64f077c4f6
stages: prepend the stage name to messages
2021-04-16 09:29:33 +02:00
7c94997023
control: add trap for bad exit
2021-04-16 09:29:33 +02:00
fb0dee4b61
exp: move exit1 experiment to slurm
2021-04-16 09:29:33 +02:00
bde54c69c5
sbatch: store queued status
2021-04-16 09:29:33 +02:00
2151e20bd6
exp: add exit1 experiment
...
Tests unit bad exits
2021-04-16 09:29:33 +02:00
886d16bcc6
garlic tool: add jq as dependency
...
So we can parse the experiment configuration in JSON
2021-04-16 09:29:33 +02:00
5c0f179830
stdexp: rename "name" to "clusterName"
2021-04-16 09:29:33 +02:00
422d359b48
script: stop on error by default
2021-04-16 09:29:33 +02:00
60248ab06b
article: remove not used figures
2021-04-16 09:29:33 +02:00
1cb63b464d
osu: adjust figures for publication
2021-04-16 09:29:33 +02:00
821b4f0d15
rplot: patch scales and fontconfig
2021-04-16 09:29:33 +02:00
0cf35decc5
osu: add mtu and eager experiments
2021-04-16 09:29:33 +02:00
26e3a86c78
garlic tool: check the presence of all the units
...
This check prevents a user from removing units between the
execution of the experiment and the fetch.
2021-04-16 09:29:33 +02:00
b96c39e0ba
noise: add srun signal bug to the list
2021-04-16 09:29:33 +02:00
f842f1e01d
slurm: add sigsegv experiment
...
Ensure that we can catch a sigsegv signal before and after the
MPI_Finalize call.
2021-04-16 09:29:33 +02:00
71c06d02da
stages: add baywatch stage to check the exit code
...
This workaround stage prevents srun from returning 0 to the upper stages
when a signal happens after MPI_Finalize. It writes the return code to a
file named .srun.rc.$rank and later checks that exists and contains a 0.
When the program is killed, exits with non-zero and the error is
propagated to the baywatch stage, which aborts immediately without
creating the rc file.
2021-04-16 09:29:26 +02:00
604cfd90a3
test: add sigsegv after MPI_Finalize test
...
The current srun version used in MN4 returns 0 if the program crashes
after MPI_Finalize, as shown by this test.
2021-04-16 09:28:02 +02:00
07253c3fa0
fwi: update figure index
2021-04-14 17:18:46 +02:00
eab323a13a
fwi: update io figure
2021-04-14 17:18:24 +02:00
8ce2a68cd7
fwi: update strong scaling figure script
2021-04-14 17:16:12 +02:00
99c6196734
fwi: update granularity figure
2021-04-14 17:05:09 +02:00
dd75a840ce
fwi: use enableIO instead of ioFreq
2021-04-12 20:09:17 +02:00
e49e3b087f
fwi: rename big io experiment
2021-04-12 19:49:31 +02:00
59040d9355
fwi: fix inverted resources
2021-04-12 19:31:35 +02:00
6422741cb7
fwi: merge io experiments into one file
...
The enableExtended parameter control if the experiment runs with
multiple nodes or only one.
2021-04-12 19:27:45 +02:00
99beac9b23
fwi: generate the model in every node
...
As we are using local storage, we need a copy of the input in every
node. The current method is to run the generator only in the rank which
has assigned the cpu 0 in the mask.
2021-04-12 19:01:10 +02:00
58dc277d3d
fwi: refactor ss-io with common.nix
...
Also, keep the names short and consistent.
2021-04-12 17:57:46 +02:00
47b326c646
fwi: generate the input at runtime
2021-04-12 17:46:07 +02:00
419e7f95cc
fwi: avoid input generation
...
The ModelGenerator is now included in the fwi-params, so that the input
can be generated at runtime.
2021-04-12 17:43:30 +02:00
b0af9b8608
srun: add postSrun hook
2021-04-12 17:41:59 +02:00
4afda7dbfb
fwi: use common.nix in sync_io experiment
2021-04-12 16:27:18 +02:00
02a103565c
fwi: use common.nix in reuse experiment
2021-04-12 15:48:59 +02:00
788dd13ebd
fwi: merge mpi pure experiment
...
The getResources function is used to assign the proper cpu binding
depending on the version. However, additional contraints are required to
ensure that we have enough points in Y.
By default the mpi+send+seq branch is disabled.
2021-04-12 15:37:39 +02:00
41665bc6fc
fwi: refactor config generation into common.nix
2021-04-12 15:01:25 +02:00
9aa07993b2
fwi: refactor ss and granularity experiments
...
A common.nix file contains the shared stages
2021-04-12 14:41:26 +02:00
e0a68c077c
fwi: merge forkjoin ss experiment into one file
...
Additional options are only active with enableExtended = true
2021-04-12 12:51:10 +02:00
989f6ee018
fwi: adjust input size to meet timing constraints
...
The previous iniput size for both granularity and strong scaling tests
where too big to meet the timing constrains needed for garlic. This
patch sets a new, smaller, input size.
Also, a minor cleanup is applied to the rest of the fwi experiments
and figures.
2021-04-07 12:44:14 +02:00
3e5a56ebdb
fwi: add tampi non-blocking variant
2021-04-07 12:44:14 +02:00
3ef4a505d3
fwi: add strong scalability tests
2021-04-07 12:44:14 +02:00
aadce016e1
fwi: add granularity and data reuse experiments
...
The data reuse experiment shows the effect of poor data locality versus
task granularity.
2021-04-07 12:44:14 +02:00
1d9a5c4721
fwi: fix input derivation
...
The fwiInput derivation must be the same used when compiled the fwi app
as the fwi-input used in the experiment.
2021-04-07 12:44:14 +02:00
11e400abb5
fwi: remove old experiment
2021-04-07 12:44:14 +02:00
a8477b1b05
fwi: add test figure with the time
2021-04-07 12:44:14 +02:00
7a6cbd3a9e
fwi: update test experiment
2021-04-07 12:44:14 +02:00
3de7b5a0b6
fwi: save the params and frequencies files
2021-04-07 12:44:14 +02:00
485b9150e5
fwi: add problem size parameters
2021-04-07 12:44:14 +02:00
fa0e9f591f
fwi: update repo url to PM server
2021-04-07 12:44:13 +02:00
de175b2380
fwi: fix input name
2021-04-07 12:44:13 +02:00
bfbbc294ae
fwi: split into input and solver
...
All branches compile with several hacks.
2021-04-07 12:44:13 +02:00
9bea3cc264
fwi: add oss experiment
2021-04-07 12:44:13 +02:00
