Closed sjspielman closed 1 year ago
Attached to this comment is the stdout
from running RUN_LOCAL=0 bash scripts/run-manuscript-analyses.sh
, at the bottom of which is the output fromtime
, run with 64 GB RAM on 4 cores:
real 25687.98
user 3.60
sys 7.14
That's 25687.98/60 = 428.133
minutes.
For posterity, here's what I actually ran on the AWS instance -
nohup time -p docker run --rm --memory=64g --cpus=4 --volume="/home/ubuntu/OpenPBTA-analysis:/rocker-build:" ccdlopenpbta/open-pbta:latest bash run-manuscript-analyses-wrapped.sh &
Where the contents of run-manuscript-analyses-wrapped.sh
are simply: RUN_LOCAL=0 bash scripts/run-manuscript-analyses.sh
. This script was really only needed for me to get that RUN_LOCAL=0
variable in there before bash
without docker complaining.
The final nohup
file was renamed to stdout_run-manuscript-analyses.txt
and uploaded to this comment.
Attached to this comment is the stdout
from running RUN_LOCAL=0 bash figures/generate-figures.sh
, at the bottom of which is the output fromtime
, run with 64 GB RAM on 4 cores:
real 6273.08
user 0.25
sys 0.44
That's 6273.08/60 = 104.5513
minutes.
I want to note that the vast, and I mean vast (not formally profiled, but my guess is about 90% of the runtime) is running and exporting snv-callers
PBTA (not TCGA) figures:
https://github.com/AlexsLemonade/OpenPBTA-analysis/blob/ba5b09344b0f9c78d196d0d8c97a891b999bf0ea/figures/generate-figures.sh#L137-L140
For posterity, here's what I actually ran on the AWS instance -
nohup time -p docker run --rm --memory=64g --cpus=4 --volume="/home/ubuntu/OpenPBTA-analysis:/rocker-build:" ccdlopenpbta/open-pbta:latest bash run-generate-figures.sh &
Where the contents of run-generate-figures.sh
are simply: RUN_LOCAL=0 bash figures/generate-figures.sh
. This script was really only needed for me to get that RUN_LOCAL=0
variable in there before bash
without docker complaining.
The final nohup
file was renamed to stdout_generate-figures.txt
and uploaded to this comment.
64 GB RAM, 4 cores
Analyses: 25687.98 s Figures: 6273.08 s
Woops, this needs to stay open as there are code changes from the Big Run that need to be PR'd!
We want to do Big Run on AWS to obtain a time/cost estimate.
This requires ^ scripts be all set for resubmission.