►
From YouTube: NUG monthly Meeting, Aug 2021
Description
Recording of NUG Monthly meeting in august - meeting page and slides are at https://www.nersc.gov/users/NUG/teleconferences/nug-meeting-aug-19/
A
A
A
I
suspect
that
most
of
us
by
now
are
already
on
slack,
but
just
in
case
you're,
not
here's
the
link
in
the
chat.
This
is
a
great
we
thought
forum
place
for
discussions,
amongst
you,
know,
users
or
nurse
users,
including
chat
during
this
meeting.
One
of
the
advantages
of
turning
in
slack
compared
to
the
zoom
chat
is
that
the
chat
continues
to
be
visible
after
the
meeting's
finished
and
continue.
The
conversation.
A
Okay,
so
we'll
follow
our
normal
agenda.
This
is
intended
to
be
a
yeah,
an
interactive
forum.
It's
a
it's
a
discussion,
not
a
presentation,
so
please
unmute
yourself
and
speak
up
and
you
have
something
to
add.
A
A
And
agenda
will
follow
a
normal
pattern,
we'll
start
with
looking
for
stories
of
win
of
the
month
and
it's
flip
side.
Today
I
learned
we
have
a
whole
bunch
of
announcements
and
cfps
and
we'll
be
interested
in
hearing
any
more
that
our
users
have
and
then
we'll
go
into
our
topic
of
the
day,
we're
going
to
talk
about
the
nugg
executive
committee
and
what's
involved
and
how
you
can
participate
in
that
a
brief
look
at
what's
coming
up
and
last
month's
numbers,
so
keeping
it
off
with
win
of
the
month.
A
So
the
purpose
of
this
segment
is
to
get
for
our
users
to
be
able
to
show
off
an
achievement
or
to
shout
out
somebody
else's
achievement
that
they're,
aware
of
so,
for
instance,
if
you've
just
had
a
paper
accepted
if
you've
been
working
on
a
challenging
bug
and
found
a
solution
or
you
know,
stumbled
across
something.
That
is
a
really
good
tip.
A
Scientific
achievements
are
always
good.
We
have
every
year
we
have
scientific
achievement
awards
and
also
awards
for
anybody
who
slipped
high
performance
computing.
So
this
this
forum
is
a
good
kind
of
your
opportunity
to
talk
about
things
that
you've
seen
or
done.
That
might
actually
be
candidates
for
those.
A
Also,
candidates
for
our
science
highlights
interesting
and
attractive,
so
we'd
love
to
hear
about
them.
A
Interesting
successful,
even
award
war
worthy
that
you've
achieved
or
seen
in
the
last
month
or
so.
A
A
Maybe
we'll
jump
straight
to
the
next
one
or
extend
it
out
into
the
other
side
of
the
queen.
The
today
I
learned
and
this
this
can
be
something
that
tripped
you
up.
It
could
be
something
that
you're
still
stuck
on
either
or
just
something
that
you
stumbled
across
that
might
benefit
other
nurse
users
or
might
help
us
to
identify
things
that
we
can
tweak
in
our
documentation
as
well.
A
Very
quite
meaningless.
C
I'll
give
some
so
I
added
some
documentation
on
the
interactions
of
corey
gpu
module
with
the
regular
queues
earlier
last
week,
like
you,
cannot
submit
to
haswell
or
know
if
you
have
the
cgpu
module
loaded.
C
A
I
do
remember
seeing
that
go
through
and
yeah,
that's
a
really
helpful
tip,
but
that's
a
really
easy
one
to
trip
up
on
as
well.
The
the
slurm
error
messages
are
not
always
obvious,
so
so,
thanks
for
putting
that
in,
it's
actually
also
a
good
reminder
if
you
haven't
discovered
it
already
when
you're
looking
at
the
nurse
doc's
pages.gov.
A
Up
here
this
is
the
gitlab
source
for
the
nurse
docs.
So
if
you
see
an
error
or
you
notice
something
that
would
be
helpful
to
add,
you
can
go
there
and
that
will
take
you
to
the
gitlab
gitlab
page,
where
you
can
clone
the
repo
and
make
a
merge
request
against
it.
A
D
A
Can
go
back
and
check
that
for
the
details,
important
things
to
watch
out,
for
we
have
a
upcoming
corey
os
update
planned
for
september
there'll,
be
a
new
programming
environment,
new
default
modules
and
a
particular
change
to
be
aware
of
and
prepared
for
is
that
if
your
executables
are
statically
linked,
you
will
likely
need
to
re-link
them
there.
Previously,
the
existing
static,
executables
may
not
work.
Dynamically
linked
executables
should
be
fine
but
yeah.
So
so,
once
that
update,
it
has
happened.
You
may
need
to
do
a
relink
of
things.
A
It's
just
a
heads
up
that
it's
coming.
You
may
have
noticed
that
the
corey
burst
buffer
is
currently
unavailable.
We
did
have
a
a
few
users.
I
think
who
noticed
this
by
by
seeing
their
jobs,
I
guess
sit
in
the
queue
rather
than
seeing
the
message
of
the
day.
A
So
there
is
a
note
about
it
on
the
my.nurse.gov
yeah
system
status,
dashboard
page,
but
it
means
that
if
you
have
burst
buffer
jobs,
they're,
probably
just
sitting
in
the
queue
waiting,
we're
anticipating
enough-
you
know
a
fix
for
that
problem
soon,
but
it's
not
there
yet
so
keep
watching
email
for
updates.
On
that
other
significant
thing:
that's
coming.
A
Soon
is,
it
is
very
nearly
ercap
season
october
sorry,
september
7
is
when
the
submissions
will
be
open.
A
There
are
a
couple
of
differences
in
this
year's
process
compared
to
previous
years,
and
so
next
meeting
clinton
will
walk
us
through
the
process
of
submitting
a
nocap
request,
and
you
know
what
what
you
need
to
know
for
this
year.
So
don't
miss
that
also
on
september,
7th,
it
will
be,
is
when
the
quarterly
allocation
reductions
are
planned
to
occur.
A
So
that's
a
way
for
program
managers
to
get
more
available
hours
to
get
or
to
redistribute
the
hours,
rather
by
taking
a
few
hours
away
from
allocations
that
not
using
them
at
a
fast
enough
rate
to
use
them
all
up
and
making
them
available
for
program
managers
to
redistribute
sort
of
as
needed.
A
Let's
see
so
we
have
even
more.
This
is
a
good
source
of
today
I
learned
stuff.
Actually,
I've
learned
a
lot
of
interesting
things
out
of
this.
This
series,
so
the
better
scientific
software
series
is
part
of
the
sap
ideas
project
and
it's
aimed
at
improving
software
development
practices
and
tools,
particularly
with
an
eye
to
scientific
software
development.
So
it's
quite
close
to
our
hearts.
A
Here
I
think,
and
so
that
organization
I
guess,
has
just
recently
announced
a
fellowship
program,
so
the
better
scientific
software
2022
fellowship
program
applications
are
now
open.
So
this
is
targeting
work
that
fosters
practices,
processes
and
tools
to
improve
scientific
software,
productivity
and
sustainability,
and
there
are
grants
available
for
for
this
that
are
up
to
25k.
A
So
this
could
be
a
you
know,
a
good
source
if
your
research
and
work
is
in
that
direction.
The
weekly
email
there's
details
and
the
only
needs
for
how
to
apply
more
more
information.
A
There's
a
bunch
of
calls
for
participation
out
there
at
the
moment,
some
some
of
the
more
recent
ones
that
we've
seen
the
women
in
hpc
workshop
has
a
cfp
out.
That
will
happen
in
conjunction
with
sc21,
also
in
conjunction
with
sc
21.
The
international
workshop
drop
on
performance
portability
and
productivity
in
hpc,
and
also
supercheck,
which
is
the
checkpoint
restart
forum
that
is
being
called
shared
by.
A
So
this
is
the
the
summary
of
the
current
ones
that
I
know
of.
Does
anybody
else
have
or
know
of
any
calls
for
participation
that
would
be
interesting
or
beneficial
to
the
nest?
Users.
A
See
there's
a
question
in
the
chat
about
the
prognosis
first
buffer.
Well,
it
will
come
back
but
we're
waiting
on
a
fix.
So
I
don't.
D
D
Yeah
steve
we
can,
we
can.
We
can
follow
up
on
the
exact
time.
You
have
it
and
see
been
getting
updated
today.
A
A
bunch
of
training
things
coming
up
in
the
near
future
in
a
very
near
future.
Next
week
there
is
some
cmake
training
for
the
first
four
days
of
the
week.
A
A
Spelling
error
in
this
next
one,
I
believe,
there's
an
el
valorium
before
portable
power
management
code
has
sessions
tomorrow
and
in
about
a
week's
time.
A
Further
ahead
in
september,
there'll
be
a
session
on
facility
testing
of
b4s,
which
is
the
extreme
scale.
Scientific
software
stack
so
facility
testing
using
the
efrs
test,
suite
spec
test
and
build
test
that
will
be
in
the
middle
of
september.
Also
in
the
middle
of
september
is
the
next
ideas,
sap
webinar
on
leading
open
source
projects.
So
this
is
attached
to
that
same
bit
of
scientific
software
as
the
fellowship.
A
And
then
further
ahead
in
october
will
be
a
p4s
deep
dive.
So
e4s
has
got
quite
a
good
range
of
scientific
software
and
particularly
libraries.
A
Are
there
any
others
that
I
didn't
know
about.
A
This
is
a
good
opportunity
to
you
know
shout
out
for
conferences
that
you're
you
know
participating
in
or
on
the
committee
for.
A
A
Committee,
so
do
a
bit
of
a
walk
through
what
the
committee
is,
what
does
and
so
richard's
online
and
may
yeah
speak
up
for
at
any
point
and
and
that's
which
has
been
involved
with
nuggets
for
from
nurse
side
for
quite
a
while.
A
So,
to
begin
with,
what
is
none?
Hopefully
everybody
knows
that
since
you
know
us
or
the
people
here
now,
it's
the
nurse
music
group-
and
it
inherently
includes
all
active
nurse
users
and
it's
designed
to
be
a
forum
for
users
to
communicate
with
each
other
and
with
nurse
staff
for
the
benefit
of
the
nurse
music
community.
A
So
activities
include
discussions
around
the
the
use
of
nurse
sharing,
acknowledge
and
experience
hearing
from
representatives
of
nurse
and
stakeholders
and
engaging
in
other
related
activities,
including
participating
in
special
interest
working
groups
and
having
interactions
with
other
relevant
groups
to
nurses
such
as
the
society
for
science.
That
use
a
research
facility,
and
I
see
yeah
dan
has
his.
I
can
icon
on
zoom
yeah.
A
And
so
you
know,
this
meeting
is
a
nug
activity
and
those
are
pretty
much
exactly
the
goals
that
we
have
here.
Sharing
knowledge
and
experience
and
discussing
the
use
of
desk.
A
So
nurse
hosts
a
nurse
users
slack
channel,
which
I
think
we're
all
aware
of
at
this
point
and
fairly
active
on.
So
this
is
a
forum
where
users
can
interact
with
other
nurse
users
and
in
a
more
ad
hoc
way,
nursing
staff
often
jumps
on,
and
that
contributes
to
the
conversation
there
and
the
other
thing-
and
you
know
didn't
mention
this
in
the
upcoming
events.
But
it
is
an
upcoming
event.
A
Nurse
hosts
an
annual
note
meeting,
which
is
open
to
all
users,
and
it
generally
includes
presentations
by
nurse
by
doe
sponsors
by
nurse
users
by
vendors
and
also
some
training
and
workshops,
and
so
yeah.
We're
actually
planning
for
that
to
happen
this
year
in
october,
the
eats
that
details
and
date
will
be
announced
soon.
A
So
how
does
it
work
and
what's
involved
so
all
nurse
users
are
invited
to
volunteer
to
serve
on
nugets?
You
don't
have
to.
You
know,
have
a
particular
length
of
time
using
their
score
or
a
particular
role.
We're
interested
in
a
live
range
input
and
nurse
will
select
representatives
with
the
aim
of
representing
as
diverse
and
broadly
inclusive
nurse
users,
projects
and
hpc
needs,
as
we
can.
A
So
we
want
to
make
sure
that
you
know
everybody's
needs
are
heard
and
addressed
so
so
we're
very
interested
in
having
a
a
really
good,
diverse
group
in
in
nuggets,
so
nugets
will
meet
regularly,
typically
up
up
to
about
one
hour
per
month.
So
it's
you
know
intended
to
be
a
not
too
heavy
commitment,
yeah
a
meeting
with
nest
staff
for
activities
such
as
the
ones
we
talked
about
before
identifying
needs
from
nurse
users
in
some
scientific
community.
A
So
if
you
join
nuggets
typically
nugex
members
will
serve
for
two
to
three
years
as
long
as
they
continue
to
be
active
nurse
users.
If
you,
if
you're
no
longer
using
nurse,
then
it's
difficult
to
be
representative,
but
we,
you
know
it's
good
to
have
some
sort
of
continuity.
A
So
the
nuggets
chair
and
co-chair
are
chosen
from
amongst
the
nuggets
members
they
kind
of
serve
as
the
the
primary
points
of
contact
for
nuggets
and
they
didn't
help
to
advise
on
the
structure
and
the
content
of
the
annual
meeting
and
the
chair
and
co-chair
serve
one-year
terms
in
kind
of
a
rolling
fashion.
So
each
year
the
co-chair
succeeds
from
the
previous
year's
chair
and
yeah.
A
new
co-chair
is
is
selected
so
and
the
terms
for
nuggets
members
are
staggered.
A
A
A
It
also
sort
of
puts
you
in
the
inner
circle.
You
get
some
closer
insights
into
how
nurse
works
and
you'll
also
have
a
an
avenue
to
alert
nurse
to
concerns
or
events
that
are
happening
in
your
scientific
community,
that
you
know
we
might
not
have
a
direct
connection
or
a
direct
interaction
with
or
as
as
loudly
so
yeah
you'll
be
able
to
kind
of
be
in
that
inner
circle
and
there's
also
kind
of
a
cross-pollination
aspect,
because
you'll
be
working.
A
You
know
reasonably
closely
with
other
people
from
diverse
scientific
domains,
yeah
who
have
all
got
different.
You
know
innovative
approaches
to
challenges,
there's
a
sort
of
opportunity
to
see
how
other
areas
of
science
are
solving
problems
that
may
actually
be
related
to.
You
know
scenarios
that
you're
working
on
so
there's
a
great
opportunity
to
you
know,
swap
ideas
in
fact
that
this
meeting
here
is
also
a
great
opportunity
to
swap
ideas,
but
in
the
mega
experiments
you're
in
a
you
know,
a
smaller,
tighter
interacting
group.
A
So
timeline,
so
we
were
planning
for
a
nug
annual
meeting
in
october,
which
means
time's
actually
getting
a
little
bit
tight
for
volunteering
or
nominating
people
to
join
nuggets.
A
We
so
we're
looking
for
volunteers
and
nominations,
kind
of
you
know
by
the
end
of
august,
so
that
we
can
select
new
members
for
negative
early
in
september
and
then
we'll
have
the
annual
meeting
in
october,
at
which
they're
retiring
and
continuing
community
member
committee
members
will
be
recognized
and
the
incoming
committee
members
announced
and
then
the
new
committee
officially
begins
its
tenure
at
the
at
the
start
of
the
new
allocation
year
in
january.
A
But
the
new
members
will
be
invited
to
join
the
meetings
during
that
period
between
between
october
and
january.
So
they
can
get
a
feel
for
what
happens
and
become
a
bit
more
familiar
with
it.
A
So
hopefully
this
is,
you
know,
picked
your
interest
and
desire
to
to
be
a
part
of
the
community
and
of
this
community
so
to
to
volunteer
or
if
it's
making
you
think
of
somebody
in
in
your
scientific
community
who
you
think
would
be
really
good
to
have
on
the
committee.
You
can
also
nominate
a
colleague
to
do
that.
Please
fill
in
this
short
form.
A
A
Here's
a
longer
version
of
the
link,
so
this
is
what
the
form
looks
like
it's
not
too
long,
a
few
simple
questions
and
then
some
optional
additional
information
that
helps
us.
A
So
that
link
will
be
posted
in
weekly
emails
coming
up
and
also
on
the
webpage
really
soon.
A
So
that's
an
overview
of
nuggets
and
what's
coming
up,
do
you
have
any
questions,
suggestions,
comments
or
other
input,
they'd
like
to
add.
F
F
Then
I
have
been
had
this
question
sometime,
but
is
it
possible
for
like
in
our
my
division
of
pinellas
working
on
atmos
sciences
and
global
change
simulations
we
have
sort
of
informal
seminar
each
month
that
you
know
we
bring
in
guests
from
outside
and
talk
about
some
ideas,
exchange
information
for
this
kind
of
seminar
event
at
each
individual
institution?
Can
we
is
it
possible
to
invite
one
of
the
nas
help
desks?
Member
as
a
speaker
or
maybe
joint
speaker?
F
I
can
be
one
of
the
presenters
together
with
one
or
two
and
ask
staff
members
to
give
some
presentation
about
how
how
mask
works
and
what
kind
of
sort
of
common
sense
procedure
is
assumed
and
what
kind
of
events
and
then
maybe
we
can
advertise
those
how
you
know
people
can
get
involved
in
nurse,
because
I
didn't
know
about
this
new
gex.
Actually,
until
I
know
one
of
my
colleagues
they're
a
committee
member
and
yeah,
so
yeah
so
again,
going
back
to
the
question.
F
D
Yes,
steve
I'll.
Take
that
one.
I
guess
yes
very
much,
so
we
we
would
love
to
do
that
and
we'd
like
to
encourage
anybody
else.
That's
listening
that
would
like
to
have
something
similar
to
just
contact
us.
Yes,
that's
not
it's
a
great
way
for
us
to
communicate
with
the
different
communities
is
to
participate,
and
things
like
that.
So
yes,
definitely.
F
You
I'll
get
contact
maybe
later.
A
D
I
don't
think
so
other
than
you
know,
we've
had
nuggets
has
been
in
existence
for
quite
a
while
and-
and
we
really
appreciate
the
people
that
have
taken
part
in
it.
But
we
don't
feel
like
whatever
model
we've
had
in
the
past
has
been
as
consistent
and
effective
as
we
had
hoped
and
part
of
us
everybody's
busy
and
we're
just
not
sure
what
the
model
is,
and
you
know
we'd
like
we'd
like
to
have
more
interactions
and
kind
of
a
more
dynamic.
D
Nuggets
and
so
that's,
why
we're
trying
something
a
little
bit
different
here
and
asking
for
volunteers
and
and
trying
to
be
a
little
bit
more
specific
about
the
kind
of
activities
that
we
would
have
and
so
we're
just
we're
just
trying
to
tweak
things
a
little
bit
to
to
make
it
a
more
effective
organization.
D
Because
we-
and
we
think
it's-
it's
really
important
for
us
to
kind
of
interact
and
can
keep
keep
our
finger
on
the
pulse
of
the
community,
and
you
know
all
of
our
community
all
of
our
different
communities.
The
nurses
are
very
have
a
very
diverse
user
community,
and
this
is
our
attempt
to
to
to
try
to
enhance
that
a
little
bit.
A
Yeah-
and
that's
probably
part
of
you
know
that
having
a
really
diverse
community
at
nursery
is
probably
part
of
why
it's
it's
been
kind
of
difficult
in
past
years
to
get
sufficient
votes
in
an
election.
You
know,
because
I
suspect
a
lot
of
the
nominees.
A
Most
of
the
nurse
user
community,
you
won't
actually
know
they
might
be
known
in
their
own
scientific
community,
but
you
know
not
so
widely
so
by
by
taking
this
approach
this
year,
where
we're
hoping
to
you
know,
I
guess
I
kind
of
widen
the
representation
yeah
and
help
make
the
the
different
scientific
communities
a
little
bit
better
known
to
each
other.
A
So
see
stephen
has
a
comment
in
the
chat.
That's
good
to
see
that
was
enjoyable
and
and
useful
different
categories
of
fields.
D
And
we
would
find
that
we,
we
weren't,
really
capturing
the
broad
demographics
that
we
thought
we
needed
to
all
the
time.
Sometimes
yes,
sometimes
no.
It's
just
kind
of
inconsistent.
A
Yeah
and
actually
to
that
in
you'll,
see
on
the
the
participation
form.
A
One
of
the
optional
questions
that
we're
asking
is
you
know
for
a
little
bit
about.
You
know
the
volunteer
or
nominee
so
because
we're
interested
in
getting
you
know
both
experienced
users
and
early
career
or
graduate
users
and
people
who
you
know
directly
run
jobs
as
well
as
people
who
use
data
from
jobs
that
somebody
else
has
run
people
who
manage
projects
we're
trying
to
cover
that
the
entire
range
of
experiences
that
people
are
coming
to
nurse.
A
F
F
Probably
many
of
my
colleagues
mainly
running
jobs
using
existing
community.
You
know
code
without
knowing
how
to
build,
for
example,
software,
because
they
only
gives
users
a
really
nice
tailored
outdoor
box
script
also
once
users
to
get
try
to
do
more
advanced
or
specific
experiments.
They
have
to
go
into
those
building
scripts,
but
until
then
autobox
already
prepared
the
case.
I'm
mainly
talking
about
those
used
for
the
climate.
Like
csm,
e3sm,
wolf,
you
don't
need
to
learn
pretty
much
anything
about
how
to
build
a
script
for
wolf
model.
F
You
have
to
know
about
the
user
guide.
It
gives
all
the
steps
you
just
copy
and
paste.
That's
it.
So,
once
something
error
happens
or
like
in
the
query,
regretted
status
gives
some
error
message.
They
don't
know,
we
don't
know
really
what
to
do.
Even
even
don't
know
how
to
ask
questions
to
the
help
desk.
F
So
that
comes
back.
What
is
a
user's
current
demography?
How
many
of
the
users
are
that
kind
of
more
really
beginning
beginner
level
of
hpc
and
how
many
are
more
like
experts,
building
social
for
hpc
environment,
and
if
we
have
such
a
statistics,
I'm
just
very
interested
in
to
see.
D
Sure
I
can
take
that
one
too
steve.
That's
that's
a
great
great
point.
We
we
have
an
annual
survey
and
we
asked
some
limited
number
of
demographic
questions
like
how
long
have
you
been
using
earth?
So
we
have
that
data,
but
I
think
that
would
be
a
great
question
for
us
to
add
to
the
annual
survey
steve.
Maybe
you
could
take
a
note.
D
It's
just
you
know
kind
of
how
do
you?
How
do
you
use
nurse
can
what's
your
ex
experience
level?
In
other
words,
are
you
are
you
developing
code
building
code?
Are
you
just
running
pre-pre-built
applications?
Are
you
having
to
create
batch
scripts?
You
know
what
that
that
sort
of
thing.
I
think
that
would
be.
I
think,
it'd
be
great
information
for
us
to
have,
and
we
don't
explicitly
have
it
now.
F
We
may
need
to
use
like
easier
long,
technical
terms
for
certain
section
of
the
document,
and
I
try
to
suggest
some,
but
I
just
didn't
have
a
chance
to
go
into
the
yeah,
the
the
good
luck
and
then
copying
those
last
documentation
myself,
but
that's
another
use
of
those
kind
of
demography
data,
but
yeah
thanks
for
listening.
A
Yeah,
so
that's
that's
one
of
the
things
we're
looking
for.
I
guess
for
the
for
the
nuggets
committee.
A
As
a
as
a
forum
to
you
know,
make
sure
that
we
hear
about
some
of
those
and
there's
also
some
good
input,
I
think
to
be
you
know,
gained
about
you
know.
What's
what
are
good
things
to
focus
on
for
training
sessions
for
training
events,
or
you
know
what
what
training
events
should
we
add
to
our
repertoire,
for
instance,.
D
Yeah,
that's
a
that's
a
great
topic
and
this
a
great
suggestion,
so
you
might
want
to
volunteer
for
nuggets.
You
know
one
thing
that
nugex
has
is
working
groups
and
right
now
I
think
there's
only
one
working
group
around
experimental
and
observational
data,
but
it'd
be
it'd,
be
great.
D
If
we
could
have
somebody
in
nuggets
that
was
interested
and
maybe
formed
a
working
group
and
maybe
brought
in
others
and
nurse
staff
could
participate
too,
and
maybe
you
had
you
know
regular,
if
not
that
frequent
meetings
to
discuss
you
know
effective
ways
to
present
the
documentation
or
what's
lacking
what
what
what
we
could
do
better,
because
you
know
tapping
ideas
from
our
community
is
a
fantastic
way
to
help
us
improve
how
we
do
things
because
there's
there's
so
much
knowledge
and
talent
and
perspective
out
there.
F
Thanks
for
the
comments
yeah,
and
indeed
that
kind
of
thing
for
documentation
and
training
program
might
be
really
nice,
and
I
I
do
like,
by
the
way,
youtube
playlist
for
the
new
user
training
event
and
ask.
I
recommend
that
phrase
to
every
new
users
in
our
group
who
study
to
use
now.
So
we
might
replace
that
more
as
well,
yeah
anyway,
yeah
thanks
for
the.
E
The
pittsburgh
supercomputing
center
used
to
have
a
very
interesting
support
model
in
which
one
staff
member
would
be
assigned
to
a
research
group,
and
he
would
be
your
primary
contact.
So
he
got
to
know
your
code
and
got
to
know
the
users
and
you'd
kind
of
go
there
with
your
first
line
inquiry,
and
I
found
that
was
quite
effective
at
the
time.
A
D
We
do
do
that
at
some
level
with
our
nissan
teams.
We
have
about
50
different,
it's
mostly
based
on
codes,
but
not
entirely
projects,
and
each
one
has
a
designated
contact
person
and
liaison.
E
Yeah
we
do
have
that,
but
I
guess
I
consider
that
person
mostly
available
for
porting
codes
to
the
new
machines,
not
so
much
for
the
production
running
problems,
we're
having.
D
Right,
yeah.
Okay,
that's
that's!
That's
a
good
idea
to
bring
up
again.
You
know
we
have
we've
thought
about
how
we
might
do
that
in
the
past
we
haven't
really
implemented
anything
too
much
except
we
we
do.
We
do
have
some
people
that
work
with
different
communities,
but
it's
not
on
a
real
organized
structure.
D
We
don't
have
a
real
organized
structure
around
that
right
now,
but
we'll
think
more
about
how
we
might
be
able
to
do
that
you're
able
to
identify
communities
that
could
benefit
from
this
kind
of
interaction
that
you
know
would
reduce
the
number
of
groups
that
were,
as
it
were,.
F
Sure
can
I
further
follow
up
on
steve's
suggestion
sounds
so
good
to
me.
Even
if
the
steve
mentioned
I
mean
steve
rick
mentioned.
If
the
resources
is
one
problem,
can
we
ask
pis
to
even
provide
some
funding
so
that
pi
can
provide
some
funding
to
mask
help
desk
and
then
in
in
return
we
can
have
some
designated
semi-designated
point
of
contact
to
put
the
code
and
change
the
code,
those
kind
of
framework,
maybe
as
part
of
the
rcap
application.
D
Yeah,
you
know,
that's,
that's
an
interesting
idea,
an
idea
I've
had
in
the
past,
and
the
thing
that
I
think
would
be
really
useful
is,
if
I
think
it'd
be
really.
I
think
it'd
be
really
useful.
If
projects,
particularly
from
the
larger
projects
were
able
to
have
a
designate
within
that
group,
that
is
their
nurse
liaison
or
nurse
contact
and
or
whatever,
and
they
could,
they
could
sit
within
that
group.
D
Maybe
they
can
sit
with
the
nurse,
maybe
half
and
half,
but
they
would
really
be
the
coordinator
for
the
interactions
between
nurse
and
and
that
group.
I
think
that
would
be.
I
think
that
would
be
potentially
a
really
nice
model.
E
D
It
that's
a
good
question
to
ask
see
we're
getting
lots
of
good
ideas
here.
I
don't
know
if
anybody's
taking
notes,
steve
you're,
taking
notes.
The
one
thing
we
could
ask
on
our
cap
is:
do
you
need
extended
support
and
and
or
would
you
like
extended
support
and
describe
what
it
is?
I
think
it'll
be
a
good
question.
A
Ideas
coming
out
of
this
group,
we
should
yeah,
we
should
definitely
yeah
fill
in
fill
in
the
form
and
volunteers
will
be
yeah
to
join
x.
D
Well,
we're
gonna
try
to
we're
gonna,
try
to
create
more
opportunities
to
have
this
sort
of
interactions
within
nuggets
too.
A
D
D
Well,
not
the
not
the
procedures
by
which
you
do
things
that
pretty
much
stay
the
same,
but
the
big
changes
are
going
to
be.
We
are
going
to
allocate
the
promoter
gpu
nodes
separately
from
the
other
collection
of
cp
cpu
nodes
that
we
have,
which
include
corey,
and
then
we
will
have
some
cpu
only
nodes
on
parameter,
so
there'll
be
two
pools
of
allocations,
gpu
time
as
it
were,
and
cpu
time
so,
you'll
have
to
put
into
two
different
numbers.
D
If
you
want
to
use
both
resources
next
year
and
then
the
other
is
that
we're
moving
away
from
this
kind
of
funky
nurse
gower
allocation
unit,
which
has
been
based
on
kind
of
a
cpu
hour
on?
I
think
it
was
hopper
years
ago
and
we're
moving
to
the
concept
of
node
hours,
which
I
think
will
put
us
in
more
alignment
with,
certainly
with
insight
and
alcc,
and
a
lot
of
other
allocation
programs.
D
E
G
Yeah,
so
I
just
wanted
to
ask
real
quick
since
richard
brought
it
up
the
as
far
as
perlmutter.
If
they're
going
to
be
different
units
on
there,
is
there
going
to
be
an
opportunity
to
to
try
to
benchmark
code
before
you
would
put
in
an
allocation
request?
So
you
know
how
much
allocation
to
ask
for,
or
do
you
have
to
kind
of
rely
on
a
calculator
that
may
or
may
not
be
very
accurate
for
your
particular
code.
D
That's
a
that's,
an
excellent
question
and
the
current
timeline
that
we've
put
up
there,
but
I
hope
that
that
is
the
case
that
people
can
get
on
the
on
the
system
itself
and
actually
do
those
tests.
D
If
that
doesn't
happen,
we
might
think
about
reconsidering
the
timeline
to
see
see
what
we
can
do,
because,
because
I
do
agree
that
you
know
you
need
you
need,
you
need
something
to
to
be
able
to
judge
the
size
of
your
request.
D
I
mean
we're
small
if
you
want
to
do
small
benchmarks
right
now,
we
can
get
you
access
to
some
a100s
that
are
attached
to
cory
and
see
if
I
remember
what
the
process
is
for
doing
that,
but
but
that
is
a
possibility,
but
it's
only
how
many
notes
is
it
there's
only
a
handful
of
nodes,
it's
yeah
they're.
A
Like
you
know
a
couple
or
six
or
something
like
that,
so
it's
so
it's
quite
small,
although
it's
probably
enough
to
at
least
get
a
sense
of
the
difference
in
performance
between
the
a100
and
the
v100.
A
So
that's
and
that's
still
going
back
to
kind
of
you
know
a
model
and
kind
of
calculating
and
estimating,
but
so
the
corey
gpu
nodes
v100s,
which
can
be
and
there's
there's
18
of
those
with,
I
think
eight
gpus
in
each
so
yeah.
There's
hopefully
enough
there
to
you,
know
to
get
a
a
bit
of
an
idea
of
performance
and
scaling.
D
A
Yes,
thank
you
all
right.
So
things
coming
up
as
richard
was
just
saying:
september's
no
monthly
meeting
we'll
have
a
walk
through
the
ocap
process
and
particularly
that
the
things
that
are
new
for
the
coming
year.
A
Especially
around
belmatter
and
gpu
allocations,
then,
in
october,
we're
planning
to
have
the
nugget
annual
meeting
kind
of
in
lieu
of
this
meeting.
It
may
not
be
on
the
same
day
we'll
announce
the
date
a
little
bit
closer
november.
A
A
presentation
of
some
of
the
work
of
one
of
our
regular
participants
and
on
that
note,
we're
very
interested
to
hear
kind
of
you
know
lightning
talk
presentations
about.
You
know
what
you're
doing
and
how
you
use
nurse-
and
you
know
I
think,
other
nurse
users
also
find
this
very
interesting.
So
if
you,
if
you
would
like
to
present
your
work,
it
could
be
also
a
good
opportunity
to
do
like
a
almost
a
practice
run
or
or
a
preparatory
run
for
something
you're
presenting
at
a
conference.
A
We'd
love
to
hear
from
you
in
the
last
couple
of
minutes,
quick
scan
through
last
month's
numbers,
so
our
scheduled
availability
was
was
also
quite
high.
Overall
availability
was
a
little
bit
lower.
The
bulk
of
the
reason
for
this
being
that
we
had,
you
might
remember,
a
fairly
large
center
power
maintenance.
This
is
a
once
every
three
years
kind
of
maintenance
that
is
important
to
do,
but
it
meant
that
we
were
unavailable
for
a
couple
of
days
there.
A
Over
a
weekend
there
was
a
a
number
of,
fortunately
quite
short,
outages
and
not
all
of
them
were
full
out
ages.
Actually,
I
think
a
few
were
system
degraded,
I
think,
lower
utilization.
A
It
was
that's
not
the
wrong
otp
one-time
password
issues,
making
it
difficult
to
log
in
a
couple
of
times,
so
you
can
see
on
the
chart,
that's
kind
of
what
it
looks
like
core
utilization
overall
was
you
know
nice
and
high
up
over
94
the
large
jobs
numbers
hadn't
come
in
yet
at
the
time
of
writing
these
slides,
but
we'll
do
an
update
of
these
before
posting
the
slides
to
the
website.
After
this.