-
Notifications
You must be signed in to change notification settings - Fork 11
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
docs: add MPICH #47
base: master
Are you sure you want to change the base?
docs: add MPICH #47
Conversation
OpenMPI will fail test. See HDFGroup#46 .
@@ -39,7 +39,7 @@ To build the DAOS VOL connector, the following libraries are required: | |||
|
|||
+ `libhdf5` - The [HDF5](https://www.hdfgroup.org/downloads/hdf5/) library. | |||
Minimum version required is 1.13.3, compiled with | |||
support for both parallel I/O and map objects | |||
support for both [MPICH](https://www.mpich.org) parallel I/O and map objects |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We shouldn't mention a specific MPI implementation here since the connector will operate fine with implementations other than MPICH.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I attended DAOS tutorial today and @mchaarawi said MPICH is necessary.
@jhendersonHDF, are you going to patch CTest for openmpi? If not, I think user should be aware that MPICH is required to make test pass.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I only use OpenMPI with the DAOS VOL connector, so no, MPICH isn't necessary. It's necessary currently for the tests to run via CTest, but it's not necessary in general. It's fine to document that MPICH is needed for the tests and/or to create an issue explaining that, but I don't think this is the correct place for that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Then, where is a good place to add this information for user?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ideally, we should just fix the issue so the tests aren't restricted to a particular MPI implementation.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This should say HDF5 1.14.0 instead of 1.13.3. I don't want to make it sound like that's a legit, released version of the library.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
probably a misunderstanding. mpich is not necessary for the DAOS VOL.
mpich is necessary if you want to use the DAOS ADIO driver (MPIIO VFD - native VOL). there is a difference.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@mchaarawi , thank you for the clarification.
I understood what you meant.
My point was: MPICH can work well for everything in DAOS stack so why should a DAOS user bother with OpenMPI?
Is there any extra benefit of using OpenMPI + HDF5 DAOS VOL?
OpenMPI will fail test.
See #46 .