►
From YouTube: Node.js Benchmarking WG Meeting - June 5th 2018
Description
A
Okay,
so
we
should
be
live
so
welcome
everybody
to
the
node.js
benchmarking
meeting
for
June
5th
2018
and
we're
gonna
follow
our
standard
agenda,
so
I'm
just
gonna
pick
that
up
find
that
hold
on
open
up
the
minutes.
Talk
if
people
can
hear
I'll
paste
the
link
into
the
chat
if
everybody
could
add
themselves.
That
would
be
great.
A
Okay,
if
no
announcements
then
on
to
the
issues
extracted
by
looking
at
the
benchmarking
and
gender
tag
across
the
repos,
the
first
one
is
brainstorming,
capturing
and
documenting
areas
where
people
can
contribute,
but
I
think
we
should
leave
that
to
the
end,
because
we
can
probably
use
whatever
time
we
have
there.
The
next
one
on
the
agenda
was
footprint,
increase,
acne
rafter
load
that
one
again
we've
talked
talked
about
for
the
last
few
times.
A
B
Performance
or
to
whatever
the
number
the
size,
but
some
I'm
not
able
to
reproduce
what
we
see
on
the
web
page
and
I
don't
have
access
to
any
of
our
benchmarking
machines.
Maybe
I
forgot
the
process
because
I'm
not
accessed
for
now.
I
would
like
to
get
an
accent
maybe
a
day
for
a
couple
days
and
then
I
try
it
on
there
sure
yeah.
A
A
A
Have
people
who
have
access
to
that
benchmark
key
be
able
to
log
in
in
general?
Otherwise
you
know
individual
keys.
We
remove
after
the
need
has
passed,
I
see,
okay,
so
it's
it
just
hasn't
been
pressing
enough
for
me
to
actually
set
up
the
other
one.
So
I
think
for
this
one
again,
I
can
just
give
you
access.
You
know
for
a
defined
period
of
time.
It
yeah.
A
A
B
B
A
B
A
A
Okay,
so
I'll
coordinate
with
you
to
get
access
in
the
end.
What
I'll
need
is
your
SSH
key
yeah.
So
if
you
can
send
me
one
through
email
will
work
will
work
to
get
you
access
there.
So
you
can
see
if
you
can
recreate
there.
That's
good.
Okay,
the
next
one
was
add,
ghost
GS
workload
to
the
benchmark.
I
think
that
was
your
next
thing,
though
right
so
maybe
yeah.
B
So
there
was
one
problem:
why
I
wasn't
able
to
contribute
till
now
so
goes
by
default,
runs
in
to
the
ghost
runs
in
two
modes:
one
is
a
dev
mode
development
mode,
and
there
is
one
in
production
mode
and
I
was
last
couple
weeks
ago.
I
was
just
doing
the
analysis
and
I
found
that
the
performance
or
the
perform
production
mode
is
almost
twice
that
of
a
dev
mode,
which
is
a
default
mode
for
our
testing.
B
C
B
C
B
B
A
A
B
A
I
think
it
makes
sense
at
this
point
to
try
and
start
using
docker
I
can't
I
think
we
talked
a
little
bit
about
it.
Laughs
I,
didn't
think
having
doctor
installed
would
actually
cause
any
performance
degradation
to
the
other
things
like
I
guess
is
the
demon
itself
running
gonna
be
an
issue.
Was
the
question.
B
A
B
B
A
A
A
B
D
A
Not
necessarily
the
right
answer,
but
we
should
I'm
just
thinking
once
we
have
doctors
like
as
opposed
to
installing
things
on
the
machine,
the
more
we
sort
of
use
the
containers
to
avoid
pollution,
probably
the
better
okay,
but
that
if
it's
just
this,
who
knows?
Maybe
it's
already
there
too,
if
it's
part
of
a
standard
install.
So
we
can
take
a
look
at
that
first
yeah.
B
D
E
A
C
A
Okay,
so
this
one
here
is
a
J
is:
is
kind
of
working
across
the
repo
to
help
pull
together
things
that
people
can
get
involved
in
and
to
start
working
on
when
they
join
the
community,
and
so
we'll
kind
of
done
doing
this
in
a
few,
the
different
working
groups
to
sort
of
brainstorm,
a
list
of
things
that
we
could
think
of
where
people
could
jump
in
and
participate
I
mean
so
maybe
a
did.
You
want
to
give
a
bit
more
of
an
introduction
than
that
sure.
C
So
I
can
directly
direct
those
people
towards
whatever
items
we
can
get
on
that
list
and
they'd
be
able
to
help
with
those
specific
tasks.
The
different
working
groups
or
initiatives
come
up
with.
In
the
meantime,
we're
kicking
off
a
our
first
cohort
for
the
mentorship
team
on
June
15th
tentatively,
or
is
that
right,
yeah
and
we
hope
to
have
some
some
items
on
the
list
buy
them
to
show
those
people
soon.
C
A
A
See
if
we
can
get
any
comments,
okay,
so
I
guess,
the
idea
is
to
spend
a
bit
more,
that
you
know
the
time
that
we
have
now
to
kind
of
brainstorm
things
that
we
think
on
the
benchmarking
working
group
front.
We
could
get
new
people
who
are
joining
to
help
out
with
now.
Obviously
there's
the
right
new
benchmarks
based
on
the
matrix.
You
know.
So,
let's
add
that
in.
A
And
this
can
kind
of
be
the
ideas.
This
could
kind
of
be
like
a
wish
list.
So,
even
though
you
know
one
of
the
existing
members,
don't
doesn't
have
time
to
jump
in
and
do
that
it's
still
good
to
have
kind
of
a
wish
list.
So
the
other
we'll
have
some
time
they
might
be
able
to
see
something
interesting
there
and
jump
in
how.
A
C
A
B
A
Think
the
idea
here
is
that
I
mean
as
much
details
we
can
add
would
be
good,
but
hopefully,
as
long
as
it's
enough
to
for
somebody
to
say,
hey
I
want
help
out
with
that
or
not.
Then
we
could
fill
in
the
ones
that
people
volunteer
for
the
general
discussion
we
had
before
was
like.
We
don't
want
to
if
it
takes
too
much
time
to
add
things
to
this
list,
it'll
be
harder
for
people
to
throw
things
on
the
list.
Yeah.
C
B
C
B
B
A
Okay,
yeah,
like
a
the
problem,
would
be
integrated
into
what
we
already
have
to
say.
Okay,
once
I'm
once
I've
posted
today's
result,
do
I
look
at
the
last
two
results
or
yeah.
Somehow
we
basically
have
to
keep
the
list
of
results
and
and
figure
out
a
delta
when
something's
gone
down
by
more
than
a
certain
amount.
A
A
B
A
C
A
A
A
A
A
C
B
A
B
B
B
A
B
D
B
A
B
D
B
A
A
E
Mean
even
even
picking
a
very
small
list,
for
example
like
pick
in
the
top
five
buffer
benchmarks,
or
something
like
that,
those
getting
some
of
those
into
a
nightly
run.
If
we
can
get
a
set
of
benchmarks
that
only
take
what
five
minutes
temperance
to
know
something.
Then
it's
potentially
the
sort
of
thing
that
you
can
get
something
in
the
chats.
D
A
A
A
A
A
A
E
A
C
A
B
I
was
thinking
that
since
the
Garrett
added
core
benchmarks
for
benchmarking
correct,
he
finished
that
one
of
the
issue
at
127,
but
we
don't
publish
any
of
them.
Is
there
a
way
we?
Maybe
we
can
think
about
what
we're
best
way
to
highlight
that
or
like
there're.
The
run
ad
hoc
right
here
now
like
a
developer,
can
go
and
start
a
run
like
yet.
E
A
E
I
published
it
first
of
all,
I
asked
if
there
was
any
feedback
from
people
who
I
have
a
go,
find
something
that
doesn't
work
or
maybe
it's
not
particularly
clear,
because
here
I
put
it
together.
So
for
me,
it's
really
if
they're
I'm
well
aware
that
yeah,
when
you're
right,
when
you
put
something
together
yourself,
ease
yeah,
sometimes
difficult
to
see.
B
B
A
B
Yeah,
if
there
is
a
date
and
when
we
run
it
Intel
and
we
do
see
some
of
the
variations
and
we
have
caught
couple
of
recreation
internally-
maybe
it
could
be
because
of
the
machine
was
down
or
something
else
was
running.
Sometimes
we
one
particular
v8
was
a
problem.
Something
like
that.
Would
it
be
useful
for
a
benchmarking
side.
Well,
do
you
have
a
subset
like.
B
E
Laughter
I
think
one
of
the
issues
we've
had
a
lot
of
the
lot
of
the
benchmarks.
If
you
just
tell
it
goin
buffers,
it
can
take
many
hours
to
run
I.
Don't
know
that
sounds
like
a
defeatist,
so
I
thinking,
yeah
I
mean
I
fully
agree
that
we
need
getting
something
running
is
better
than
running
nothing
at
all,
because
I
mean
I.
Think
at
the
moment
we're
if
we
were
to
look
at
the
utilization
of
the
machines,
especially
the
benchmarks
ones,
there
they're
not
exactly
running
so.
A
E
I
mean
maybe
we
could
have
a
good
just
pick
out.
I,
don't
know
see
if
see
if
we've
got
an
easy
way
to
yeah
again,
YouTube
is
another
thing:
that's
gonna
take
some
time,
but
maybe
if
we
just
find
a
you
just
go
and
kick
off
you
in
a
benchmark
before
one
now
just
get
up
today.
Thinking
of
how
long
my
actual
folder
takes
to
run
an,
if
is
less
than
an
hour,
have
a
go
at
today,
I'm
running
it
on
a
nightly
basis.
E
B
E
E
A
E
E
E
E
D
A
C
A
E
A
A
E
A
B
B
A
Me
a
benchmark,
see
it
won't
master
right,
yeah,
just
control,
see
I'm
going
to
so
I
think
what
I
could
do
is
create
a
new
job
which
you
could
then
tweak
to
do
what
you
want
there
there's
no
there's
no
easy
way
to
add
a
parameter
or
something
like
that
to
the
existing
jobs
but
yeah.
Let's
just
okay,
create.