►
From YouTube: Benchmarking WG meeting 14 Aug 2017
A
I
was
on
holidays
last
week
and
was
catching
up
from
known
some
of
the
week
before
so
not
too
much.
I
think
the
one
thing
I
did
do
before
I
left
was
I
set
up
a
job
for
Gareth
to
work
on
the
core
benchmarking
job.
Since
we
have
the
new
benchmarking
machines
from
Intel.
One
of
the
things
we're
pushing
forward
is
getting
a
job
that
people
can
use
to
run.
Those
core
benchmarks
on
those
machines
and
I
do
see
that
there
was
some
progress
on
that
which
maybe
we'll
look
at.
B
Yeah
I'm,
very,
very
oh-
this
was
a
action
item
on
my
plate
from
last
meeting
is
to
verify
the
stability
of
the
benchmarking
of
the
benchmark
numbers
on
the
hog-monkey
machine.
So
they
look
stable.
Actually,
I
did
like
three
iterations
runs
and
they
are
like
difference
of
like
two
or
five
score:
okay,
yeah
I
think
yeah.
The
that
machine
doesn't
have
any
problem.
Okay,.
C
A
B
D
B
Like
to
just
hold
back
that
till
we
have
like
fair
data
there,
okay,
so,
but
it
is
like
yeah,
though
the
variance
doesn't
is
not
reflected
in
the
on
the
benchmarking
machine
right.
Okay,
so
that's
good
news
that
basically
there's
no
fundamental
problem
there
right
so
I
didn't
like
what
we
are
doing
today
is
like
just
running
one
iteration.
That
should
be
fine,
I,
guess,
yep,
okay,.
A
Great
yeah
and
just
let
us
know
once
you
get
to
the
point
where,
like
you'd,
want
to
see
some
results
published
regularly
and
we
can
work
on
that
like
they
might
we
have
to
sort
of
figure
out
what
we
wanted
to
do
like
whether
it
would
be
a
duplicate
set
of
numbers
on
a
different
page
or
whatever,
but
I
guess
we
can
wait
to
talk
about
that
till
you
get
there
right.
Yep.
B
Yeah,
okay,
the
other
thing
is
I,
was
supposed
to
set
up
meeting
with
the
thumb
about
like
running
the
new
DCIS
was
not
available
last
week
and
it's
somehow
just
like
lost
traction,
and
then
today
he
he
said
that
he
is
available
this
week,
so
I'm
going
to
meet
him
on
Thursday.
Ok,
what
time
on
Thursday
at.
A
So
I
just
want
to
figure
out
so
what
time
is
that?
So
that's
PST,
so
that's
three
hours
behind
me
right
yep,
so
it
should
be
due
in
the
afternoon
for
you,
so
in
the
afternoon,
I
could
do
2:30.
That
should
be
planned.
Okay.
So
if
you
get
scheduled
for
2:30
and
send
out
an
invite,
that
would
be
great
okay.
A
little
bit,
though,
that's.
A
B
D
Sure
I'm
also
adding
the
tactility
technical
team
I've
been
trying
to
separate
internal
benchmarking.
To
we
were
seeing
all
variants,
then
the
the
node
machine
would
be
one
of
the
reasons
that
Canal
was.
You
guys
do
have
that
variances,
no,
not
visible,
so
I
might
be
interested
learning
more
about
exactly
how
your
machines
are
configured.
So
we
can
replicate
that
and
get
less
variance,
but
I've
been
looking
into
that.
Trying
to
performance
and
I
tend
more
of
these
media
to
be
more
productive.
Okay,.
A
D
A
Sounds
good
yeah,
it
comes
with
a
machine,
it's
I
guess
like
we
just
basically
did
a
base
install
of
Thembu,
214
I
think
it
is,
and
then
we
run
like
the
community
ansible
scripts
I
could
try
and
fight
yeah.
If
you
open
an
issue,
I
could
look
in
our
like.
So
it's
it's
a
machine
and
SoftLayer,
and
so
whatever
the
default
1404,
install
and
soft
layer
is,
is
what
we
end
up
with.
So,
if
you
have
specific
questions
like
things,
sort
of
go
on
the
machine
and
help
out
that
way,.
D
A
Yeah
questions
cuz,
you
know
maybe
we'll
even
find
something
we
shouldn't
have
said
on
ours,
but
it's
entirely
default
and
then
and
then,
if
you
want
to
see
how
its
configured
the
ansible
scripts,
if
you
go
to
I,
don't
know
if
you've
looked
in
the
build
repo.
Oh,
oh,
it's
a
bear
in
the
middle
yeah.
If
you
go
to
build
so
some
in
terms
of
the
base
operating
system
how
its
configured?
A
Repo
is
the
manual
steps
that
I
took
to
install
things,
although
I
think
on
that
bench
marking
machine
itself,
where
they
run
there's
very
little
installed
there.
It
I
think
the
Java
stealth
just
does
things
like
cloning,
the
benchmarking
repo,
so
I,
don't
think
I
had
to
install
anything,
but
if
I
did
it's
written
down
there,
but
in
terms
of
like
packages
and
stuff,
you
can
see
in
that
ansible
script,
exactly
what
it
does
to
the
machine
very.
A
At
Brennan
I
mean
in
theory,
you
should
be
able
to
run
the
ansible
stripped
and
get
it
in.
You
know
and
get
the
exact
same
thing
which
would
be
dead
so
right,
so
yeah.
If
you
have
any
questions
just
let
me
know
what
I
want
wanted
to
point
you
it
back,
because
it
is
fairly
well
documented:
okay,
okay,
okay,
so
anybody
just
looking
up
okay.
So,
let's
move
on
to
actions
from
last
meeting
so
I
think
you've
already
reported
that
completed,
runs
on
on
machine.
C
B
A
A
A
So
the
PR
was
landed,
which
is
good,
I,
think
the
job
I'm
just
gonna
go.
Take
a
look
at
the
CI.
That
sounds
like
there's
a
job
which
does
a
bunch
of
things
benchmark,
node
micro
benchmarks,
I,
see
that
it's
run.
Let
me
login
and
see
if
I
can
even
right,
so
if
I
say
build
with
parameters
right,
I
kind
of
expected,
some
dropdowns
or
something
Oh.
A
A
A
A
A
B
B
E
A
E
E
A
A
A
E
A
B
A
A
A
A
Perfect,
okay,
benchmarking,
jobs,
failing
law
that
that
they
were
failing
earlier
today,
the
benchmarking
data
machine
was
rebooted
and
it
doesn't
I
haven't
gotten
around
to
saying
up
to
having
all
the
pieces
restart
automatically
which
we
should
do,
but
anyway
I
restarted
them.
It
looks
like
it's
gonna
be
okay,
but
we
already
talked
about
the
status
of
writing
the
new
core
benchmarks
in
terms
of
meet.
There
was
an
issue
about
meaning
no
confused
2017.
A
B
B
A
B
A
A
A
A
D
A
A
I
mean
I
think
we
have
like
the
two
things
pushing
forward
on
the
the
benchmarking
like
the
micro
benchmarks
and
getting
the
no
DCIS
yeah
and
the
you
know
your
work
on
track
record.
The
regular
runs.
That
seems
like
three
things
we
can
push
on
right
now
in
the
next
meeting,
so
I
think
till
we
resolve
those
there's,
no
point
necessarily
really
highlighting
anything
else.
Yep.