jedbrown closed issue rsmpi/rsmpi#215.
Build process fails on Mac with this error
When running cargo build of late versions of Mac, the build script of libffi-sys a internal dep of libff1 encounters this error; …View Comment
v-dobrev commented on issue mfem/mfem#4977.
Adding the tolerance for changing dt did not actually change the result since the actual difference in dt on the last step is slightly above 2.7569e-15 (this is for the sample run I posted above). However, this is probably fine.
…
vsoch commented on issue flux-framework/rfc#481.
Actually I don’t see your feedback yet @grondo - let me know if you have thoughts on the above!…
grondo commented on issue flux-framework/flux-docs#309.
Thanks!…
trws commented on issue flux-framework/flux-core#7191.
Honestly I just didn’t know that, that makes me think that we might want to expand on that rather than going from that to a big object though. We’ve already seen some scaling issues on large json objects in other places, and JGF is likely to get that treatment at some point, so may as well keep it that way here. Still a good idea to be able to get the events efficiently by type, but keeping it incremental would be good….
grondo commented on issue flux-framework/flux-core#7191.
> Kinda orthogonal, and I
garlick commented on issue flux-framework/flux-core#7227.
Having a way to run a local health check in rc1 sounds like it would be useful. I think if rc1 fails the broker exits with the special exit code that inhibits restart….
chu11 commented on issue flux-framework/flux-core#7191.
A random thought I had tonight, regarding my idea 1 and idea 2 that I listed above. …
vsoch commented on issue flux-framework/flux-core#7229.
Will need guidance on the mac-os test - not sure if it’s flaky (and I can’t figure out from the output what it isn’t happy about). …
grondo commented on issue flux-framework/flux-accounting#787.
It would have to be available per job, since there could be a different limit for different banks in the flux-accounting case. …
garlick commented on issue flux-framework/flux-accounting#787.
I suppose another option is to allow the jobs into sched state but withhold them from the scheduler. That is already available except not per user….
cbritopacheco commented on issue cbritopacheco/rodin#159.
@copilot Please remove these sections or add appropiate documentation under the section Documentation from the README …
JustinPrivitera open issue visit-dav/visit#20750.
Mili Global Node Ids for Uncombined Databases
When using pick, you get different node id results for combined versus uncombined databases. It seems like there are only global node ids for combined databases….View Comment
cyrush commented on issue visit-dav/visit#20729.
@Sam-Briney I tried with develop version of visit, and I see the same values. …
alecbcs commented on issue spack/spack-packages#2679.
@spackbot run pipeline…
gardner48 commented on issue spack/spack-packages#2557.
@jrood-nrel does https://github.com/spack/spack-packages/pull/2679 fix the problem?…
adrienbernede commented on issue spack/spack-packages#2558.
@alecbcs @davidbeckingsale …
becker33 commented on issue spack/spack#7890.
This appears to have been fixed at some point in the intervening years. Tested with spack install zlib tcl which does not print output for zlib twice….
vsoch commented on issue singularityhub/singularity-hpc#694.
That looks like an issue with parsing the module. Take a look at the module file that shpc generates and see how it compares to modules on your system (and make sure you are using the right one)….
vsoch commented on issue pydicom/deid#295.
https://pypi.org/project/deid/0.4.10/…
jedbrown commented on issue openjournals/joss-reviews#6099.
I think this is really an author preference matter. In general, I’d say benchmarking the official version is best. I’m not sure how much performance is expected to have changed from earlier, nor whether it’s the same machine. For dense operations, the relative performance of MKL to BLIS and OpenBLAS is routinely benchmarked….
hppritcha closed issue open-mpi/ompi#14.
Allow F90 MPI_BUFFER_DETACH to return correct pointer
The current F77 MPI_BUFFER_DETACH implementation does not return the detached buffer pointer to the caller – it simply does not make sense to do this in F77 because a) you can’t get it, b) pointer implementations between compilers seem to differ, and c) even among the F77 compilers that do support pointers, you can’t compare or use the pointer in a meaningful way. There are two precedents that support this interpretation: LAM/MPI and CT6 both do not return the pointer to F77 callers….View Comment
tzanio commented on issue mfem/web#310.
Thanks @nmnobre !…
acfisher commented on issue mfem/web#204.
@tzanio There is some good work in here and I think we should rescue it. I can make the couple of changes that I suggested and dust this off if you want to provide another review/reviewer. What do you think?…
tzanio commented on issue mfem/mfem#4986.
Re-merged in next for testing……
vladotomov commented on issue mfem/mfem#4986.
Update CHANGELOG…
chu11 commented on issue flux-framework/rfc#482.
re-pushed, adding backwards compatibility. I wasn’t sure how to organize things. So I just added a blurb at the end of the section saying backwards compatibility MAY be supported via the old response format….
trws commented on issue flux-framework/rfc#481.
Dang, it sounded like such a good idea, and it looks like it works for HPCToolkit on gitlab. It looks like roles aren’t easy to make visible on github either. I suppose a list is fine, as long as it’s a single place, that’s where lists burn is if they have to be maintained in multiple places without a way to ensure they stay in sync….
cmoussa1 open issue flux-framework/flux-accounting#788.
database: add `max_sched_jobs` attribute per-association
If the max number of scheduled jobs needs to be enforced per-association, then it should be a property in the flux-accounting database that can be configured with front-end commands such as add-user and edit-user….View Comment
tpatki closed issue LLNL/variorum#576.
Sample at 50ms in the IBM Power9 Energy APIs
Currently, we’re sampling every 250ms – as a result, when we encounter a code that doesn’t do much work and takes less than 250ms, energy deltas get reported as zero. …View Comment
adrienbernede commented on issue LLNL/radiuss-spack-configs#152.
@tdrwenski thank you for the review. Since @kab163 really needs it and I have no problems with my projects, I say we merge it and further investigate your issues. If we need a patch, no problem with me creating a minor release….
mplegendre closed issue LLNL/Spindle#94.
Clean up warnings
artv3 commented on issue LLNL/RAJAPerf#589.
TODO: remove Bt as it is not used. …
adayton1 commented on issue LLNL/RAJAPerf#592.
> > What does google benchmark do to prevent things being optimized away? …
jeremylt commented on issue CEED/libCEED#1905.
In every instruction vpd though? We’ve run into situations I think where most but not all of the QF is vpd…
v-dobrev commented on issue xsdk-project/xsdk-issues#280.
What compiler is this using? I can’t figure out what’s behind mpic++….
cmelone commented on issue spack/spack-gantry#208.
@dependabot rebase…
hppritcha commented on issue open-mpi/prrte#110.
Could you show the error message mpirun is reporting? Also what configure options were you using? …
cjvogl commented on issue mfem/mfem#5126.
Pinging some people who may be interested in this PR: @jandrej and @nmnobre …
grondo commented on issue flux-framework/rfc#482.
I’m talking about one version of Flux talking to another, e.g. as would happen with flux top or flux pstree if a subinstance is launched using a different version. We’ve been pretty good about keeping interfaces compatible up to this point….
garlick commented on issue flux-framework/rfc#482.
Wouldn’t most users of this RPC (apart from job-list) be using the python API which would also be updated? As I type that I am doubting myself. …
wihobbs commented on issue flux-framework/flux-core#7211.
That’s great! We tend to be a heavily GitHub driven project, so this + other issues and PRs for sure. We also lean on Slack, and I sent you an invite to our workspace for collaboration and quick questions….
jameshcorbett open issue flux-framework/flux-coral2#441.
Offset rabbit data movement with computation
@behlendorf recently mentioned that there are some users who want to fill up the rabbit SSDs on some chassis with data before their jobs start. I think he mentioned these users would typically use node-local file systems and that data movement could take around an hour….View Comment
trws commented on issue desul/desul#160.
Yup, that transition is officially upstream for us, and I think 20 will follow soon….
garlick commented on issue chaos/diod#164.
FYI I’m going offline for a few days for travel but next week I’ll try to set up a 32 bit test environment to run this down….
alecbcs commented on issue LLNL/hubcast#190.
No worries haha :)…
adrienbernede commented on issue LLNL/Umpire#1009.
@kab163 OK, adding the camp release and updating the umpire dependency constraint now….
adayton1 commented on issue LLNL/CARE#357.
Can you add a release note about these fixes?…
jedbrown commented on issue CEED/libCEED#1905.
In particular, the v*pd (“packed double”) instructions, which I expect GCC is indeed emitting. I like to use Linux perf and then look at the hot instructions. If those are v*pd, then vectorization has been effective. What does upstream clang do on your kernels? Is there a reason you are using legacy icc?…
Andr00dz closed issue mfem/mfem#5128.
MFEM and HYPRE without MPI
Hello, …View Comment
milroy commented on issue compspec/fractale#21.
Alright, the latest updates are pushed. Let me know if this needs anything else….
benzwick commented on issue Slicer/Slicer#8883.
What if when multiple segments are selected the tools that only work on single segments are disabled and “grayed out”. I think this would be quite intuitive and easy for users to understand….
adayton1 commented on issue LLNL/axom#1723.
@kennyweiss, @rhornung67, I don’t have permission to merge, so one of you will have to push the button….
jeremylt commented on issue freeCodeCamp/freeCodeCamp#64241.
Please use the forum to request help on the curriculum …
bgunnar5 commented on issue LLNL/merlin#596.
@doutriaux1 I can. The only thing that’s really changing here is the backend functionality and the --steps option being removed. From the frontend, the user can still use merlin monitor ${YAML} with the exact same call….
artv3 commented on issue LLNL/RAJA#1949.
> Any thoughts on blockIdx and gridDim? …
adayton1 commented on issue LLNL/MPIDiff#11.
Could you add a release note and tick the version number in CMakeLists.txt?…
jedbrown commented on issue EnzymeAD/Enzyme#2504.
I’m assuming the audience that cares about repo size will always be using a shallow clone? I just want to make sure everyone here knows it’s not possible to actually shrink the (non-shallow) repo size without invalidating old hashes. It is possible to split the history, but it changes hashes, which is confusing for people who have clones lying around and disrupts workflows that depend on existence of old hashes. I have experience doing such things, though it isn’t a decision to make lightly and I’m not recommending it….
emily-howell commented on issue Alpine-DAV/ascent#1615.
Sure thing!…
trws commented on issue vsoch/hpsf-tac#1.
> > This is spot on - afaik there is no central governance model or document (beyond a brief contributing guide). This is something we need to put together before we submit this to the upstream repository as an issue for review. I am suggesting that we make a central .github repository for all projects to share a single source of truth. …
samuelkgutierrez closed issue hpc/quo-vadis#411.
qv_scope_create() is broken with QV_SCOPE_FLAG_NO_SMT
I’m looking into a solution. I’ve found the problem, but a potential solution may involve maintaining multiple topologies. That is, if we detect the use of QV_SCOPE_FLAG_NO_SMT, we can restrict the topology, but then we can’t use that one for other scopes without the SMT flag….View Comment
jedbrown commented on issue CEED/libCEED#1905.
It would depend on the loop body. For example, if it calls functions that don’t vectorize (we did the series work with log1p because it doesn’t vectorize in common implementations) or has branches that are unsafe/hard for the compiler to transform into masked vector arithmetic….
vsoch commented on issue vsoch/hpsf-tac#1.
> I will ask to reserve that date for you during next TAC meeting. …
garlick commented on issue vsoch/hpsf-tac#1.
> This is spot on - afaik there is no central governance model or document (beyond a brief contributing guide). This is something we need to put together before we submit this to the upstream repository as an issue for review. I am suggesting that we make a central .github repository for all projects to share a single source of truth. …
grondo commented on issue flux-framework/flux-core#7215.
Thanks! …
trws commented on issue vsoch/hpsf-tac#1.
Given this state, maybe it would be useful to do just a couple of tweaks in this PR then merge so we can collaborate via further PRs? Guess it depends on what all we want/need to do before trying to move this forward….
imguoguo commented on issue sipeed/NanoKVM-USB#82.
Does it manifest as a dialog box prompting you to select a serial device, but then displaying “Failed to connect serial. Please try again” after selection? If so, it’s likely because the current user doesn’t have permission to connect to the serial port. You need to add the current user to the uucp group (this worked successfully on my Arch Linux) or the dialout group (this might be necessary on Fedora or Ubuntu)….
vsoch commented on issue pydicom/deid#294.
I’m not sure I’m following - so let’s say you have a dicom with some set of Rows/Columns:…
Andr00dz open issue mfem/mfem#5128.
MFEM and HYPRE without MPI
Hello, …View Comment
alecbcs open issue dmy3k/gnome-adaptive-brightness#11.
Allow for Additional Brightness Level Categories
It would be very cool if users could dynamically add/remove brightness level categories. …View Comment
cbritopacheco commented on issue cbritopacheco/rodin#156.
@copilot Add interpolation tests on 8x8 grids. Also add tests to verify the case of one cell meshes. Across all geometries for K = 3….
markcmiller86 commented on issue LLNL/Silo#503.
Resolved in #505 on 4.12RC and #508 on main….
alecbcs commented on issue FrameworkComputer/EmbeddedController#73.
I’m also experiencing this issue on an AMD Framework 13 when using gnome-adaptive-brightness (issue linked above)….
v-dobrev commented on issue xsdk-project/xsdk-issues#279.
> BTW: I’m not yet sure if I can use hypre-3 in xsdk - due to mfem dependencies [as there is no mfem release with hypre-3.0 fixes]. cc: @v-dobrev…
JustinPrivitera commented on issue visit-dav/visit#20666.
@IdoAkkerman I have added your changes here: #20737 …
v-dobrev commented on issue mfem/mfem#4028.
Re-merged in next for testing……
becker33 commented on issue spack/spack#51566.
Closing this – now that we’re not trying to get it in for v1.1.0, we have time for a better solution….
hppritcha commented on issue openpmix/openpmix#3722.
this is the commit that tag 5.0.10rc1 points to …
hppritcha commented on issue open-mpi/ompi#13484.
okay i got this to run but i don’t observe this heap-use-after-free warning in the voluminous output….
tzanio commented on issue mfem/mfem#4996.
Merged in next for testing……
benzwick commented on issue benzwick/StreamVorti#10.
Thanks @will-wz-li this fixed the problem. I updated the build badges on the README.md as well. …
adrienbernede commented on issue spack/spack-packages#2558.
@spackbot re-run pipeline…
tgamblin commented on issue spack/spack#51633.
@haampie: I think -p should do exactly this for the new installer, with default of 0 (for which p == j), and then we should just allow j to exceed p (I think we currently cap it, right?)…
wihobbs commented on issue flux-framework/flux-core#7211.
Great summary, thanks @kush-gupt! And it was nice to meet you at SC25….
grondo commented on issue chaos/pdsh#171.
Looks good, thanks!…
garlick open issue chaos/diod#163.
diod_getlock: user@host:/path: Invalid argument
Problem: I noticed a dozen or so of these messages while running flux-core regression tests in a diod home directory in host picl7:…View Comment
markcmiller86 commented on issue spack/spack-packages#2346.
Just wanted to give praise for all of the CI checks here :clap: :tada:. Although it was sometimes a bit difficult to paw through the CI logs to diagnose failures, I was able ultimately to find and fix many issues without much hand-holding from other busy developers. …
dylan-copeland commented on issue mfem/mfem#4936.
Should this PR have the “bug” label? If so, I’m not sure what the bug is….
jandrej commented on issue mfem/mfem#5120.
I’m loading cuda/12.9.1 and use clang version 21.1.5 to build mfem with debug flags. The generated output file is ~15MB. I can share that with you internally if needed….
cmoussa1 closed issue flux-framework/flux-accounting#783.
`fetch-job-records`: properly log errors from script
Any errors or exceptions raised by the fetch-job-records script are not properly logged, which makes them hard to track down. It would be easier if they were logged in a similar fashion to the update-usage command….View Comment
adrienbernede commented on issue LLNL/Umpire#1009.
@kab163 Let me know if you need me to merge https://github.com/LLNL/radiuss-spack-configs/pull/150. I
vladotomov commented on issue CEED/Remhos#47.
@justinlaughlin is there a more recent version to use or do we merge this?…
vladotomov commented on issue CEED/Laghos#195.
Hi @lpottier, sorry we missed this….
gardner48 closed issue LLNL/sundials#722.
CVode does not return CV_TOO_CLOSE when initial step size is set
## Issue description…View Comment
markcmiller86 commented on issue spack/spack-packages#2346.
@patrickb314 sorry for false alarms…I discovered I was missing the compression lib options in the CMake builds….
camierjs commented on issue mfem/mfem#5120.
In serial, CUDA mode, compiling fem/integ/bilininteg_vecmass_pa.cpp does not show up the issue so far:…
vsoch commented on issue flux-framework/flux-sched#1420.
#$#%# I just ran pre-commit on the entire flux-sched codebase… …
garlick commented on issue flux-framework/flux-core#7209.
Thanks, fixed those and I’ll set MWP….
grondo commented on issue chaos/pdsh#171.
Meh, the commit validator is complaining that the commit message subject is too long. Not a huge deal, but you could reduce it to something like just …
emily-howell commented on issue Alpine-DAV/ascent#1621.
Looks like the only thing that failed was the dray-cuda issue! So that’s a win at least. …