNL/RAJA into feature/cuneo3/static-tensor-refs-and-indexes
Merge branch ‘develop’ into task/biagas/add_vtk9_support_phase2
Merge pull request #4470 from garlick/issue#4465
job-list: avoid segfault after job purge</small>
broker: content-cache `acct_dirty` can be greater than the number of flushable entries
While working on #4267, I noticed that sometimes the
acct_dirty field can be greater than the length of the
flush list, even when a backing module is not loaded. a
flux content flush will therefore hang, b/c
acct_dirty will never become zero, so the content flush will never respond to the original request.
Docker images resulting from the above changes have already been pushed to
I think you use enough scikit-build stuff. I opened an issue: https://github.com/scikit-build/scikit-build/issues/736…
Looks great! Thank you @fritzgoebel and @CameronRutherford. The Marianas issue seems to be beyond the scope of this PR….
Fix BLT target exporting to match camp and RAJA (#767)
error in using expression with "jitable" attribute
Hi, …View Comment
> Exporting still results in just the time and cycle being output. …
I think you could create a coefficient which returns the normal of boundary triangles. …
> What makes the small difference here? Is that related to where the matrix is assembled (host vs device)? …
I have been testing hypre on Perlmutter and it has been working well as a single solver or in mfem. But thank you @sjcrs for providing the hypre installation configuration in another issue. It was very helpful…
Are you sure the problem is symmetric positive definite? Have you tried GMRES or BiCGSTAB with your problem? …
flux-sched configure fails on TOSS 4 tce
Problem: configure fails here in the default TOSS 4 tce environment …View Comment
Just adding the link here for the live preview of this PR while we build this up: https://maestrowf–403.org.readthedocs.build/en/403/…
Only intel 18 failing now, will wait for #767 to be merged….
> Since the TPL builds take a bit, I will build from what you have here and create a Docker container that contains pre-built oneapi TPLs. …
Issue isn’t vtkm_filters. These are, for the most part, friendly wrappers that call vtkm functions that utilize vtkm worklets. With the exception of Threshold, which needs to have its worklet migrated to vtkm. …
Sycl build fix (#707)
Move fill functions in device utils back into the general functions block for use with sycl backend.</small>
Merge pull request #952 from grondo/news-0.24
add release notes for v0.24.0</small>
`--exclusive` nodes do not have their full resources emitted in R
When match policy is not
hinodex and nodes are requested exclusively with
--exclusive, only the minimum matching resource set is included in the generated R, though the node is still allocated exclusively. See e.g. this output on Fluxion, where the system instance uses
lonodex but single-user instances do not.
Missing ZFP_VERSION_DEVELOP variable in version.h
Hello @GarrettDMorrison and @lindstro , …View Comment
@data-panda No, this release does not include the latest CUDA and HIP work we have been doing. That will end up in the next release. Regarding CUDA streams specifically, that is not yet something our team has looked at yet. We’ve had discussions with others who have looked at this (see this paper, for instance) and would welcome a contribution….
@lindstro was this something that got a place in this release (release notes does not mention so)? …
Looks like they deprecated the macos-10.15 container on gh actions? https://github.com/actions/virtual-environments/issues/5583…
The mac jobs are failing due to
GitHub Actions has encountered an internal error when running your job., any ideas?
That’s definitely a problem, and probably means a refactor is in order. I see the conversation on slack from a bit ago, if matchers are cheap to create then doing that sounds good for now. If not, in the end it really should be a property of a single resource node in the jobspec, so it might be worth the pain to get it there from where it is….
Increment develop version after each release.
To help CI apps testing using zfp, it would be helpful if the version of develop could be different from the released version….View Comment
@lindstro I agree, I don’t think
c_ptrdiff_t is the culprit. Thank you for taking a deeper look into this!…
Thanks for the heads up @lindstro, we’ll pick it up and begin testing it. Congratulations on getting to the 1.0.0! …
## What’s Changed
- Fix for enabling OpenMP for host-only builds by @liu15 in https://github.com/LLNL/CARE/pull/200
Full Changelog: https://github.com/LLNL/CARE/compare/v0.7.6…v0.7.7</small>View Comment