►
From YouTube: ONNX Edge WG meeting 20190904
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
A
A
Somebody
else
joined
me,
so
they
offer
right.
So
we
just
basically
started
so.
I
was
just
about
to
say
that
I
I
did
I,
didn't
upload
the
slides
that
were
using
the
workshop
workshops
meeting,
and
so
you
can
find
him
here
and
I
thought
that
he
should
just
learn
a
trick
in
a
few
minutes.
Maybe
offer
and
sorrow
aapke
actually
could
give
update
for
the
group
for
the
folks
that
we're
not
there
may
be
skinny
what
happened
at
workshop
regarding
our
or
slide
in
our
concerns
and
discussion,
topics
that
we
had
I
am
guessing.
A
A
Basically,
maybe
challenges
we
presented
our
challenges
to
the
audience
and
then
there
was
a
breakout
session
and-
and
there
was
a
discussion
on
two
proposals
that
we
put
up
and
then
there
was
a
readout
right
and
so
anything
you
know
maybe
five
minutes
just
to
kind
of
go
over
the
whole.
You
know
how
the
the
workgroup
work
group
session
went
in
in
the
in
the
workshop.
D
Okay,
so
maybe
I
can
start
and
they
saw
a
big
thing
to
add
your
perspective.
So
we
did
an
overview
of
edge
emphasizing
that
edge
is
not
just
for
mobile
devices,
but
age
is
covering
the
wider
scope
and
which
lens
for
the
side
covering
what
was
done
so
far.
We
presented
the
challenges,
kind
of
talked
about
the
fact
that
we
have
row
attendants
and
the
other
issue
of
complexity
and
defining
the
profiles
and
making
sure
that
we
wouldn't
have
too
much.
D
So
doing
the
decision
during
the
breakout
session
I
went
over
again
on
the
final
offer
and
there
was
doing
the
workshop.
There
was
some
it
was
called
higher
attendance
I
think
they
were
about
20-25
people
in
the
womb,
something
that
at
least
two
of
them
were
very
active.
There
was
a
guy
that
is
running
the
data
pipeline
working
group,
I
forgot
his
name
but
confided.
He
was
asking
questions
and
there
was
a
puddle
to
participant,
and
so
we
talked
about
some
of
the
questions.
D
D
D
A
D
That's
true
and
I
think
even
in
other
working
groups
like
training
goal,
even
the
open
I,
don't
think
that
the
attendance
is
higher.
It
appears
that
the
actual
work
is
only
by
being
done
by
two
or
three
people.
So
in
our
case
we
have
no
attendance,
which
of
course
affects
the
number
of
people
able
to
do
work
and
contribute,
but
even
in
the
working
group
that
there
is
higher
attendance,
think
they
end
up
with
a
small
number
of
contributions
on
people.
D
D
B
I
think
you
covered
our
session
pretty
well
I.
Think
outside
our
session,
you
know
as
overall
they
made
some
changes
in
the
UNIX
governance.
They
are
making
a
legal
entity
similar
to
Linux
Foundation
and
all
that
which
will
help
more
contributions,
I
think
right
now,
their
licensing
and
all
that
is
a
deterrent
and
hoping
that
that
those
changes
in
their
governance
model
will
help
more
companies
to
contribute
right.
B
B
A
B
Right
now
there
is
not
a
legal
entity
for
Onix
and
what
I
understood
right
so
now
they
are
making
a
legal
entity
which
you
know
when
we
are
contributing
something
on
get
in
the
open
source
community.
We
can
there's
a
legal
entity.
We
are.
We
can
say
that
you
know
the
agreement
has
been
signed
with
them,
so
that
open
foundation
is
will
be
able
to
govern
Onix
I
think
the
IBM
guys
were
are
driving
that
effort
that
established
moving
towards
the
Apache.
A
A
C
A
So
we
have
these
two
things
that
we
propose
in
our
workshop
and
I.
Think
as
I
get
it
like,
there
was
not
I
mean
we
need
to
decide.
Are
we
going
to
extend
our
statement
of
work
to
cover
for
the
edge
profiles
as
execution
mode
right
as
a
working
group?
You
have
to
do
that.
It
seems
like
based
on
at
least
feedback
here
or
discussion
points
from
the
workshop.
It
was
something
that
I
mean.
A
There
was
no
objections,
I
would
say
right,
I
mean
I'm,
not
sure
if
they
were
like
strict,
you
know
big
support
for
it
either
right
I
mean.
Can
we
so
we
have
to
decide
whenever
we
want
to
do
this
or
not
I
mean
you
know,
then
there's
all
this
aspect
of
Oh
API,
and
that
is
completely
you
know.
Another
say
way
of
you
know
another
dimension
that
we
need
to
consider
for
the
whole
signal
work.
One
thing
that
I
will
say
about
this
age
profile
as
execution
mode.
A
Is
that
the
way
it's
defined
rather
with
the
way
I
mean
we
agree?
But
what
you
know
in
document
now
now
I
clearly
understand
what
essaouira
meant
with
a
communal
executive
compliance
discussion.
We
had
before
the
workshop-
and
maybe
this
wasn't
clear
then,
but
now
I
understand.
So
this
is
like
about
having
this
as
a
contract
or
execution
contract
for
the
model
executor
and
to
to
kind
of
obey
bit,
and
so
you
do
either
true
true
by
embedding
a
bending
this
into
the
model
itself,
onyx
model
or.
A
Also,
and
or
actually,
I
guess,
extending
or
providing
some
sort
of
API.
That
device
is
support
to
provide
their
capabilities
right.
What
what
kind
of
profiles
they
can
run
now
that
one
thing
that
I
am
gonna
say
is
that
to
me
that
this
just
goes
even
beyond
the
beyond
the
edge
right
you're
talking
about
you
know,
embedding
a
profile
into
the
model
which
is
generally
applicable
right.
It
doesn't
have
to
do
with
the
edge
anything
it
could
be
if
you're
saying
okay
execute
this
model
with
this
profile.
That
means,
like
you,
know,
I.
A
You
know,
I
want
you
to
run
with
these
with
these
execution
parameters.
Well,
that
that's
that
could
be,
you
know,
that's
applicable
in
the
cloud
and
the
edge
everywhere,
it's
not
just
edge,
and
so
from
that
perspective,
I'm
just
having
a
question
I
like
the
idea,
it
is
kind
of
interesting
whether
we
want
to
make
it
part
of
the
working
edge
working
group.
That's
another
discussion.
You
know
it
seems
like
it's.
It
is
of
more
general.
You
know.
A
It
is
a
more
general
purpose
than
just
edge.
That's
that
is
one
concern
that
I'm
thinking
of
so
I'm
just
curious
to
I
would
like
to
you
know
think
you
know
here,
maybe
others
views
of
this.
If
you
agreed
to
extend
and
add
this
to
the
working
group
yeah
we'd
done,
we
should
update
the
state
of
our
work
and
make
sure
that
is
covered
right
fix
it.
So
you
know
I'm
just
wondering
what
you
guys
think
should
this
be
part
of
the
edge
working
group,
a
statement
of
work.
B
B
C
B
We
haven't
seen
much
need
like
like
they
don't
because
they
have
another
orchestration
mechanism
through
VMs
and
through
that
mechanism
they
can
do
resource
allocation,
and
you
know
your
again:
they
don't
care
about
it.
Much
because
of
that
very
reason
again,
they
can
control
the
execution
environment
using
the
VM
knobs
that
are
available
to
them.
However
an
edge.
You
know,
some
of
these
things
are
not
needed
or
not
available
or,
and
that's
why
the
motivation
is
coming
from
the
edge
side.
B
A
Other
than
recommending
that
this
is
an
interesting
option
to
consider
for
you
for
for
onyx
body,
because
now,
if
you
can
start
getting
to
the
API
definition
all
this
stuff,
this
is
all
recommended
to
me.
Those
are
recommendations
we
made
to
to
only
intro
sake
group.
Six,
okay,
listen!
Consider
these
two
things
you
know
and
do
we
need
to
I,
wouldn't
necessarily
think
that
we
need
to
get
into
it
that
this
working
group
needs
to
get
into
the
defining
the
onyx
or
proposing
changes
to
the
onyx
format
or
proposing
changes
to
introduce
new
API.
A
A
It
is
just
general
concept
that
is
applicable
right
because
even
in
the
document
here
that
I'm
reading
and
I
read
this
morning
through
that
you
put
together
syrup,
yeah
I
mean,
and
it
talks
about
a
yeah
I
mean
execute
release
to
allocate
the
proper
memory
or
schedule
the
execution
workload
allocate
proper
resources.
Well,
that's
very
much
also
cloud
applicable
right.
A
You
can
say
yeah
Randy's
on
the
cloud
but
execute
with
this
profile,
because
I
don't
need
it
any
faster
right
and
that
is
gonna
help
Club,
because
they
don't
use
many
resources
and
you
know-
and
they
could
you
know
properly,
balance
the
resource,
loading
and
all
that
stuff,
and
so
it's
very
much
good
idea.
I
agree
that
this
could
be
proposed.
I'm.
Just
not
one
person
convinced
that
this
working
group
needs
to
kind
of
deal
with
it.
D
That
for
now,
the
first
challenge
that
we
have
is
to
say
that
we
managed
to
design
all
fairness
in
a
in
a
way
that
we
have
satisfied
with
in
a
way
that
is
consistent
and
represents
what
we
wanted
to
represent,
and
we
are
able
to
capture
the
collective
set
of
attributes
and
restrictions,
as
we
set
out
to
do
and
using
the
profile
as
an
execution
mode
is
kind
of,
depending
on
being
able
to
correctly
describe
for
fun
in
a
way
that
is
sufficient.
So
to
me
it
looks
like
the
first
task.
A
B
A
B
One
thing
to
add
here
is
and
I
think
I
touched
there.
It
was
also
discussed
in
the
comments
section
there
too.
You
know
like
the
way
we
are
defining
right
now
as
static
profiles
like
fixed
profiles.
I
think
the
applicability
of
such
concept
would
remain
restricted
to
fixed-function
devices
then
like.
B
If,
if
we
want
to
extend
the
concept
of
edge
to
devices
which
can
do
more
than
a
fixed
thing,
then
we
need
to
also
get
into
this
space
of
execution
modes.
Otherwise
we
will
only
be
able
to
tackle,
like
you
know,
fixed
devices
like
you
know,
vision
or
audio
or
speech,
or
something
of
the
third
shot.
Well.
A
A
You
know
it's
quite
possible
that
the
single
device,
but
I
guess
you
call
it
static,
but
I
mean
single,
defy
device
could
support
multiple
of
the
profiles
right.
We
agree
about
that.
Right,
I
mean
it
could
be
that
you
know,
even
even
smart
speaker
is
a
composite
device.
Small
speaker
can
support
to
the
2d.
A
You
know
it
basic
profile
in
audio
basic
profile.
For
example,
right
I
mean
that
even
the
smart
speaker
is
a
completely
composite
device.
Even
you
know
PC.
Obviously
it's
very
much
composite
device
right
and,
and
so
from
that
perspective,
I
don't
see,
I
guess
when
you
say
we
only
can
can,
can
focus
on
a
static
profiles.
I'm
not
sure
that
I
understand.
What
do
you
mean
to
me?
Any
device
is
pretty
much
could
be
composite
device
combined
from
one
or
more
profiles
right.
Any
device.
B
So
in
case
of
a
PC
kind
of
environment,
you
have
hundreds
of
processes
that
are
running
simultaneously
mm-hmm
and
the
scenario
requires
certain
quality
of
service
and
right
now,
there
is
not
a
way
to
express
that
there
is
not
a
way
to
express
that
this
model
requires
this
much
memory,
this
much
power.
This
much
all
these
different
attributes
that
we
have
already
defined
like
this
model
requires
it
to
go
to
the
network.
Also
right
like
for
the
fifth
attribute
that
we
talked
about.
So
some
of
these
things
are
not
just
static
elements.
B
They
are
sort
of
dynamic
in
the
sense
that
a
PC
can
probably
support.
You
know
n
number
of
profiles
and
not
just
affect
you
know
like
a
bigger
number,
then
just
5,
4
or
5
things
that
we
end
up
defining
or
a
hard-coding
some
fixed
attributes
that
these
are
the
things
that
would
define
as
an
audio
profile
or
a
2d
profile
in
case
of
a
PC
environment.
Like
you
know,
depending
on
the
model
model,
it's
it's
very
closely
associated
with
the
model
file.
D
A
Yeah
I
mean
it
seems
like
to
me
that
if
you
are
to
use
this
profiles,
attributes
to
actually
change
them
dynamically
afterwards
for
the
photo
service
purposes
say
say:
we
have
the
profile
that
is
basic
2d
and
it
defines
a
you
know:
maximum
input
image
size
and
how
much
smoke
inference
and
all
that
stuff,
and
that
is
like
to
say
that
proof
that
that
profile,
but
vendor
says
iron
compliant
with
this
profile.
Meaning
I
can
support
this
right.
A
But
what
you're
now
saying
is
that
at
the
runtime
you
could
even
say:
oh,
this
guy
can
support
this
profile.
But
let
me
dynamically
change
some
of
these
activities
and
say
instead
of
running
20
frames
per
second
run
2015,
because
that
is
within
the
boundaries
of
the
profile.
But
we
are
now
talking,
maybe
about
energy,
changing
this.
These
attributes
at
the
execution
time
to
control
execution
right
so.
A
A
To
me,
exit
I
mean
that's,
that's
my
face,
but
you
know.
Maybe
we
don't
go
right
away
into
this.
Changing
the
really
scope
of
work
for
the
edge
working
group,
but
rather
continue
on
on
the
part
that
we
envisioned,
but
then
clearly
we
can.
We
can,
as
we
define
these
profiles,
we
can
always
you
know,
maybe
cross
check
this
thinking
of
execution
control
and
see
how
this
kind
of
a
lines
any
profile.
A
Indeed,
is
this
the
way
we
are
defining
the
profile
with
the
attributes
we
identified
is
something
that
is
sufficient
to
control
execution
as
well
in,
in
the
general
sense,
the
quality
of
service.
When
you
want
to
execute
the
model,
controls
may
be
mainly
require
more
more
knobs
right,
not
just
the
attributes
that
we
have.
You
know
currently
and
I,
don't
know,
and
that's
why
I
think
even
that
execution
control
and
a
cocoa
service
control
is
perhaps
the
it's
a
separate.
It's
a
separate
topic.
C
A
And
okay,
so
the
other
thing
was
that
so
I
guess
there
was
there
was
this.
You
know
document
that
you
put
together
as
a
proposal
in
context
of
execution
compliance
that
offered
this
some
comments
on
and
did
some
suggestions
to
clarify
to
disambiguate
this
compliance
versus
execution,
because
that
is
what
is
creating
a
bit
of
maybe
confusion.
A
But
if
you
all
agree,
I
mean
I
guess
this
did.
We
should
not,
then
necessarily
continue
about
discussing
you
know
using
or
updating
the
extending
edge
profile
definition
to
to
be
used
for
the
I.
Guess
in
this
occasion
would
control
for
now
right
so
I,
don't
know,
I
mean
it's
up
to
syrup.
You
could
you
could
we
could
table
this
document
for
now
and
then
come
back
to
it?
You,
you
are
free
to
update
the
document,
their
comments.
A
Actually,
if
you
want,
but
at
least
we
captured
drinking
here
and-
and
so
that
is
good
enough
for
content
right.
What
in
originally,
when
I,
when
we
started
talking
about
this
and
that's,
why
we're
compliance
but
confusing,
is
that
I
thought
we
are
talking
about
Onix
compliance
workflow,
which
is
inch
profile,
compliance
workflow,
which
is
a
part
of
our
scope,
which
is
about
to
say
these
are
the
profile
parameters.
A
This
is
where
you
know
that
that
come
that
compliance
we
talked
about
is
about
right,
and
so
we
need
to
definitely
make
that
part
of
the
profile,
and
actually
it
doesn't
be
great
great
if
you
could
actually
elaborate-
and
you
know
and
create
a
separate
document
for
that
type
of
compliance
right.
It's
it's
about
edge
profile,
certification.
A
It's
about
vendors
certifying
their
devices
to
be
compliant
with
a
defined
edge
profile,
and
so
what
is
that?
Workflow,
right
and-
and
it's
actually
kind
of
similar
in
the
context
to
this
document,
because
there
are
certain
parameters,
but
you
know,
obviously,
once
we
get
into
definition
and
scope
of
this.
It's
clear.
It
becomes
clear
that
we
are
talking
about
the
execution
mode
and
so
that
that
is
different.
But
you
know
the
the
parameters,
attributes
and
all
that
stuff.
You
know
they
have
similar,
meaning
or
same
meaning
in
both
cases.
A
B
That's
something
that,
but
that
you
could
pick
up
on
yeah
I
need
to
better
understand
it.
You
know
things
a
little
bit
on
where
we
going
with
the
static
profiles
again:
I'm
sorry
I'm,
using
the
term
static
profile
here,
because
to
me
it's
still
very
much
centered
around
IOT
devices,
not
the
multifunctional
devices
which
have
a
lot
more
to
offer
right.
A
A
Speed,
for
example,
for
the
you
know
has
to
be.
Let
me
see
that
document
here,
for
example,
to
the
basic
profile.
Yes,
if
we
have
static
set
of
attributes
and
say:
okay,
for
example,
speed
is
30
frames
per
second
static,
but
it
could
be
up
to
20
frames
per
second.
It
could
be
at
least
30
frames
per
second
right
of
2
being
could
be
more
right
or
depends
how
we
define
so
it
has.
It
will
have
a
range
of
values,
potentially
that
we
kept
a
low
profile.
A
B
A
A
A
I
want
to
run
this
this
model,
but
you
know
bounded
to
20
frames
per
second,
for
example,
I
mean
that
is
completely
you
know,
within
the
bounds
of
the
supported
profiles,
but
somebody
there
has
to
be
mechanism
that
somebody
established
in
onyx
world
to
enable
that
control
right
of
dynamic
fitting
into
the
like
desired
execution
parameters.
I,
understand,
I,
understand,
I,.
D
That
was
not
necessarily,
let's
say
that
I
define
a
profile
that
says
that
now
you
need
to
pause
this
twelve
different
images
at
4k
at
100
frames
per
second
okay,
let's
say:
I
have
this
profile?
Okay,
not
every
PC
can
now
hold.
This
is
an
automotive
Bopha.
Okay,
not
every
PC
can
process
input
for
12
cameras
at
100
frames
per
second
4k
resolution,
so
even
the
PC
has
limitations.
Yeah.
C
D
Well,
so
it's
only
a
question
of
how
high
we
define
some
often
to
be.
Maybe
we
can
define
it
in
some
high
heaven
that
even
some
of
the
pcs
would
not
be
able
to
meet
these
profiles.
So
I
kind
of
I
somewhat
disagree
that
PC
can
support
any
of
the
four
define
four
five.
It's
a
question
of
how
high
we
define
the
profiles,
and
this
is
well
well.
B
B
D
A
A
B
B
For
a
platform
which
is
scalable
as
what
I
am
talking
about,
a
scalable
platform
will
have
less
benefit
of
these
profiles.
A
scalable
platform
will
not
be
able
to.
You
know
like
if
you
create
a
2d
profile,
basic
intermediate
in
advanced
for
a
scalable
platform.
It
doesn't
mean
a
whole
lot.
Yes,
it
there
is
some
value,
but
it
it's
not
a
huge,
exciting
thing.
A
scalable
platform
can
benefit
a
little
bit
more.
B
If
we
would
say
that
you
know
these,
these
are
things
the
same
5
things
that
we
have
talked
about,
what
they
are
actually
embedded
in
the
motor
file,
then
a
scalable
platform
can
read
it
and
execute
it
accordingly,
it
can
decide.
Oh
I
need
to
use
that
accelerator,
that
is
in
my
system
or
decide
not
to
use
it
like
the
example
that
you
gave
about
automotive,
it
could
very
well
be
that
in
that
example,
it
can
choose
to
run
it
accordingly
or
not.
B
The
scope
of,
or
the
definition
of
edge
that
we
defined
earlier
if
it
is
going
to
cover
scale
as
device.
Also,
then,
that
element
is
also
important,
otherwise
we'll
be
restricted
to
just
IOT
edge
devices
and
at
least
that's
how
I
see
things
but
I
guess
we
need
to
discuss
a
little
bit
more
further
down.
A
So
the
device
has
certain
characteristics,
compute
characteristics
that
somebody
defined
to
begin
with,
given
PC
right,
what
do
you
think
about
scalability
in
terms
of
somebody
upgrading
or
changing
compute
characteristics
of
the
device
or
scaleable
in
terms
of
capabilities
of
that
device?
In
terms
of
the
execution?
Let.
B
Me
give
you
an
example
right,
let's
say
we
know
Lenovo
XYZ,
that
platform
came
up
right
and
we
certified
that.
Oh,
it
can
only
comply
it
to
the
audio
basic
device
profile.
Only
right
we
certified
it.
It
can
only
comply
to
the
audio
basic
profile
now
most
of
the
devices
that
are
in
the
market
right
now
they
have
a
Thunderbolt
port.
It's
a
PCI
extension.
Basically
right.
B
They
have
USB
ports
so
device
which
is
scalable.
It
can
now
suddenly
be
able
to
support
profiles
which
are
more
advanced.
Now
when
we
advertise
the
platform
to
begin
with,
we
said
it's
only
compliant
to
a
basic
profile.
Now
I'm,
not
saying
that
you
know
the
profile
cannot
be
updated,
it
can
be
updated.
However,
the
value
of
such
static
profiles
is
relatively
less
for
a
scale,
a
scalable
device.
B
B
A
Guessing
the
way
I
see
it,
I
get
what
you're
saying,
but
the
thing
is
like
somebody
will
say:
well,
you
define
when
somebody
will
say
I'm
compliant
or
certified,
for
these
profiles.
They
have
to
describe
device
characteristics,
they
have
to
say
on
this
PC
running
in
the
nominal
mode,
not
in
turbo
mode
right
running
in
you
know
the
disk
lock
frequency,
whatever
they
will
need
to
describe
device
characteristics
that
includes
PC
right
and
they
will
have
to
say
in
this-
is
within
these
parameters
of
the
device.
I
am
complying
with
these
profiles.
B
The
characteristics
are
always
dynamic
in
a
PC
environment,
at
least
the
ones
that
I'm
working
at
Intel
at
this
all
these
things,
these
characteristics
that
you're
saying
they
are
constantly
being
managed
by
the
system,
frequency
memory.
A
lot
of
these
things
are
coming
and
going
and
unlike
a
fixed
function
device
where
you
have
these
things
are
fixed
in
a
PC
environment.
These
things
are
always
changing.
You,
you
load
an
app.
A
I'm
trying
to
say
is
when
you
declare
your
compliance
when
somebody
declares
compliance
to
certain
profile,
they
have
to
explain
and
describe
their
test
and
where
that
means
fully
powered.
So
it's
just
going
to
some
of
your
examples.
They'll
have
to
cover
okay,
you
know
fully
powered.
You
know
fully
charged
battery.
No
just
like
fresh
boot
from
the
from
the
start,
I
mean
you'll
have
to
describe
your
test
environment
to
say
when
I
run
in
this
test,
environment
I
can
certify
that
I
am
compliant.
This
is
very
much
the
same
as
with
benchmarks.
A
You
cannot
mean
benchmarks
the
dictate
that
you
specify
your
configuration
right,
I
mean
otherwise.
Benchmarks
are
moved
right.
If
you
just
say
you
know,
I
can
run
this
speed,
but
without
saying
how
you
can
run
that
you
know,
then
it's
gonna,
be
you
know,
kind
of
data
point
without
you
know
much
much
value
bottom
line
you
everybody
will
have
to
describe
how,
under
what
circumstances,
certain
hardware
configuration
is
compliant
with
a
certain
profile.
A
Now,
how
much
of
that
value
is
to
the
end-user,
for
example,
in
the
PC
space,
and
because
it's
like
very
dynamic
and
not
changing
good
question?
I,
don't
know
to
me:
it's
a
it's
a
it's
a
part.
It's
a
sort
of
ballpark
estimate.
If
you
run
it
in
this
configuration
I
can
guarantee
you
can
achieve
these
profiles
right
and
that's
it.
If
you
change
this
these
execution
environment,
your
mileage
will
vary
right
because
it
didn't,
it
hasn't
mean
within
parameters
of
the
test
environment,
the
tile
used
to
certify-
and
that
is
the
way
I
see.
B
B
A
To
me
that
seems
reasonable
and
I
get
the
challenges
that
PC
may
face,
but
I
also
feel
that,
within
certain
predefined
set
of
you
know,
platform
hundred
configuration
parameters
that
somebody's
using
it's
still
valid.
You
can
say
within
the
index
within
these
hundred
spec
parameters
and
execution,
environment,
I
can
say,
I'm
compliant
and
that's
all
it
means
right.
I
mean
it
means
if
you
change
these
parameters
in
any
significant
way.
Well,
your
your
suddenly
may
not
be
compliant
with
this
profile
or
you
may
be
compliant
with
with
the
higher
profile.
A
It
is
the
way
you
know,
I
think
it
could
be
managed
yeah
that
that
could
lead
to
many
variations
for
the
scalable
highly
scalable
device
like
PC
in
terms
of
ok,
how
many
hundred
configurations
and
execution
modes
I
need
to
test
and
certify
I?
Guess
that's
up
to
the
vendor,
then
to
decide
how
many
of
these
proffered
wear
setups
I
would
like
to
certify
for
so
that's
extra
kind
of
burden.
A
But
it's
not
you
know
in
an
unmentionable.
You
know
to
be
done,
and
so
I
am
just
you
know
wondering
if
that
is
gonna
be,
like
you
know,
really
a
blocker
showstopper
for
further.
You
know
you
know
discussion.
Maybe
if,
if
we
go
down
the
path
of
defining
these
profiles,
we
we
start,
you
know,
thinking
through
it
again
and,
and
you
know,
try
to
kind
of
maybe
some
as
we
kind
of
review.
Some
of
these
things
will
potentially
present
yourself
how
it
could
be
managed,
and
so
we
we
can.
A
A
Okay,
so
listen,
we
are
out
of
time.
I
may
actually
need
to
run
another
meeting.
I
guess
you
know
some
action
items
for
next
next
meeting,
perhaps
I
could
take
the
stab
of
you
know.
Defining
one
of
these
profiles
like
2d,
2d
basic,
inter
maybe
just
maybe
even
all
three
of
them
I
actually
had
proposal
in
this
direction.
A
Already
we
discussed
so
maybe
I
could
more
extend
that
for
next
time,
and
so
we
kind
of
we
can
consider
engineering
that
next
time
and
then
you
know
then
then
use
that,
maybe
as
a
potential
template
for
others
right
and
then
then
some
other
folks
can
take
on
others
and
and
then
we
can
kind
of
can
establish
you
know,
set
of
we
helped
to
establish
set
of
profiles.
We
are
originally
gonna,
be
saying
we
are
gonna,
define,
and
now
we
have
these
six
I
think
I
don't
know.
A
If,
if
we,
we
need
to
go
further
than
that
right
now,
I
would
like
that
we
define
something
that
is
kind
of
tangible
and
then
the
addition
of
more
is
something
that
could
come
afterwards
and
even
after
working
group
finishes
in
a
sense
like
okay,
here's,
the
here's,
a
set
of
things
that
we
define
and
then
there
is
a
part
to
extend
this
right.
If
need
be.
Does
that
make
sense
for
next
time
that
we
kind
of
go
through
the
one
of
these
profiles
more
in
depth
and
discuss
and
review
makes
it
make
sense?