►
From YouTube: 9. Q&A and Closing Remarks
Description
From the NERSC NVIDIA RAPIDS Workshop on April 14, 2020. Please see https://www.nersc.gov/users/training/events/rapids-hackathon/ for all course materials.
B
What
I'm
gonna
do
is
suggest
that
if,
if
there's
no
other
immediate
questions
that
people
wanna
ask,
then
we
can
move
to
going
ahead
to
the
closing
remarks.
But-
and
I
don't
have
slides
prepared
for
this,
but
there
are
some
things
I
want
to
make
sure
to
to
touch
to
touch
on
one
of
court.
B
The
first
one,
of
course,
is
the
survey
that
laurie
has
cut
and
paste
in
there,
and
I
notice
it
doesn't
render
as
a
link
so
you'll
have
to
paint
it
and
then
copy
it
over
in
order
to
be
able
to
use
it.
But
do
please
fill
that
out
we'll
be
following
up
with
an
email
for
you
to
fill
that
out
with
another
one.
B
Is
that
again
for
the
training
event
today
there
were
notebooks
for
you
to
test
things
out
with
yourself
and
you
can
run
them
on
the
query,
gpus
and
if
you
didn't
get
a
chance
to
do
it
before
the
event
today
or
during
the
event
today,
then,
if
you
don't
otherwise
have
gpu
access,
then
you
should
be
able
to
keep
keep
that
access
until
april
20th.
So
that's
about
a
week
from
now
to
try
those
out,
so
we
encourage
you
to
do
that.
B
If
you
have
any
questions,
you
know
just
just
email,
lori
or
me.
B
Next
thing
I
wanted
to
say
of
course
was:
I
wanted
to
really
express
our
gratitude
from
nurse
to
the
presenters
today
from
nvidia,
not
just
for
the
work
that
they've
done
and
looking
at
our
workflows
and
using
that
to
inform
their
product
development
and
also
give
us
feedback
about
the
workflows
like
you
saw
today,
but
also
their
their
presentations
and
the
time
that
they
took
to
prepare
those.
B
I
know
that
they
that
you
do
do
this
a
lot,
but
you
did
incorporate
into
the
things
you
were
showing
today
stuff
that
was
really
specific
to
our
workflows
and
really
important
to
the
kind
of
concerns
that
our
users
have.
So
I
I
want
to
express
that
on
gratitude
again.
B
A
Short
question
sure
regarding
the
cuml
notebooks:
where
is
the
parallelism
happening
in
that?
So
these
are
scikit-learn,
you
know
regression
or
workflows.
Where
exactly
is
the
parallelism
happening?
Is
it?
Is
it
data
parallelism.
C
So
in
the
in
the
notebooks
there's
parallelism
happening
all
over
so
not
with
scikit-learn
necessarily
but
with
the
qml
and
the
gpu-based
algorithms.
The
parallelism
is
is
happening
all
over,
so
there's
only
one
machine,
so
there's
already
a
model.
It's
like
there
isn't
model
parallelism
because
there's
only
one
model
here,
because
there's
one
there's
one
worker
one
machine,
and
so
it's
it's
just
a
parallel
in
the
really
in
the
linear
regression
one.
I
forgot
this
whatever
it
depends
on
the
solver.
C
It's
either
a
parallel
closed
form
solution
or
it's
a
parallel.
I
guess
I
can
decomposition
solution.
I
forgot
what
it
was,
but
it's
it's
just
a
parallel
algorithm
on
a
single.
A
D
That's
correct.
I
just
wanted
to
add
to
that
you're
right
nick
it
was
the
eigen
decomposition
solver
and
the
notebook
example,
and
it's
just
apparel
algorithm,
and
we
didn't
have
that
example
in
this
notebooks,
but
we
have
other
ones
for
das
ml2.
B
I
think
the
the
first
presentation
had
a
final
slide
that
included
some
handy
links
and
we'll
make
sure
if,
if
we
haven't
got
that
one
already
that
we
will,
that
will
include
that
one
in
there.
Okay,
thanks
thanks,
sarah
okay.
B
So
with
that,
I
think
what
we're
gonna
do
is
we're
gonna
shut
off
the
recording
and
then
we're
gonna
go
ahead
and
let
everybody
go
a
few
minutes
early
here
it's
been
a
long
day.
I
want
to
thank
everybody
for
putting
up
with
zoom
and
doing
this
remotely,
and
I
hope
that
people
found
it
useful
and
that
they're
able
to
use
the
the
notebooks
and
materials
that
have
been
shared
with
them
today
in
their
future
workflows.