►
From YouTube: NUG Monthly Webinar, November 17, 2022
Description
Topic of the month: Transitioning from Cori to Perlmutter
A
A
A
A
So
our
agenda
is
same
as
usual,
so
we've
got
first
win
of
the
month
or
we
want
to
hear
from
people
and
today
I
learned
and
then
some
announcements
and
then
the
topic
of
the
day,
which
is
transitioning
from
Quarry
to
perl,
Mudder
and
then
coming
meetings.
If
you
all
have
any
topic,
suggestions
or
requests
and
then
last
month's
numbers-
okay,
so
it's
the
win
of
the
month
time
so
tell
us
about
an
achievement
or
a
shout
out
to
someone
else's
achievement
because
we'd
love
to
hear
about
people's
successes.
A
Yeah,
so
this
so
there's
the
strike
of
the
graduate
students
and
postdocs
going
on
yeah
and
yeah
I.
Guess
it's
been
going
on
for
a
few
days
now
so
I
don't
know
much
about
it.
I
haven't
been
to
the
campus
or
anything
I,
don't
know
what's
happening
there,
but
hopefully,
hopefully
they're
going
to
be
able
to
reach
an
agreement.
That's
going
to
be
good
for
everyone.
A
Okay,
well
maybe
we
can
move
on
to
today.
I
learned
so
did
you
learn
anything
exciting
or
new?
That
might
have
surprised
you
or
that
you
think
that
others
might
benefit
from
hearing
about
like
something
you
got
stuck
on.
You
hit
a
dead
end
or
you
were
wrong
about
it
or
you
have
a
tip
for
using
nurse
or
something
else
that
you
learned
that
might
benefit
other
nurse
users.
A
A
A
I
feel
I
feel
like
I
feel,
like
Ben,
Stein,
right,
Bueller
Bueller,
we're
not
getting
anything
today,
that's
okay,
all
right!
So
we
can
go
on
to
announcements
and
cfps.
So
mostly
I
think
these
are
included
in
the
weekly
email,
but
so
there's
a
few
calls
for
papers
still
going
on
so
the
the
past
conference.
A
Then
we've
got
some
upcoming
training
events.
So
next
week
we've
got
debugging
GPU
accelerated
applications
with
Nvidia
developer
tools,
migrating
from
query
to
prometer
training
that
that
nurse
is
putting
on
on
Thursday
and
then
well,
I,
shouldn't
say
next
week:
it's
the
week
after
next!
Isn't
it
because
next
week
is
Thanksgiving
so
the
week
after
that
is
when
these
are
then
there's
the
olcf
crusher
user
experiences,
which
is
on
December,
1st
and
9th
and
then
using
hip
and
GPU
libraries
with
openmp,
that's
on
December
14th,
so
those
are
all
coming
up.
A
A
We
we're
glad
that
the
people
are
coming
and
we
encourage
you
to
come
too.
If
you
haven't
come
yet
to
one
of
our
office
hours,
so
we're
holding
about
Thursdays
and
Fridays
from
9
A.M
to
noon,
Pacific,
Time
Of
course.
So
the
next
office
hours
are
going
to
be
Friday,
December,
2nd.
So
the
day
after
the
Quarry
to
Perlmutter
training
I
will
have
an
office
hours
for
people.
A
If
you
want
to
follow
up
and
then
the
next
week
we'll
have
office
hours
on
Thursday,
December,
8th
and
then
the
week
after
that
on
Friday
December
16th
we'll
take
a
holiday
break
and
we'll
come
back
on
Friday
January
6th,
followed
by
Thursday,
January,
12th
and
probably
we'll
schedule
more
after
that.
But
those
are
the
ones
that
we
currently
have
on
the
schedule
for
you,
foreign.
A
A
You
know,
as
you
know,
Corey
we
installed
it
in
2015,
so
I
think
it's
six!
It's
almost
it's
really
seven
years
old
at
this
point,
probably
I'd
have
to
check
the
exact
dates,
but
it's
probably
our
longest
lasting
system
that
we've
ever
had
at
nurse
we've
already
allocated
for
the
next
allocation
year,
based
only
on
Pearl
Motors
capability,
but
we're
going
to
give
you
all
some
more
time
to
transition
from
Corey
to
Perlmutter.
C
A
A
So
here's
what
our
retirement
plan
for
Corey
kind
of
looks
like
so
the
things
that
are
that
are
in
that
kind
of
teal
color,
we've
already
done
right.
So
in
October
we
had
a
software
freeze,
so
we're
not
installing
any
new
user-facing
software
on
Corey
I
mean
you
can
continue
to
compile
your
codes
on
Corey,
but
nothing
new
is
coming
from
nurse
and
then
we've
already
decided.
You
know
we
have
our
allocations,
they're
going
to
be
based
on
promoters
capacity
only
for
the
next
allocation
year.
A
We're
gonna
have
the
training
on
December
1st
that
I
really
encourage
everybody
to
attend,
and
so
that's
that's
where
we're
at
right
now
and
then,
hopefully
in
late
January
or
early
February.
That's
when
we'll
be
able
to
announce
our
final
date
for
decommissioning
Corey
and
let's
call
that
date
t
for
the
sake
of
argument
and
so
at
T
minus
one
week
we
will
put
it
in.
We
will
put
in
a
reservation
that
will
prevent
any
new
jobs
from
running
at
T
or
later
and
then
on
the
day
at
that
time.
A
When,
when
she
is
we're
going
to
delete
all
the
jobs
from
the
queue,
you
won't
be
able
to
submit
any
new
jobs.
But
we'll
continue
to
allow
you
to
log
in
to
retrieve
any
files
from
the
Cory
scratch
file
system
and
then
at
TA
plus
one
week
we
will
close
the
login
nodes
permanently.
You
will
be
able
to
access
the
Corey
scratch
system
anymore
and
then,
probably
a
month
later
or
so,
we
will
begin
disassembling,
Corey
and
sending
it
back
to
the
manufacturer
for
recycling.
A
A
A
D
A
Yes
and
I
will
share
the
results
with
you
all
I,
don't
know
if
this
is.
This
is
probably
not
showing
up
on
the
recording,
but
so
I'll
tell
everybody
here:
we've
got
two
out
of
16
folks
here
who
are
ready
to
use
Pearl
meter.
Okay,
wonderful,
thank
you
good
job,
and
then
we
have
five
out
of
16,
so
that's
30
31
who
have
begun
to
get
ready
to
use
promoter.
A
So
it's
good
on
you
excellent,
and
then
we
have
one
person
who
says
they
know
what
they
need
to
do
to
get
ready
to
use
Pearl
Mudder,
but
maybe
you
haven't
done
it
yet:
okay,
that's
good
and
then
the
remaining
half
50
eight
people
say
I,
don't
know
what
I
need
to
do
to
get
ready
to
use
Pro
matter,
okay,
excellent!
Well,
we
will
work
on
that.
So
hopefully
this
this
presentation
will
help
you
to
understand
what
you
may
need
to
do
so
I'm
not
going
to
be
able
to
solve
everyone's
problems
today.
A
Obviously
I'm
just
one
person,
but
hopefully
I-
can
sort
of
provide
you
with
some
overarching
high
level
strategies
to
know
what
you
need
to
do
and
then
you
can
follow
up
by
attending
our
training
on
a
Corey
to
prometer
transitioning
and
also
any
of
our
office
hours
that
we're
offering
and
if
you
can't
attend
either
of
those
know
that
the
training
will
be
recorded
so
you'll
be
able
to
view
it
online
along
with
professional
captioning,
so
that
you
don't
have
to
even
listen
to
it.
You
just
watch
and
then
the
office
hours.
A
If
you
can't
attend
office
hours,
that's
fine
feel
free
to
send
in
a
ticket,
and
we
can
get
somebody
to
help
you
either
through
the
ticket
system
or
if
we
really
need
more
of
a
like
a
one-on-one
sort
of
a
dialogue,
then
you
know
we're
happy
to
organize
something
for
you
we're
here
to
help.
We
really
want
you
to
be
able
to
be
successful
on
Pro
matter.
A
Okay.
So
let's
move
on
okay,
so
I
just
first
I
want
to
talk
about.
There
are
some
similarities
between
Cory
and
prometer.
You
know
they're
both
they're,
both
cray
hpe
systems.
So
they
they
have
the
cray
user
environment.
They
have
compiler
wrappers
that
you're,
probably
already
familiar
with
this
CC
Capital
Cc
or
ftn
for
C,
C,
plus,
plus
and
Fortran
respectively.
A
A
They
both
uses,
learn
as
the
batch
system
and
scheduler,
and
they
have
very
similar
cues
set
up.
So
we've
got
regular
premium
overrun,
there's
still
some
of
the
less
utilized
cues
that
that
aren't
quite
the
same
and
that
we're
still
working
on
perhaps
but
but
it's
a
very
familiar,
it
should
be
a
very
familiar
queue
setup
to
you.
There's
there's
not
really
any
surprises
there.
A
The
AMD
CPUs
are
similar
to
Haswell
in
in
the
clock.
Speed
like
they're,
fast
clock,
speed
but
they're
similar
to
KNL.
In
the
sense
that
they
have
a
large
number
of
cores
per
notes,
so
they've
got
64
cores
per
node,
so
anyway,
those
are
some
of
the
similarities.
You
know
that
I
see
between
the
two
systems.
Now
there
are
a
few
differences,
so
one
big
difference,
I
would
say
in
terms
of
the
user
environment,
is
that
we
use
lmod
instead
of
the
standard
modules
that
we've
used
on
Corey
Edison
Hopper.
A
All
of
our
past
machines,
we
used
modules
we're
using
one
called
lmod
which
stands
for
Lua
modules
and
so
that
it's
pretty
similar
I
mean
there's
this
the
same
sort
of
commands
that
you
would
use
with
lmod
versus
modules,
but
one
big
difference
that
we've
seen
people
struggle
with
is
that
sometimes
there
are
hidden
modules
that
that
they
only
come
out
of
hiding.
If
you
will,
when
you
load
prerequisite
modules,
so
one
of
them
is
netcdf.
A
We've
seen
a
lot
of
people
struggle
with
finding
cray
net
CDF
module,
because
it's
not
there
and
if
you
do
module
Avail
crane.cdf,
you
won't
find
it
and
that's
because
it
has
dependencies
that
you
have
to
load
before
you
can
actually
load
crane.cdf.
So
there's
a
command
in
lmod,
that's
called
module
spider,
and
that
is
what
you
need
to
use
instead
of
module
of
veil
in
order
to
find
any
module.
A
So
if
you
do
module
spider
crane.cdf,
you
will
find
crane.cdf
and
it
says
you
need
to
load
these
other
modules
before
you
can
get
this
module.
So
so
that's
that's
the
trick
to
remember
for
lmod
I
would
say
now
also
on
promoter.
We
have
these
GPU
nodes
where
we've
got
one
CPU
and
one
and
then
four
gpus
on
the
on
the
Node
and
of
course,
GPU
architectures
require
substantially
different
programming
models
than
your
CPU
code.
A
A
So
a
big
thing
is
that
on
Corey,
our
default
programming
environment
is
the
Intel
programming
environment
and
on
Pro
matter
we
don't
even
have
an
Intel
programming
environment
and
we
don't
plan
to
provide
one.
So
that
is
a
big
difference,
something
that
we
we've
seen
some
people
struggle
with
all
righty.
So
if
we
move
on
to
preparing
for
parameter
we're
going
to
talk
about
logging
in
I
see,
there
are
some
some
questions
here.
Okay,
is
there
a
shared
cue?
Yes,
there
is
a
shared
Q
for
the
CPU
nodes.
A
A
We
do
not
currently
have
a
big
job
discount,
because
Corey
is
still
considered
our
Flagship
system,
so
we
don't
actually
have
so.
The
reason
we
have
a
big
job
discount
is
because
we
have
a
big
job
metric,
where
we
have
to
have
certain
percentage
of
hours
that
go
towards
these
big
jobs,
and
so
we
offer
a
discount
as
an
incentive
for
you
to
make
those
big
jobs
right.
So
we
do
not
currently
have
that
on
on
Pearl
letter.
A
I,
don't
know
when,
when
we're
going
to
decide
exactly
about
that
for
promoter,
because
we're
also
looking
at
trying
to
change
our
our
metric
from
being
so
exclusive
to
big
jobs,
but
if
we,
if
we
were
going
to
do
that,
it
would
start
in
the
new
allocation
year.
It
definitely
wouldn't
start
any
time
before
that.
C
C
A
Yeah,
okay,
so
logging
into
Pearl
Mudder,
so
I've
actually
encountered
some
people
who
were
not
aware
that
you
can
use
their
same
account
that
you
have
been
using
our
quarry
on
Pro
matter.
Everyone
is
enabled
to
use
promoter
at
first
we
did
gate
people
to
using
Pearl
better.
You
know
early
on,
but
now
for
I,
don't
know
at
least
six
months
or
longer.
Everyone
has
been
able
to
to
log
on
to
Perlmutter
and
so
to
connect.
A
You
just
do
SSH
Elvis,
if
that's
your
username
Elvis
at
promoter,
Dash,
p1.,
nurse.gov
right
now.
That's
what
that's!
What
the
name
of
it
is
we'll
we'll
change
that
we'll
update
that
to
just
be
prometer.nurse.gov
soon,
but
not
yet,
and
then
you
just
use
your
password
and
one
time
password
in
the
same
way
that
you
would
do
it
on
Corey.
A
So
if
your
password
is
I,
love,
cheese
and
your
you
know,
your
one-time
password
is
one
two
three
four
five
six,
then
you
just
type
I
love
cheese,
one,
two,
three,
four,
five:
six
with
no
spaces
or
anything.
You
know
the
same
way
that
you're
already
doing
it
on
Corey.
You
can
also
use
the
SSH
proxy
to
reduce
the
the
frequency
the
need
for
you
to.
A
You
know
authenticate
every
time
you
log
in
and
of
course,
you
know
you
can
learn
more
about
this
in
the
nurse
documentation,
but
it
it's
no
surprises.
It's
exactly
the
same
way
that
Corey
is
is
used.
Okay,
so,
let's
say
you're
a
person
who
has
codes
and
they
are
CPU
only
and
you
don't
have
any
plans
to
use
gpus
great,
just
go
and
recompile
your
codes
on
promoter.
That's
the
thing
that
you're
going
to
need
to
do
most
of
the
time
this
just
works.
A
If
it
doesn't,
then
you
know
please
talk
to
us
and
we'll
try
to
help.
You
out.
I
have
seen
reports
that
the
promoter
default
compiler,
the
gnu
compiler,
is
more
pedantic
than
the
Intel
compiler,
which
is
the
default
on
query.
So
I've
seen
that
to
be
true,
there'd
be
people
who
they
they're
able
to
compile
on
on
quarry
without
any
warnings
whatsoever
they
go
and
they
try
and
compile
it
on
Perl
butter.
They
get
warnings
and
errors.
A
So
there
are
some
flags
that
can
reduce
the
strictness
of
of
the
compiler
and
then
there's
a
flag.
That's
that's
minus
W
pedantic
and
that
can
inform
you
about
which
standard
a
lot
of
your
code
is
breaking
and
then
you
can
go
back
and
use
a
flag
that
tells
the
compiler
to
use
that
particular
dialect
or
not
use
that
particular
dialect
of
C
or
or
Fortran
I,
guess
or
C
plus
plots
for
that
matter.
A
I've
mostly
seen
it
with
people
doing
c
anyway,
so
there
are
workarounds
for,
if
you're,
having
trouble
getting
your
code
compiled
because
of
this
kind
of
strictness
of
the
gnu
compiler.
If,
if
that
is
the
case,
like
I
said,
send
in
a
ticket
come
to
us
in
office
hours,
you
know
we'll
help
you
out.
That's
what
we're
here
for
okay
now,
let's
say
that
you
are
a
person
who
wants
to
use
the
gpus
on
prometer
excellent.
A
One
of
these
tools
is
called
Cody.
It's
so
likely
in
early
2023
we're
going
to
have
another
training
on
how
to
use
Kodi.
For
this
purpose
and
I
see
I
have
a
question
in
the
chat.
Oh
I,
don't
have
a
question.
I
have
Richard
giving
us
a
a
link
to
the
applications
on
yes,
so
if
you
go
to
docs.nurse.gov
Applications,
then
that
will
show
you
all
of
the
applications
that
we
have
on
machine
or
that
we
plan
to
have,
and
you
can
take
a
look
at
that.
A
A
D
A
You
know
gotchas
or
hiccups,
so
please
test
it
out
well
before
Corey
is,
is
gone
and
again,
if
you're
running
into
issues
talk
to
us
we'd
love
to
help
you
with
this,
so
that
we
can
make
sure
that
you
have
what
you
need
and
that
it's
that
is
working
for
you,
okay,
so
if
you,
on
the
other
hand,
want
to
use
Jupiter
with
the
gpus
On
Prom
header,
then
you
probably
need
to
rebuild
your
environment
to
exploit
gpus
because
chances
are.
It
doesn't
already
do
that
now.
A
If
you
have
used
the
Corey
GPU
test
bed
and
have
an
environment
set
up
for
that
you're,
probably
okay,
you
can
probably
use
that
on
Pearl
Mudder.
But
again
it
goes
back
to
the
same
thing.
Please
test
it
out
so
that
you
can
know
for
sure
that
it's
going
to
work
and
then
we
can
help
you
figure
it
out
if
you
can't
figure
out
how
to
fix
it.
A
A
If
what
they're
doing
is
correct,
it
makes
it
makes
life
so
much
easier
to
use
the
JavaScript
generator
I'm
a
huge
fan.
Okay.
So
now,
let's
talk
about
data
migration,
so
I've
I've
received
some
questions
with
a
bit
of
confusion
about
the
data
that
is
stored
on
Nurse
systems
and
whether
it
is
retrievable
from
both
promoter
and
Corey.
So,
let's
start
with
the
first
thing.
A
So
if
you
have
any
files
in
your
home
directory
or
in
your
project,
Community
file
system
directory,
those
will
be
available
on
promoter
no
action
required
so
because
these
are
Global
file
systems
and
these
Global
file
systems
are
mounted
on
all
nurse
resources.
So
when
I
say
home
directory,
of
course,
I'm
talking
about
the
directory
where
you
land
by
default,
and
it's
going
to
take
the
form
of
something
like
slash:
Global,
slash
homes,
slash
e
Elvis,
assuming
your
username
is
Elvis,
and
then
your
project
Community
file
system
directory.
A
This
is
a
directory
that
is
shared
by
members
of
your
project
and
it's
on
the
community
file
system
and
and
so
that
file
system
is
going
to
be
something
like
this
slash.
Go
Global,
slash,
CFS,
slash,
C,
dears,
slash
m1234.
If
your
project
was
m1234
now
before
we
had
the
community
file
system,
we
had
a
file
system
called
project
and
when
we
migrated
everybody
over
to
the
community
file
system,
we
left
this.
A
Otherwise
it
should
just
work
okay.
So
now,
if
you
use
the
hpss
archive
system
that
is
available
on
prometer,
so
you
can
access
hpss
from
promoter
just
like
you
can
access
it
from
Corey.
You
know
using
htar
or
other.
You
know
h
started
commands
that
start
with
eight
to
access
it.
Just
like
you
do
on
on
Corey.
It's
there's
no
difference
there.
Now.
The
one
remaining
thing
is
the
Corey
scratch
file
system,
okay,
that
is
attached
only
to
Pro
matter
or
I'm.
A
Sorry,
the
Corey
one
is
attached
only
to
Corey
and
not
to
Pearl
mutter
Pearl
Mudder
has
its
own
scratch,
which
is
attached
only
to
Pearl
butter
and
not
Corey.
Now,
I
want
to
remind
you
that
scratch
file
systems
are
used
for
temporary
storage
only
while
your
data
is
actively
being
used
and
it's
subject
to
purging.
So
if
you
leave
something
on
there
for
longer
than
12
weeks,
then
we
reserve
the
right
to
delete
it.
A
A
B
A
Maybe
I
I,
don't
know
that's
a
good
question
and
so
please
send
in
a
ticket
and
find
out
because
I'm
sorry
I,
don't
know,
that's
not
that's
what
I
do
okay
and
then
the
path
to
the
user
scratch
directory.
Okay,
so
on
all
machines
there
is,
there
is
a
a
it's
called
a
macro,
but
it's
an
environment
variable,
and
so,
if
you,
if
you
do
dollar
sign
scratch,
it
will
tell
you
the
the
path
to
your
scratch
directory.
A
So
that's
you
can
find
that
on
yeah
I
didn't
I
didn't
replicate
those
here,
sorry
mangshi,
but
but
yeah.
You
should
be
able
to
find
out
yourself
for
your
scratch
directory.
Okay.
So
next
we
are
going
to
talk
a
little
bit
about
how
we're
here,
to
help
I've
been
saying
this
the
whole
time
but
I'm
going
to
say
it
again.
We
are
here
to
help
you.
A
We
want
you
to
be
successful
in
in
using
Pro
matter,
so
we
really
want
to
help
you
move
over
to
prometer,
so
I
told
you
earlier
I'll
tell
you
again:
we've
got
this
migrating
from
Corey
to
Pearl
Mudder
training
coming
up
on
December
1st.
So
it's
really
focused
on
building
codes
and
running
jobs
and
there's
going
to
be
a
Hands-On
session
at
the
end
where
you
can
bring
your
own
code
and
bring
your
own
jobs
and
try
them
out
and
get
help
from
nurse
experts.
A
So
so
please
attend
that
training
and
sign
up
anyways,
even
if
you
can't
make
it
because
we're
gonna
record
the
training
and
we're
going
to
post
it
on
our
YouTube
channel,
complete
with
a
professional,
closed
captions
on
the
on
the
video.
So
please
do
come
to
that
training
and
then
also
we've
got
the
Corey
to
perlmeter
office
hours.
A
All
of
them
are
going
to
be
from
9am
to
noon
the
next
one
after
that
is
going
to
be
the
Thursday,
December,
8th
and
then
Friday
December
16th,
and
then
we'll
have
two
more
in
early
January
and
then
we'll
probably
have
more
after
that.
We
just
planned
through
this
allocation
year
for
now-
and
so
all
this
is.
A
This
is
a
zoom
meeting
that
we
put
up,
and
then
you
just
drop
in
anytime
anytime,
between
nine
and
noon
Pacific,
and
we
have
people
there
who
will
help
you
we've
got
breakout
rooms,
it's
how
we
do
it.
We
have
so
first
of
all,
I
have
at
least
three
or
four
people
who
are
there
to
help
you
and
we
have
breakout
rooms.
And
then,
if
you
have
kind
of
a
long,
in-depth
problem
that
you
want
to
help
solving,
then
we
send
you
and
an
expert
to
the
breakout
room
to
resolve
that
issue.
A
So
it's
super
fun,
I
love!
It
I,
get
to
meet
a
lot
of
users
and
and
kind
of
understand
more
about.
What's
going
on
with
with
your
problems
that
you're
encountering
as
you
make
this
transition
so
anyway,
I
just
wanted
to
to
once
again
stress
that
we
are
here
to
help
you
and
we
would.
We
would
love
to
help
you
more.
A
Okay,
do
we
have
let's
see
okay,
so
that's
the
end
of
the
Quarry
to
perl
meter
transition
presentation.
So
do
you
have
any
other
questions
for
me?
I
saw
there
were
some
in
the
chat,
but
I
think
I.
Think
thanks
Richard
for
answering
some
of
those
questions
for
me
about
the
scratch
quota
and
the
Globus.
Yes,
thank
you
all
any
questions.
A
A
Okay,
so
hopefully,
hopefully,
we've
moved
at
least
most
of
those,
the
50
of
people
that
I
pulled
at
the
beginning
of
this,
who
did
not
know
what
they
needed
to
do
to
prepare
for
a
pearl
letter.
Hopefully,
we've
moved
you
up
at
least
one
level
where
you
at
least
know
what
you
need
to
do
in
order
to
prepare
for
Pearl
meter.
A
But
if
that's
not
the
case,
then
of
course
happy
to
answer
any
questions
either
right
now
or
you
know,
in
in
a
ticket
or
in
one
of
our
office
hours,
all
right,
good
question
all
righty.
Well,
so
let's
talk
about
what's
coming
up
next,
oh
maybe
we
have
another
question:
oh
okay,
so
in
December
at
the
next
neck
meeting
we're
gonna
talk
about
preparing
for
allocation
year
2023
and
then
in
January,
we'll
talk
about
user
Community
engagements.
A
So
we've
developed
some
plans.
We
need
some
feedback
from
users
about
what
we're
going
to
do
for
the
next
year
and
then,
of
course,
I
haven't
figured
out
exactly
what
what
we'll
do
in
February,
so
I
welcome
any
ideas
or
suggestions
about
topics
that
you
all
would
like
to
hear
more
about,
and,
of
course,
we
always
welcome
like
lightning
talks
from
nurse
users
about
the
research
that
you
use
nurse
before
you
know
your
research
is
kind
of
what
what
makes
our
jobs
you
know
really
outstanding
and
worth
doing.
A
A
A
Your
job
like,
if
you
do
so,
you
did
SQ
and
you
looked
at
your
particular
job
and
it
says
wreck
node
not
available.
All
that
means
is
that
there
currently
aren't
any
nodes
available
to
run
your
job,
like
your
job
will
run
when
there
are
nodes
available.
But
typically
you
see
something
like
that
because
because
your
job
just
doesn't
have
high
enough
priority
to
get
in
and
at
this
point
in
time,
and
and
that's
why
it's
not
running
yet,
but
it
will
run
soon.
A
Okay,
Francois
Francois
asks
what
does
large
jobs
mean
good
question
Francois,
so
what
I
mean
by
large
jobs?
I
mean
jobs
that
use
a
substantial
portion
of
the
resource.
So
in
the
case
of
our
large
jobs
metric,
we
we
take
all
jobs.
Okay,
jobs
are
considered
large
jobs
if
they
use
at
least
1
8
of
the
the
total
machine.
Node
count
rounded
down
to
the
dearest
Power
of
Two
number
of
nodes,
okay,
so
on
on
Corey
in
the
k,
l
nodes,
we
have
like
about
9000
nodes.
A
A
It's
a
very
quiet
group
today,
but
I'm
glad
that
you
were
able
to
join
and
I
hope
to
see
you
next
time
and
and
thanks
again
and
please
don't
hesitate
to
contact
us
if
you're
running
into
any
issues
getting
on
to
prometer
and
using
prometer
and
after
this
is
over.
I
will
post
a
PDF
of
these
slides
to
to
the
web
page
for
this
meeting
and
then
I
will
upload
the
the
meeting
recording
to
the
nurse
YouTube
channel
as
well.