►
From YouTube: Node.js Benchmarking meeting
Description
A
Ok,
this
is
the
nodejs
benchmarking
meeting
for
September
21st
and
in
attendance
is
myself
and
Gareth.
So
far
there
are
a
few
other
people
who
did
indicate
they
could
make
this
time
but
figured
we
should
get
started
anyway
in
terms
of
the
agenda.
Think
we'll
just
go
through
a
quick
round
table
then
dive
into
the
agenda
items
and
go
from
there.
A
So
I
guess
I'll
start
haven't
done
too
much,
since
the
last
meeting
have
looked,
did
help
out
to
fix
an
issue
with
the
the
acne
or
benchmarks
that
were
running
you
know
with
with
Jerris
health.
We
fix
some
scripts
that
had
stopped
working
due
to
changes
in
one
of
the
module
dependencies.
That's
about
it
for
my
side,
I
guess.
A
No,
the
other
thing
I
also
did
was
I
landed,
the
the
matrix,
which
shows
are
our
youth
cases
and
key
runtime
attributes
versus
the
benchmarks
that
we
already
have,
which
I
want
us
to
use
as
discussion
in
terms
of
you
know
where
we
focus
on
filling
in
additional
test
cases.
I
Gareth
over
to
you,
yep.
B
So
I've
been
working
on
a
few
different
things,
so
the
first
one
but
I've
pull
request.
Earth
info
at
the
moment
in
the
benchmarking
work
group
in
a
prototype
will
be
now
to
run
a
subset
of
benchmarks
from
the
node
sauce.
So
that's
the
benchmark,
folder
in
nodejs,
/,
node
and
Gibson
is
actually
trying
that
out
today.
I
hope.
So
very
you
get
a
feedback
if
it
worked
okay
and
we
could
potentially
have
a
look
at
seeing
if
we
could
get
that
into
Jenkins.
B
B
B
B
B
So
I
mean,
if
you
have
a
look
at
the
chart,
the
dust
do
seem
to
go
up
and
down
quite
a
lot
now.
Obviously,
those
are
running
all
different
versions
of
node.
So
does
it
chance
that
changes
that
we're
making
to
node
are
impacting
it,
but
from
the
so
the
range
of
scholars
that
I
was
getting
and
so
I
mean
the
numbers
that
we
get
from
the
benchmark.
It
gives
us
the
number
of
operations
per
second
were
able
to
achieve
so.
B
The
sort
of
range
I
was
getting
was
on
the
cached
job
from
240,000
operations,
a
second
look
to
290,000
and
then
for
the
new
one.
So
this
is
where
we're
going
clear,
the
cache
we
get
between
thirty
thousand
and
thirty
three
thousand
operation.
Second,
so
it's
a
fair
range,
but
again
that
that
was
one
that
was
on
my
on
one
of
our
local
machines.
So
I've
not
had
a
go
on
the
bench
marking
machine.
There
may
be
something
that's
causing
laughter
yeah.
A
B
But
now
I
mean
perhaps
if
Michael
was
on
yeah
michael
paulsen,
if
he
was
on
the
call
he
made,
we
seized
their
only
guest.
He
may
have
had
a
bit
more
experience
than
I
have,
because
it's
a
apart
from
looking
at
the
results
have
not
done
too
much
with
it.
Apart
from
the
last
few
days,
right,
okay,
but
as
part
of
the
pr
there
was
a
comment
that
it
could
be
useful
to
get
a
module
that
has
some
native
code
included
as
well.
B
C
B
So
I
think
it
was
flying
that
acid,
nodejs
benchmarking
and
now
we've
been
added
to
the
group.
It
came
through
as
a
notification
for
Roy,
okay,
I
thought
I
may
as
well
have
a
go
at
doing
that.
That's
good,
okay,
but
yeah
me
nuts.
That's
mainly
what
I've
been
up
to
the
last
up,
lemons,
I,
suppose,
okay,.
A
B
Right,
it's
probably
getting
on
a
year
since
we
discussed
that
first
go
I
wouldn't
know
if
it's
the
sort
of
thing
that
we
just
need
to
accept
the
night's
game
as
I'd
imagined.
The
code
base
now
has
changed
even
more
significantly
since
okay,
the
issue,
so
I
suggest
probably,
if
you're
happy
with
it,
we
can
probably
just
drop
it
up.
Kitra
genders.
Okay,.
A
A
A
Kos
I
went
back
that
the
table
shows,
you
know,
we'd
already
defined
our
ki
use
cases
and
our
key
runtime
attributes
in
the
table
base
that
shows.
Ok,
here
is
the
mapping
between
those
and
the
benchmarks
we
already
have
and
I
guess.
The
question
for
us
to
think
about
is
what's
the
best
way
to
fill
in
the
table
so
that
we
have
coverage
of
more
of
the
use
cases
and
more
of
the
key
runtime
attributes.
A
So
looking
at
the
gap,
so
we
have
some.
You
know
the
first
one
service
oriented
architecture.
We
don't
have
any
coverage
microservice
based
applications.
We
don't
have
anything
generating
serving
dynamic
web
page
content,
I
think
our
Acme
air
operations
per
second
and
act
year,
latency
color
that
Gareth.
B
Would
you
groove
that
yeah
and
one
of
the
tests
we
could
potentially
do
just
as
a
more
basic
serving
web
page
content
is
just
to
put
together
a
fairly
simple
by
hello
world
type
application
in
express
great
I?
Never
I've
already
got
some
Jamie
to
scripts.
We
could
run
for
that,
so
I
could
go
and
include
those
and
just
to
give
us
another
data
point
for
okay.
B
Acne
I
uses
xpress
so
I
suppose
the
you're,
the
sort
of
thing
that
we
could
do
is
maybe
use
one
of
the
other
framework.
So
that
happier
something
like
that,
possibly
right
and
then
at
least
we're
using
code
I
mean
I'd.
Imagine
probably
under
the
cover
they're
doing
fairly
similar
things
right,
but
I'm
guessing.
A
B
We
don't
have
anything
there
now
we
don't
have
anything
that,
as
as
where
the
HTTP
to
support
is
still
under
development
right.
What
thought
I
did
have
do
is
put
together
something
using
the
current
HTTP
module
and
then,
as
work
is
turn
on
the
HTTP
two
side.
We
can
then
modify
modify
that
to
try
and
have
a
HTTP
one
and
HTTP
to
benchmark
and
have
been
doing
similar
enough
things
that
we
may
be
able
to
compare
rim
games
or
drops
when
moving
to
http
teeth
right.
A
So
that
one
eh
yeah
we
should
look
for.
Is
there
a
web
sockets
benchmark
and
if
so,
maybe
try
and
use
that
otherwise
try
and
write
one
yep,
two
agents
and
data
collectors.
We
don't
have
anything
there
that
one
I
guess
we'd
also
have
to
look
to
see
if
there's
something
out
there,
just
so
that
we
know
what
the
characteristics
are:
yep
a
small
scripts.
A
A
B
B
C
A
Page
summary
for
in
a
start-up
memory,
footprint
afterload
metrics
are
collected
so
so
memory
footprint
startup.
So
we
have
to
start
up
Brenda
see
a
footprint.
Afterload
metrics
are
collecting
okay
rates.
I'm
here
for
product
load,
no
GS
process,
CP
usage.
When
idle,
we
don't
have
anything
max,
Rupert,
so
I
think
acne
or
give
us
some
coverage
there.
B
A
A
B
B
Yeah
so
I
guess
it
depends
how
we're
building
if
we
were
to
build
node
using
the
so
make
binary
option.
That
would
give
us
already
the
install
package
size
I.
Think
if
we're
talking
about
the
other
genes
head.
So
we
could
you
just
modify
how
we're
building
node
at
the
moment
and
then
obviously
unpacking
have
a
look
at
how
large
youngster
what.
B
C
B
C
C
B
C
A
B
So
the
dockerfile
is
there
and
ready
to
go
okay.
It
would
be
good
if
we
I'm
going
out
well
it'd
be
good
if
people
could
look
out
for
potential
problem
report
type
all
requests
where
we
could
be
running
this
to
possibly
at
the
moment.
We've
not
got
the
facility
to
add
it
to
the
inter
Jenkins,
but
it
could
be
the
sort
of
thing
that
someone
could
run
locally
Rick.
B
Potentially,
we
could
also,
if
the
pull
request
for
doing
their
benchmarks
from
the
nerd
sauce
goes
well.
We
could
potentially
doing
something
similar
for
acne
air
and
whether
that
was
too
would
use
darker
or
just
use
the
stuff
we've
got
already
on
the
bench
marking
machine.
I
don't
know,
but
right.
A
B
A
A
A
C
A
A
A
Ok,
I'm
just
going
to
comment
on
that,
we'll
see
what
other
people
who
are
watching
so
that
one's
still
I
think
a
wait
and
see
anyway,
I've
release
dates
to
church,
so
I
haven't
done
anything
I'm
not
still
think
it's
a
good
idea
of
capture
commit
hash.
Think
it's
still
good
idea.
So
I'm
done
anything
micro
tests,
I,
don't
think,
there's
really
anything
their
truck
start
up.
Speed.
A
B
B
So
the
benchmark
j/s
we
already
use
that
for
at
least
the
require
cash
to
acquire
new
tests.
Okay,
you
had
to
go
install.
We
could
I
suppose
we
could
compare
it
with
bench
d,
but
I,
don't
know
how
much
has
happened
on
that
since
they,
sir
that's
a
bench
d
which
she
says
is
a
slightly
modified
version
of
bench.