Steven Vandenbrande
Steven Vandenbrande
My timings for a test with 16 beads: ## IPI 1.0 RESULTS  ## IPI-DEV RESULTS (master branch)  Clearly the scaling is improved a lot using the i-pi-dev code....
> I'd say that if we want to make this more efficient it should not be regarded as a bugfix, but as a code optimization I agree. The i-pi-dev code...
> My two cents. The counters you are considering are related to DBCSR. Could you add `dbcsr_multiply_generic` and `multiply_cannon` timers to your last table? You are right that those two...
> Well, DBCSR does communications between neighbor ranks and my understanding is that you are using multiple ranks per node, so it would be good to run the OSU benchmark...
> Sorry, I did a mistake in my previous post, I meant "OpenMPI 4.1" + new CP2K (really sorry). I would like to exclude any issues introduced by changes in...
``` vsc33716@tier2-p-login-3:software-layer$ ./eessi_container.sh --verbose checking for duplicates: 'software.eessi.io,access=ro' and 'software.eessi.io' Using /tmp/eessi.oXWyJFQEsD as tmp directory (to resume session add '--resume /tmp/eessi.oXWyJFQEsD'). EESSI_TMPDIR=/tmp/eessi.oXWyJFQEsD HOST_INJECTIONS=/tmp/eessi.oXWyJFQEsD/opt-eessi SINGULARITY_CACHEDIR=/tmp/eessi.oXWyJFQEsD/singularity_cache Pulling container image from docker://ghcr.io/eessi/build-node:debian11 to...
> Maybe you're not correctly seeing a shell prompt appear? Can you try running a command, like `ls /cvmfs/software.eessi.io`, even though it still seems to "hang"? > > Can you...
It looks like the proposed fix does not cover all cases. When I run the [reproducer](https://github.com/TACC/Lmod/issues/780#issuecomment-3088633685) with Lmod 8.7.65 and add the following: ``` mkdir foo module use ${PWD}/foo ```...
> Hi there, thanks for catching this. It has been fixed and is available for testing in IS780-patchfix. You are welcome to test this; we will merge it into the...