►
From YouTube: NUG Meeting 2014: Waters
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Welcome
back,
I
think
our
next
speaker
needs
little
introduction
for
most
of
you,
but
just
in
case
bill
is
the
project
manager
at
the
blue
waters
at
ncsa
in
the
university
of
illinois,
my
alma
mater
modder,
and
he
was
here
at
nurse
for
12
years.
I
think
your
title
was
nurse
general.
A
B
500
capable,
yes,
so
I'm
very
happy
to
be
here
and
very
honored
to
be
asked
to
talk
a
little
bit
at
this
event.
So
what
my
goal
is
is
just
to
to
take
a
step
back.
We
had
an
awful
lot
of
fun
hearing
about
reminiscences
and
things
that
went
on
on
our
individual
basis
in
the
early
days
of
nurse.
Can
I
take
a
step
back
and
say
what
has
nerves
contributed
to
super
computing
overall,
the
super
computing
community.
We
have
heard
some
of
the
impacts
nurses
has
had
scientifically.
B
I
wanted
to
take
a
step
back
and
look
at
from
the
super
computing
community
and
give
you
my
thoughts
on
what
nurse
has
done
these.
These
are
entirely
random
thoughts
in
their
mind.
For
about
a
hundred
million
time
of
improvement
in
performance,
we
heard
a
little
bit
about
that
a
little
bit
earlier
about
what
we
can
do
today
versus
what
we
could
do
many
years
ago,
40
years
ago.
So
it's
spanning
a
lot
of
that
time.
B
The
other
thing
is
when
francesca
sent
me
an
email
asking
me
to
participate
as
an
old
timer.
I
objected
to
that.
I
I
don't
think
that
that
that
fit
myself
image
of
of
this,
but
she
persisted
in
saying.
Well,
that's
all
right,
but
because
now
she's
labeled
me
as
an
old
timer,
I'm
going
to
use
the
prerogative
of
old-timers
in
my
talk
right.
So
the
first
one
is.
B
These
are
my
thoughts
they're
based
on
many
supercomputers,
many
storage
systems
facilities,
many
many
users
and
science
projects
through
I
come
from
in
being
involved
with
three
world-class
organizations
in
high
performance
computing
and
having
worked
in
industry,
national
centers
governments
and
having
many
relationships.
So
the
first
thing
is
these:
are
my
thoughts
and
they're
not
scientifically
provable?
B
B
Second,
is
I'm
an
old
timer,
and
I
may
not
remember
things
correctly
and
I'm
going
to
talk
about
things
that
I'm
interested
in
or
that
I
remembered
over
the
last
couple
days,
particularly
flying
here
last
night
and
don't
take
that,
to
the
mean
any
contributions,
any
other
organizations
have
made.
It
is,
after
all,
nurses,
birthday
anniversary,
so
we're
going
to
talk
about
nurse
and
ignore
all
the
other
contributions
that
have
been
made
for
the
most
part.
B
The
next
thing
is
to
me
nurse
is
the
clients
and
users
I
like
to
call
them:
clients
or
partners,
not
just
users.
Users
are
addicted
to
machines.
I
know
some
of
you
are
clients.
You
have
a
long-term
relationship
with
nurse
philosophy.
You
heard
some
of
that
earlier
today,
the
staff
and
the
resources,
the
physical
resources.
B
You
notice
that
that's.
This
is
a
little
bit
in
priority
order
where
the
resources
are
not
the
primary
thing.
Even
though
we
all
talk
about
how
many
times
we've
been
on
some
list
or
how
many
aggregate
floating
point
operations
we
have.
B
The
funders
deserve
an
awful
lot
of
recognition
as
well
they're
incorporated
when
I
say
the
term
nurse.
The
leaders
going
back
all
the
way
to
john
clean
all
the
way
up
through
all
the
leaders
today,
the
academic
partners
like
berkeley
like
davis
and
like
some
of
the
other
partners,
and
all
the
associates
collaborators
and
friends.
So
when
I
say
nurse,
I'm
talking
about
the
big
picture
nurse,
not
just
simply
machines,
I
was
at
supercomputing.
A
B
And
that's
what
he
said
he
caveated
computing
to
say:
keep
capacity
computing
in
the
world.
That
was
a
political
thing
that
was
on
his
the
text
of
his
slides.
That's
not
what
he
said.
So
you
remember
that,
but
that
is
actually
quite
an
accomplishment
and
someone
else
chimed
in
in
the
universe
rather
than
just
the
world
at
that
session.
B
So
I
thought
maybe
it
takes
a
moment
to
have
an
historical
perspective,
not
just
of
nurse
and
what
systems
or
where
nurse
happened
to
be
and
what
it
was
titled
or
labeled
at
the
time.
But
how
does
it
play
in
the
general
community
of
high
performance
computing
and
I
restricted
myself
to
the
u.s
high
performance
areas,
so
in
the
beginning
there
was
nursk
and
nurses
mfb
cc
or
whatever
cut
it
off
a
little
bit,
but
by
nursing
and
es
net.
B
I
don't
talk
too
much
about
esnet,
but
they
certainly
play
a
significant
role
too.
The
stars
are
notable
things
we
can.
I
won't
go
into
each
one
of
those
things
that
I
thought
was
notable
and
then,
from
my
perspective
next
was
nasa
in
particular
nasa
ames
and
I'll.
Tell
you
why?
I
think
that
was
next
in
the
next
slide,
or
so,
but
the
next
major
facilities
that
are
national-based,
science-based
facilities
that
are
open
to
some
multiple
sets
of
communities
is
what
I'm
looking
at.
B
Then
the
nsf
followed
soon
after
nasa,
with
a
set
of
five
centers
down
to
three
centers,
then
added
a
few
more
centers
and
whatever,
but
also
in
there
is
in
car
tom,
is
here
incar
actually
predates
nurse.
That's
the
only
the
only
program
I
found
that
predates
nurse
in
terms
of
a
computing
facility,
but
it
is
a
single
disciplined
domain,
at
least
for
the
most
part,
where
nurses,
multiple
domains.
I
think
that
that's
important
later
on
and
then
the
dod
high
performance,
modernization.
A
B
Comes
in
and
the
classified
computing,
ascii
and
its
other
activities
about
that
time
period
and
then
the
leadership
facilities
for
the
office
of
science.
So
this
is
roughly
my
my
time
frame
everything
in
the
gray
arrow.
There
are
always
center-based
activities
or
small
activities
or
experimental
activities
going
on,
but
these
are
national
infrastructure,
national
facilities
for
people
to
do
the
science
as
opposed
to
to
studying
things.
B
B
You'll
see
in
my
comments,
but
basically
from
what
nurse
had
started,
the
open,
interactive,
multi-disciplinary
support
for
many
things
from
there
directly
had
major
influences
both
on
the
nasa
program
where
it
started
with
nas
around
center-based
machines
and
the
nsf
program.
So
if
you
go
back
and
look
at
those
programs,
they
derived
much
of
their
philosophy,
much
of
their
approach
from
what
nurse
was
doing
earlier,
and
this
is
probably
one
of
the
most
important
contributions.
B
B
Same
type
of
philosophy
here
in
the
dod,
but
then
these
groups,
just
like
everything
else
in
science
and
engineering,
the
cascading
effect
of
what
is
discovered,
then
cascades
into
other
things
discovered
or
revealed,
and
these
programs
became
very
strong
and
others
modeled
off
of
that.
So
I
think
that
this
is
an
important
thing
that
that
we
should
keep
in
mind
about
the
influences
that
nurse
has
had.
B
So
I'm
going
to
go
through
my
view
of
some
of
the
great
legacies
of
nurse
over
this
year
and
the
first
is
it
was
the
first
multi-discipline
nationally
based
hpc
resource
for
diverse
and
open
science.
So
some
of
the
changes
that
we
talked
about
this
morning
or
this
afternoon
about
moving
from
livermore
computing
central
computing
on
forward
was
to
make
it
open
and
a
little
more
open.
B
Only
local
centers
were
before
that
mf
ecc
and
his
precursors
started
to
open
up
with
some
of
the
connections
we
heard
jed's
wherever
he
is
opera
connection
that
that
was
a
subrosa.
B
But
in
most
cases
there
were
local
centers
that
maybe
you
could
dial
into
with
a
modem
or
something
like
that.
Contemporary
shared
supercomputing
was
a
single
discipline.
In-Car
even
mfecc,
where
it
was
originally
envisioned
before
it
started
to
open
up
with
our
triple
piece,
was,
was
focused
on
a
discipline
and
what
nurse
did
is
taught
the
rest
of
us
how
to
do
that
type
of
computing.
B
They
certainly
engaged
with
nasa,
and
that
was
my
first
interaction
with
nurse
about
that
program,
as
it
was
starting
up
in
the
early
1980s,
nsf
dod
and
obviously
then
now
international
centers,
so
nurse
was
was
first
there
and
has
pursued
that
and
expanded
it
over
and
over
again.
B
Next
legacy
the
first
center
to
commit
production
for
mpp
or
whatever
you
like
to
call
it
highly
parallel,
distributed
computing.
Whatever
you
like
to
call
it,
when
the
t3
came
here,
nurse
had
made
the
decision
with
doe
that
parallel
computing
was
the
production
computing
of
the
future.
Almost
every
other
place
around
was
experimenting
with
it.
Trying
it
doing
things
like
we
heard
about
the
t3d
at
nasa
ames,
we
had
an
an
intel,
ipsc
860,
a
connection
machine,
a
couple
other
parallel
machines
on
ibm
sp.
Even
but
they
were
not
the
production
resource.
B
So
the
t3
was
that
machine
again,
that's
another
thing
to
thank
livermore
and
mike
mccoy
for
because
he
was
the
one
that
figured
out
how
to
do
that
contract
along
with
lin
back
there
and
daryl
hammer
and
then
quickly
switched
it
to
from
a
t3d
to
a
p3e
that
arrived
here,
and
it
was
the
first
major
production
resource
that
at
least
I
know
of
and
then
with
the
user
community
nurse
was
committed
to
going
down
that
path.
We
they
did.
B
We
did
have
some
vector
machines,
but
within
about
a
year
of
the
full
p3
being
here,
all
of
the
disciplines,
fusion,
all
the
way
through
high
energy
physics
through
astrophysics,
all
the
disciplines
had
more
time
and
we're
doing
more
computing
in
the
parallel
system
than
on
on
the
more
traditional
vector
systems,
and
now
we
know
there's
only
that
style
computing
in
the
supercomputing
environment.
Until
we
get
the
maybe
clouds.
B
B
That
was
the
precursor
to
also
applying
linux
or
I'm
sorry
unix,
which
we
did
at
nasa
ames,
but
several
of
the
nsf
centers
use
ctss,
as
well
as
a
number
of
other
sites,
including
some
industrial
sites.
Because
of
this,
and
realizing
it
was
more
productive
and
the
exceptional
customer
service
that
was
commented
on
is
also
an
integral
part
of
nurse.
It
wouldn't
be
just
having
machines,
nobody
expects
it
there,
although
in
many
other
situations,
many
other
organizations
you
just
have
the
machine
and
people
suffer
through
trying
to
make
use
of
it.
B
So
nursing
is
a
standard
candle.
I
know.
Standard
candles
are
a
special
term
here
at
berkeley
with
supernovas
nurse
gives
a
standing
candle
for
quality
customer
service
and
in
the
high
end
computing,
but
we
also
know
that
takes
a
lot
of
work,
hard
work
and
innovation
to
provide
that.
So
it's
not
just
the
innovation
comes
from
the
next
generation
ship
innovation
comes
from
actually
helping
and
figuring
out
how
to
make
that
set
of
technology.
B
The
best
it
possibly
can
be,
and
almost
all
those
methods
are
adopted
in
the
doe
leadership
centers,
as
well
as
nurse
being
the
place
where
people
come.
If
you're
going
to
do
a
document
on
best
practices
for
how
to
have
a
center
or
how
to
provide
facilities,
nursing
is
a
place.
Everybody
needs
to
come
to
find
out
how
to
do
that.
B
Other
great
legacies
is
a
whole
variety
of
software,
either
improvements
or
basic
software
that
come
from
nurse
and
it's
associated
other
parts
of
activities
be
that
fully
within
nurse
or
being
associated
with
nurse
like
some
of
the
research
organizations
here
or
at
livermore,
the
ones
I
thought
of
were
the
enhancements
to
mpi
and
its
use,
converting
from
or
moving
the
community
from
pvm
to
mpi
data
management,
alright
shoshone
and
its
implementation
in
all
the
large
data
activities
that
go
on
visualization
adaptive
and
load
balancing
techniques
like
amr
the
I
o
and
storage
activities,
we'll
talk
a
little
bit
more
on
programming
models,
upc
other
types
of
things.
B
Cyber
protection,
I
think,
is
one
that,
maybe
we
don't
think
of
all
that
much,
but
I
think
at
least
from
my
view.
There's
no
place
better
than
berkeley
and
nursed
in
terms
of
cyber
protection
for
large
machines
that
don't
interfere
with
how
you
can
use
those
machines
and
then
obviously
lots
of
applications,
astrophysics,
microwave
background
radiation,
supernova,
all
those
other
things,
climate
materials
and
goes
on,
and
the
contributions
to
those
applications
that
were
made
by
nurse
staff
and,
along
with
the
science
teams,
are
all
great
legacies.
B
So
I
I
submit
to
you
that
nurse
merged
big
compute,
with
big
data
before
we
even
know
knew
what
those
terms
were,
and
nurse
has
been
doing,
that
for
at
least
a
decade
and
a
half.
If
not
longer.
We
all
know
that
in
high
performance
computing.
We
all
think
that
we
handle
big
data
along
with
a
big
compute.
That's
true!
B
That's
inherent
in
some
of
the
activities
of
simulation
activity,
but
the
fact
is,
I
think,
nurse
who
was
the
first
place
that
merged
high
performance,
simulated
simulation
computing
or
computational
science
with
high
performance
data
analytics
of
observed
data.
B
Be
that
bringing
the
observations
in
for
studying,
supernova,
detecting
supernova
or
for
the
pdsf
high
energy
physics
and
nuclear
physics
activities,
but
they
were
merged
here,
and
I
think
that
was
probably
first
in
any
organized
manner,
with
an
infrastructure
with
an
activity
that
meant
to
merge
those.
In
fact,
they
mentioned
the
t3
and
the
pdsf.
B
They
actually
arrived
here
at
berkeley.
On
the
same
day,
all
of
the
craig-
and
I
I
should
say,
the
all-
the
the
craig
components
from
livermore
right
here
with
the
pdsf
on
the
same
day.
They
arrived
in
one
truck
for
all
the
cray
stuff
and
then
four
trucks
for
the
pdsf
all
taken
out
of
the
the
superconductor.
B
So
they
arrived
from
the
very
beginning
and
from
that
day
on
this
synergy
of
data
and
computation
data
analytics
and
computation
has
been
an
important
thing.
That
nurse
has
pioneered
the
hpss
collaboration
of
which
nurse
is
one
of
the
developing
partners.
The
global
unified,
parallel
file
system
that
was
generated.
The
first
ideas
for
that
were
generated.
A
B
The
experiences
with
the
t3
when
we
had
two
three
t3s
of
people
remember:
we
had
a
small
one
and
a
big
one
for
a
while
and
turned
out.
We
thought
people
would
get
on
the
small
one,
do
development
interactive
and
then
they
would
move
to
the
big
one
for
production
runs.
We
realized,
after
several
months
that
nobody
was
moving.
B
They
had
two
separate
file
systems
and
people
did
not
want
to
go
through
the
hassle
of
moving
their
data
back
and
forth
between
the
two,
so
the
ones
people
that
started
using
a
small
one,
stayed
on
that
and
went
all
the
same
services
about
a
year
later.
Those
two
machines
were
merged
into
a
single
machine
for
that
purpose,
but
that
was
the
spark
that
said
gee.
This
idea
of
having
a
global
name
space
is
important.
B
Everybody
else
did
it
with
the
network
file
system,
slow,
unreliable,
insecure
and
nurse
it
envision
that
with
a
highly
parallel
high
performance
file
system
and
it's
emulated
elsewhere.
Sometimes
we
ship
people
off
and
let
them
stay
a
while.
It's
placed
like
oak
ridge
and
then
have
them
come
back.
Other
places
have
incorporated
these
ideas
as
well,
but
I
will
say
that
also
in
the
same
time
period
many
fail
trying
to
do
the
same
thing
and
nurse
succeeded.
B
So
more
so
deep,
zero
to
slide
more
data
comes
to
nurse
than
leaves
that's
not
new.
That's
been
going
on
for
a
long
time.
The
other
thing
that's
important
is
at
least
in
my
time
here.
A
high
percentage
of
that
stored
data
is
actually
used,
which
is
unusual.
Everybody
says:
oh,
it's
the
closet,
you
throw
it
away
and
never
think
about
it
again.
It's
in
your
basement.
B
It
may
be
different
now,
but
we
looked
at
that
and
a
lot
of
that
data
is
actually
used
at
least
once
and
it's
useful.
So
this
merger,
now
what
maybe
people
call
bdac
or
or
all
types
of
things
actually
has
been
going
on
for
15
18
years,
that
nurse
the
focus
on
real
performance
is
another
legacy
real
time.
The
solution
equals
real,
sustained
performance.
B
The
system
choices
going
way
back
in
livermore
days,
we're
all
based
on
what
really
is
driven
by
the
science
needs
in
getting
that
work
done,
and
I
think
that
that
while
I
like
to
think
other
organizations
do
that.
Well,
I
think
that
at
nasa
we
did
something
very
similar,
but
I
think
for
consistency
and
beginning
to
do
that.
Nurse
has
been
one
of
the
leaders
for
that,
and
also
some
of
the
methods
here
have
been
adopted
by
many
others,
and
I
think
that
that
will
continue
with
that.
B
Another
thing
I
think
nurse
paves
the
way
for
shows
is
how
to
express
the
impact
of
the
use
of
the
resource
for
the
real
science.
So
this
is
the
thing
that
we
struggle
with
in
ncsa
and
nsf
space.
How
do
you
show
the
impact
of
what
you're
really
doing
as
opposed
to?
If
you
didn't
have
any
of
those
resources?
B
Nurse
has
done
an
outstanding
job,
and
now
lots
of
people
emulate
that
in
trying
to
show
this
very
hard
thing
right,
so
we
all
have
the
traditional
methods
of
citations
or
papers
or
whatever
some
of
those
are
volume
based
the
more
users
you
have
the
more
papers
you'll
have,
but
how
impactful
are
they?
Well,
we
think
about
citations
right.
They
take
a
long
time
to
lead
up
to
nurse,
has
a
wonderful
history
of
showing
that,
but
also
the
uniqueness.
B
The
fact
is,
what
you
have
is
unexpected
insights,
as
well
as
the
planned
insights
or
the
plan
campaigns.
So
nurse
has
done
a
great
job
with
that.
I
think
many
other
organizations
are
trying
to
do
similar
things,
and
not
only
are
we
trying
to
do
similar
things,
we
also
benefit
from
what
nurse
has
done
and
we
leveraged
that
there.
B
So
this
is
one
of
those
little
stories.
I
don't
know
how
many
people
know,
but
I'll
claim
that
nurse
was
a
primary
mover
to
reinvigorate
the
federal
super
computing
funding
and
particularly
for
open
science.
So
in
about
2004,
there
was
a
meeting
here
with
ray
aubach.
He
happened
to
be
a
friend
and
a
colleague
of
chuck,
chank
and
chuck
asked
him
up
here
before
he
reported
to
washington
to
become
the
director
of
the
office
of
science
and
a
few
of
us
bill.
B
Mccurdy
haas
myself,
I
think,
were
the
ones
in
in
the
room
with
chuck
talked
for
about
half
a
day
with
ray
about
the
importance
of
high
performance
computing,
its
issues,
its
motivations,
its
benefits,
and
this
was
surrounding
the
time
period
about
when
the
earth
simulator
was
coming
out.
But
it
was
all
focused
on
the
science,
not
the
top
500.
I
don't
think
we
ever
mentioned
top
500
in
in
that
meeting
and
it's
hard
to
prove
cause
and
effect.
B
But
we
all
know
that,
since
that
time
period,
federal
budget
and
high-end
computing
has
doubled
more
than
doubled
in
actual
dollars.
And
if
you
look
at
the
open,
computational
science,
because
in
this
time
the
federal
budget
was
dominated
by
classified
work,
particularly
ascii
and
and
some
other
things,
not
the
the
doppler
work.
The
open
science
share
of
the
funding
much
more
than
doubled,
so
something
else
that
nursk
in
berkeley
helped
create.
That's
benefited
more
than
just.
B
The
nurse
site
nerf
systems
are
entirely
open
and
on
the
internet,
that's
a
fundamental
philosophy
for
this
interactive
use
and
the
fact
is,
we
want
to
get
the
maximum
amount
of
bandwidth
and
performance
for
data
transfers
for
everything
else,
but
that
leads
to
the
need
to
still
make
sure
that
they're
protected-
and
I
think,
as
I
said,
nursk
is
a
leader
in
this,
along
with
berkeley
for
cyber
protection,
I
like
to
call
it
cyber
protection
around
security.
B
Someone
from
berkeley
actually
asked
me,
would
I
rather
have
my
my
home
in
town
secured,
or
would
I
rather
have
it
protected
and
secure?
I
think,
is
I'd
rather
have
it
protected
right,
so
the
methods
of
grow
and
all
the
other
things
that
go
into
the
cyber
protection
infrastructure
have
led
the
way
we
use
that
at
illinois.
Other
people
use
that
as
well,
and
it
actually
now
has
been
codified
in
the
science
dmz
that
are
being
implemented
with
es
net
nurse.
B
Actually,
there
was
a
paper
at
supercomputing
this
year
made
a
a
very
good
impression
on
lots
of
people.
It's
what
nurse
and
es
net
were
doing
five
or
six
years
ago,
but
now
document
the
monitor
and
react
philosophy
for
major
systems,
so
they're,
not
behind
firewalls
things
like
that.
Make
them
performant
and
easy
to
use
is
something
that's
been
trying
to
be
adopted
by
others.
B
Nurse
influenced
a
lot
of
policy,
particularly
on
those
things
that
would
have
made
more
restrictions
and
berkeley
worked
on
this
to
make
to
maintain
the
openness
and
effectiveness.
But
this.
B
Works,
you
only
have
one
you
have
one
bad
experience
and
you'll
have
lots
of
implications.
So
the
fact
is,
it's
worked
for
15
18
20
years
without
having
a
major
issue
where
many
other
places
have
major
issues
with
security
shows
that
this
methodology
is
a
legacy
that
others
should
learn
from
another
legacy.
B
I
thought
of
is
nurse
ability
to
manage
both
grand
challenge
activities
and
I
don't
like
the
term
capacity,
but
the
use
of
the
systems
for
many
many
projects,
plus
a
few
very
large
important
projects
that
takes
a
particular
skill
at
managing
that
and
working
with
the
clients
to
make
sure
that
that
works,
persistent
support.
B
Another
legacy
is
part
of
that
is
nurse.
Gotcha
started
the
d.o.e
insight
program,
so
we
were
the
first
place
that
actually
started
inside
allocations.
The
first
one
happened
to
be
that
I
remember
at
least
were
colliding
black
holes
at
seidel.
Did
that
and
ray
armbach
gave
him
a
special
allocation
of
what
became
insight
to
get
away
from
just
doing
lots
of
small
things
he
wanted
to
do
grand
things
as
well,
small
world
and
now,
as
my
boss
in
ncsa.
B
So
that's
the
first
time
I
met
my
new
boss
and
it
was
about
15
years
ago,
but
it
shows
great
steps
forward
can
be
done
while
giving
large
resources
to
a
set
of
well-qualified
teams.
So
it
proves
that,
but
it
also
meant
that
you
can
manage
very
large
work
without
also
being
very
disruptive
to
to
the
general
there's,
not
a
good
term.
Smaller
jobs
or
regular
jobs,
but
nursk
is
able
to
do
that
on
a
day
in
day
out
basis,
and
it
makes
a
very
big
difference.
B
The
other
thing
is,
I
don't
think,
there's
a
better
place
at
the
moment
for
people
in
that
this
middle
range
of
being
able
to
scale
up
but
not
having
enough
time
or
resources.
There's
not
a
better
place
or
advocate
than
nurse
is
for
the
missing
middle,
as
it's
sometimes
called.
B
Science
first
has
always
been
oops.
This
is
a
a
repeat
of
that
one.
Sorry
too
much
cutting
paste.
I
would
also
submit
that
nurse
reinvigorated
high
performance
computing
at
berkeley.
When
we
came
to
berkeley
in
96
years.
I'm
sorry,
96
berkeley
did
not
have
a
single
allocation.
It
wasn't
a
pi
with
a
single
allocation
on
nurse
about
five
years
later
nurse
had
the
majority
of
or
the
largest
allocation
share
by
institution
there,
and
it
was
through
not
just
the
fact
the
machine
is
here.
B
I
don't
think
that
that
was
a
major
major
issue
with
that.
It
was
through
the
programs
that
I
think
peter
alluded
to
the
ldrd
funding
that
berkeley
put
in
to
join
nurse,
to
build
up
a
computational
science
expertise
at
berkeley,
knowing
that
that
was
going
to
be
the
future,
maybe
more
so
than
experimental
theory,
at
least
at
the
moment.
So,
within
five
years,
berkeley
had
the
majority
by
institution
of
nurse
allegations
and
built.
B
The
strong
programs
and
the
synergies
have
enabled
new
discoveries
and,
as
you
heard
at
least
one
nobel
prize
came
from
that
type
of
synergy,
so
that
legacy
also
is
an
important
one
for
nurse.
B
So
you
can't
talk
about
nurse
without
talking
about
vendors.
Jeff
is
going
to
talk
from
from
perspective,
and
we
all
know
that
relationships
between
facilities
and
organizations
like
nurse
and
in
the
vendors
that
supply
the
technologies
have
a.
I
want
to
put
up
there.
A
love
hate
relationship,
maybe
it's
a
give,
take
relationship,
it's
a
nice
way
to
do
it.
B
The
best
value
processes
that
were
started
at
livermore
and,
I
think,
have
been
pushed
a
lot
at
berkeley
and
nurse
for
a
holistic
evaluation
is
actually
something
the
vendor
community
likes
from
everything.
They've
said
they
may
be
being
nice
to
us,
but
from
everything
they
say
they
like
that
much
more
than
the
very
prescribed
methods
of
acquiring
machines.
Before
nurse
help.
Many
vendors
get
better
get
good
all
the
way
up
through
many
variations
of
it
of
cry.
B
Vectors,
mpps
ibm
the
things
I
thought
of
were
the
sp
at
scale
and
then
making
the
sp
bigger
twice
as
big
as
it
was
before,
finding
out
how
that
worked,
the
power
ih
architecture,
starting
with
the
power
5
gps,
hpss
networking
vendors,
sell
all
those
where
there's
been
explicit
help,
and
I
also
know
storage
vendors
and
I
didn't
remember
all
their
names,
the
only
one
that
stuck
in
my
mind,
was
yada
yada.
B
B
Those
were
hard.
They
were
disruptive
to
at
least
some
parties,
but
they
ended
up
in
retrospect
to
be
the
right
decisions
for
the
hbc
community,
and
everybody
knows
that
that
nurse
is
honest,
fair
and
gives
a
meaningful
opinion
or
evaluation.
B
So,
while
you'll
always
have
this
give
and
take
in
terms
of
the
individual
interests
between
the
vendors
and
the
organizations
and
the
science
community
in
general,
I
think
there's
been
a
lot
of
strengths
and
a
lot
of
benefits
coming
from
nurse
in
this
use
of
in
interactive
our
interactions
with
the
vendors
not
too
long
ago.
B
Again,
thinking
about
how
impact
has
done
a
an
expert
in
evaluating
things,
told
me
that
a
measure
of
impact
is
this:
dispora
the
spread
of
alumni
from
organizations
and,
in
many
cases,
those
attract,
particularly
in
in
some
other
fields.
Where
do
people
go
where?
Where
are
they
having
continued
impact
based
on
what
they've
acquired
the
the
experiences
they've
acquired?
Well
first,
I
want
to
point
out
that
nurse
benefited
from
this
from
others.
B
When
we
came
here
to
berkeley,
I
think
we
were
able
to
get
best
practices
from
the
entire
world
because
we're
able
to
hire
the
people
that
were
doing
that
as
a
practice
from
other
places
like
francesca
like
david,
like
harvey
and
all
the
other
folks
that
nurse,
so
we
accumulated
that
at
nursk
and
that
benefited
us
tremendously,
but
there's
also
the
spread
of
people
who
have
been
a
nurse
and
gone
on
to
other
places.
I
would
say:
shane
brought
a
lot
to
oak
ridge
when
he
went
there
and
would
lay
days
back.
B
Other
people
have
as
well
and
that's
an
important
impact
for
the
community
and
and
for
furthering
science
as
well.
So
that's
a
thing
to
keep
in
mind.
So
my
summary
is
that
nurse
was
the
first
nurse.
Is
there
the
longest
nurse
is
open
science
at
unprecedented
scales
and
impact
nurse
is
the
perennial
leader
in
our
community
and
the
nation
doing
and
the
super
computing
community
should
not
only
be
proud
but
thankful
for
nurse.
B
So
that's
what
I
want
to
leave
you
with,
because
it
is
true
and
then
here's
paraphrasing
ray,
I
will
say
nurse
is
today
and
will
be
in
the
future.
One
of
the
two
best
run
hpc
computing
facilities
in
the
world,
I'll,
let
you
guess
what
the
other
one
and
what
the
other
one
is.
That's
on
my
mind,
so
happy
anniversary
and
thank
you
very
much.
A
Oh
wow,
thank
you
bill.
Anybody
have
any
questions
or
comments
for
bill.