►
From YouTube: 20220118 FPGA standup
Description
FPGA Stand-up for 18 January 2022.
B
A
Okay,
oh
yeah,
no,
not
being
an
expert
in
in
the
nbd
or
nfs.
Then
then,
what
I'll
do
is
go,
try
it
and
see
if
I
can
get
paul
to
go
ahead
and
implement
it
as
a
as
a
way
for
us
to
deal
with
the
file
system
use
for
the
for
the
reference
design.
I
think
it's
a
big
win
so
I'll
do
that
today,
because
I
don't
have
any
meetings
or
anything
to
do
at
work
until
6
pm
my
time.
A
So
this
is
good
news,
because
I
don't
get
a
lot
of
uninterrupted
time
because,
as
as
you
all
know,
interruptions
are
very
costly.
So
I'm
looking
forward
to
digging
into
that
and
also
helping
test
out
the
the
wonderful
stuff
that
andre
has
taught
me
about
how
to
how
to
get
around
the
limitations
with
the
port
types.
A
So,
but
back
when
I
was
doing
fpga
and
asic
work,
then
this
was.
It
was
all
very
conservative
and
the
the
the
level
of
writing
for
the
vhdl
that
we
have
here
is
a
little
bit
different
than
so
I'm
relearning
everything
that
I
used
to
know
and
learning
lots
of
new
things
and
it's
fantastic,
I'm
having
a
blast.
A
No
so
there's
a
script
that
I
think
takes
care
of
it
and
I've
I've
gone
through
the
xilinx
forums
and
learned
that
like
this
is
a
thing,
and
I
think
that
that
particular
script
that
sets
this
up,
which
I
missed,
because
it
was
in
the
build
directory
and
not
in
the
rtl
directory.
I
think
that
all
I
have
to
do
is
call
that
in
the
tecla
script
that
is
used
by
the
reference
design
build.
So
that's
what
I'm
going
to
try
to
do
today
and
then
that
should
work.
A
I
looked
at
reddit
and
I'm
like
well.
This
just
sets
everything
up.
This
is
really
very
cool
so
that
that's
what
I'm
going
to
try
to
do.
I
tried
to
get
around
to
it
yesterday
and
it
and
never
got
back
to
the
lab.
So
anyway
back
so
it's
stefano
back
to
you
for
anything
else.
You
want
to
say,
and
then
andre
has
the
floor
to
correct
me
and
to
continue
to
to
help
help,
get
things
working.
C
Yeah
so
I
yeah
the
script
was
so
it
worked
in
my
small
tests.
But
then,
if
I
run
the
script
in
in
an
existing
project,
it
will
sort
of
try
to
change
stuff
from
other
files,
and
you
know
all
messes
up
and
that's
not
great.
So
I
have
I
have.
I
have
a
fix.
I
just
didn't
submit.
Yet
it
it's
ready
and
I
can
submit
in,
like
I
don't
know,
15
minutes
or
so.
A
Oh
cool,
okay,
yeah
it'll,
take
me
longer
to
get
back
over
to
the
lab,
so
I
still
got
to
wrap
stuff
up
here
for
the
day,
so
yeah
go
ahead
and
check
it
in
and
then
I'll
go
to
the
build
directory
and
try
to
see
if
I
can
use
what
you've
got
to
to
set
it
up
to
where
the
the
overall
build,
with
the
big
reference
design
that
it
all
builds
and
everything.
So
I
think
we're
really
close
yeah.
So
I.
C
C
A
A
Yeah,
I'm
very
I'm
just
very
excited
because,
like
you
know
the
the
I'm
I'm
now
really
appreciating
the
like
the
9371
and
the
jusde
204b
is
great.
It
solves
a
lot
of
problems
but,
like
you
have
to
know
how
to
bolt
it
up,
and
I
think
that
yeah,
when
we
finally
get
all
the
integration
pain
sorted
out,
that
we'll
get
some
some
fantastic
results
so
much
better
than
than
the
stuff
I
had
to
deal
with
and
and
the
most
recent
like
for
the
9361.
C
A
Sort
of
a
completely
different
interface,
so
this
is
this-
is
pretty
cool,
so
I
think
it'll
work,
sorry
it
take.
It
takes
me
so
long
to
to
figure
this
stuff
out.
I'm
I
am
honestly
trying
to
learn
it
as
fast
as
I
can
and
I'll
do
the
best
I
can
so
I
and
now
I
so.
I
see
the
the
work
that
you've
done
and
the
wrappers
and
everything
and
we'll
we'll
get
it
working.
C
A
A
Yeah,
no,
that's
a
really
that's
a
good
point,
because
it
might
be
that
we
need
one
of
these
per
ip.
You
know
so
for
for
the
other
things
that
we
want
to
integrate
in
or
if
there's
a
way
to
kind
of
put
it
at
the
top
level
for
the
in
the
top
level
make
for
the
project,
then
that
might
be
the
right
spot.
I
don't
know
yet
you
know,
but
this
is.
C
A
A
That's
all
I've
got
going
on
here
anyway,
we're
we're
right.
We
are
working
hard
to
to
get
in
touch
with
lots
of
different
groups
at
ieee.
That
will
be
very
interested
in
this
work,
and
I've
spent
a
lot
of
time
on
the
phone
for
like
organizational
and
administrative
stuff
for
taxes,
but
today
I
get
to
to
spend
in
the
lab,
and
I
should
have
a
report
at
the
end
of
my
day.
C
Yes,
yes,
like
he
was
trying
to
get
pluto
up
and
running.
Yes,
I
did
build
pluto
and
you
know
I
had
a
quick
look
and
I
think
I've.
There
is
a
data
path
that
essentially
zacks
zach's
stream
and
it
comes
from
a
dma.
So
I
I
pointed
I
didn't
fully
complete
like
locally
the
thing,
and
I
pointed
him
how
he
could
do
I
might
yeah.
I
might
actually
get
something
like
more.
C
A
C
C
I
have
sort
of
a
script
or
you
know
something
that
resembles
the
script
and
I
need
to
actually
just
like
start
recording
and
you
know
timing
and
because
it
might
be
that
I
think
it's
gonna.
What
I
have
is
gonna
take
10
min,
I
don't
know
10
minutes,
but
it
actually
takes
45
minutes
and
therefore
I
have
to
cut
a
lot.
So
I
yeah
I
need
to
get
this.
C
C
A
C
A
B
A
C
Yeah,
I
can
probably
like
the
presentation
I'm
doing
in
a
I.
I
don't
have
the
link.
It's
like
a
it's,
not
powerpoint.
It's
a.
I
don't
know
how
to
call
it's
a
different
thing
and
you
can
put
like
some.
I
think
it's
more,
it's
less
boring,
but
the
key
thing
the
is
it
like,
I'm
not
in
a
paid
account.
So
it's
a
no
it's
it's
open
like.
I
can
share
the
link.
C
A
Yeah
they
they.
They
say
that,
because
they
eric
guth
and
the
rest
of
the
ham
expo
team.
They
want
original
work.
They
want
things,
that's
original,
so
they
don't
want
you
to
give
the
same.
Talk
that
you've
given
at
hamcation
hamvention,
tapper
dcc,
you
know
but
like
in
the
past,
when
when
a
talk
has
has
has
applied.
A
A
Thing
like
that,
we're
trying
to
get
new
material,
but
it
doesn't
mean
that
it
it
has
to
be.
You
know,
relentlessly
unique
because
that's
a
lot
to
ask
so
so
don't
worry
about
the
the
material,
some
of
the
material
being
or
what
they're
also
very
enthusiastic
about.
Is
that
if
you've
given
a
talk
in
the
past-
and
you
have
one
that
builds
upon
that
work
and
shows
continuation
and
progress-
they're
very
excited
about
that
sort
of
stuff
too.
So
don't
worry
too
much
about
that
you're.
C
C
C
A
B
A
B
B
Now
it's
it's
like
like
to
jump
onto
a
horse
that
is
already
running,
so
I
so.
For
example,
I
heard
that
there
are
some
register
problem
in
the
past.
I
I
use
it
to
create
some
kind
of
automatic
tool
to
just
describe
the
register
in
in
a
interface
definition,
language
and
then
just
start
the
script
and
generate
the
vhdl
from
one
side
and
the
software
called
the
c
code
from
the
other
side
because
then
needs
to
be
in
sync.
B
Maybe
it's
in
in
some
document,
but
it
didn't
dig
too
much
so
so.
At
the
end,
the
data,
the
payload
will
come
from
the
the
board
itself
right.
So
I
don't
know
from
the
audio
channel
or
whatever
some
kind
of
packets.
Then
we
throw
in,
and
so
this
will
so.
This
will
affect.
Obviously
the
ps
part
of
the
of
the
of
the
of
the
zinc,
and
so
so
is
the
dma.
B
A
Yeah
yeah,
I
got
it
yeah,
they
say
yeah,
the
the
automatic
generation
so
yeah.
Your
three
questions
are
like
what
do
we
do?
Where's
our
status
with
like
the
the
automatic
vhdl
generation
like,
for
instance,
hls
or
the
or
the
matlab,
hdl
coder,
or
things
like
that?
It
sounds
like
that's
the
that's
the
area
and
that's
that's
an
area.
I
know
a
little
bit
about
and
I'm
trying
to
get
the
correct
tool
toolbox
from
matlab,
and
I
got
some
some.
A
I
have
some
reports
on
that,
but
it's
still
in
the
process
of
trying
to
negotiate
what
we
use
or
or
what
I've
used
for
for
matlab
is
the
the
at-home
license,
which
is
non-commercial
and
also
non-academic,
because
we're
not
a
university
we're
just
a
non-profit
and
but
we're
not
commercial
either.
So
we
get
a
huge
discount
on
matlab
and
we
get
a
huge
discount
on
matlab
and
simulink
and
a
lot
of
the
toolboxes,
however,
not
all
of
them,
and
unfortunately,
one
of
the
ones
that
we're
not
allowed
to
buy
is
hdl
coder.
A
It
turns
out.
So
I
wrote
to
matlab,
and
I've
had
several
phone
calls
with
them
and
we
are
not
yet
at
a
point
where
they're
able
to
or
they're
willing
to
sell
us
a
very
discounted,
hdl
coder
toolbox,
because
they're
worried
that
it's
too
hard
for
us
to
understand
vhdl
and
they
they
they've
from
that.
From
their
point
of
view,
they
already
have
to
give
us
such
a
huge
amount
of
customer
and
technical
support
to
their
commercial
companies,
their
commercial
customers
that
use
hdl
and
gpu.
A
They
have
a
gpu
coder,
an
hdl
coder
and
a
matlab
coder,
which
does
general
purpose
processing.
So
they
have
all
the
heterogeneous
processing
stuff
so
to
go
from
matlab
directly
to
all
the
way
to
either
gpu
or
hdl
or
arm
code.
So
we're
still
talking
so
I've.
I've
tried
to
stick
up
for
us
and
say
you
know
we
did
not.
We
don't
use
any
of
your
technical
support
at
all.
That's
not
available
to
a
home
customer,
non-academic
and
non-commercial
customers.
Don't
get
any
technical
support
anyway.
What?
A
That's
one
of
that's
one
of
the
primary
tool
flows
that
they
want
us
to
use.
They
want
us
to
use
a
workflow
that
incorporates
matlab
in
order
to
do
that,
you
have
to
have
these
toolboxes
and
that's
what
I've
been
working
on
with
matlab
now,
if
we
wanted
to
buy
them.
If
we
wanted
to
say
okay,
we'll
pay
the
commercial
rate,
then
it's
like
13
000
a
year.
A
That's
a
lot
of
money!
So
it's
just
like.
Okay,
that's
an
awful
lot
of
money.
The
matlab
is
expensive.
The
the
license
that
we
have
is
like
one
percent
of
that.
It's
like
a
150
a
year
as
opposed
to
five
six
seven
thousand
a
year.
So
that's
where
we're
at
on
being
able
to
generate
fpga
code
from
like
either
octave
scripts,
like
you
can
work
with
a
free
and
open
source
tool,
chain.
A
A
I
I'm
hoping
that
that
is
the
same
sort
of
that.
That's
answering
some
of
the
question
here
and
that
you
know
we
also
have
an
option
to
use
hls
inside
of
auto
and
we
do
have
a
full
license
from
vivato
that
allows
access
to
dsp
high
level
system
or
whatever
it's
called
so
hls
allows
you
to
go
from
c
code
into
fpga
code.
So
we
already
have
that
in
the
license.
A
So
I'm
not
too
concerned
about
where
the
payload
comes
from,
but
I'm
assuming
it
comes
from
the
processor
side.
Originally,
we
were
going
to
set
up
programmable
logic
so
on
the
pl
side
to
do
gse
right
out
of
the
box,
but
people
like
ron
economos
recommended
that
maybe
we
should
do
it
a
little
more
simply
and
and
he's
right.
A
So
I
think
I
know
we
had
a
discussion
on
slack,
but
it's
probably
way
up
in
the
history
and
I've
got
it
down
here
as
an
action
item
to
revisit
that
and
make
sure
that
it's
understandable
and
then
dma
is
a
really
good
question.
That
was
your
third
thing:
the
direct
memory
access.
That's
the
only
way
that
you
get
data
in
and
out
of
the
9371.
A
So
it's
it's
designed
for
direct
memory,
access
through
jazz
d204b
and
that's
it
that's
how
you
get
stuff
in
and
out
and
as
soon
and
we
have
been
able
to
ship
data
using
both
the
c
there's
a
there's,
an
api
from
analog
devices
and
we've
been
able
to
get
it
to
work
using
both
the
c
code
methods,
an
interface
and
also
python,
and
so
far
python
seems
to
be
easier
to
use.
But
we're
we're
a
little
concerned
about
the
performance
like.
A
Is
it
going
to
be
slower
now
figuring
out
how
to
how
to
get
all
this
to
work
together
is
is
what
we're
working
hard
on
now.
So
in
terms
of
like
dma,
it's
like
it
seems
to
be
sitting
right
there.
You
look
at
the
diagram
and
you
look
at
the
hdl
and
the
reference
diagram
and
everything's
directly
hooked
up
to
to
dma
and
when
you
write
a
little
python
script
that
makes
a
sine
wave
or
you
know
it's
eyes
and
cues.
A
A
A
Yeah,
so
it's
like
somebody
handing
you
the
keys
to
a
car
or
like
handing
giving
you
a
car
without
any
keys,
and
it's
like
so
we're
we're
we're
sorting
it
out.
We've
been
able
to
operate
the
the
hardware
successfully
with
the
api,
and
now
it's
like.
Okay
now
it
seems
like
you,
just
set
it
up
in
the
processor
side
to
write
to
this.
This
memory
that
you've
allocated-
and
it
just
shows
up
on
jesde
and
out
the
door
it
goes
and
wow.
Is
it
fast?
It's
really
kind
of
neat.
A
A
Thank
you
yeah.
I
just
this
is
the
sort
of
thing
where
I'm
like,
I
feel
so
stupid
I've.
You
know
that
sometimes
the
hardware
makes
me
feel
dumb
like
where
where's
the
where's,
the
where's,
the
steering
wheel,
and
it's
like.
Oh
okay,
it's
it's!
It's
trying
the
tr.
This
really
is
a
a
very
nice
system
on
ships.
B
B
Absolutely
thank
you
thanks
a
lot
for,
for
the
first
step.
I
think
your
job
is
very
is
something
extraordinary,
because
you
are
in
the
middle
between
university
and
an
enterprise,
and
you
are
kind
of
taking
really
the
best
from
both
and
it's
really
keep
everything
open
and
I
totally
share
your
your
vision
is
so
you
are
opening
a
new
road
at
the
end.
B
For
the
first
point,
for
example,
just
to
say,
I
see
another
possibility,
but
I
never
verified
this
so
just
to
from
octave
to
generate
c
code,
because
at
the
end
it's
a
cheapest
plus
and
let's
see
if
this
code
can
be
somewhat
entered
into
the
exciting
hls
engine,
and
that
would
be
totally
cut
off
all
the
commercial
stuff.
But
I
I
never
spoiled
this.
B
I
said,
but
theoretically
it
seems
that
there
are
also
some
tools
to
convert
from
c
to
to
vhdl
to
verilog,
and
then
it
will
be
up
to
the
stylings
compiler
to
cylinders,
place
a
route
to
address,
to
fit
the
clb.
The
cells
in
better
way
use
dsp,
don't
use
dsp,
and
this
will
keep
this
philosophy
to
keep
everything
open
and
everything
usable
by
everybody.
A
A
It
might
not
be
equal.
You
know,
there's
something
some
advantages
to
to
matlab
like
as
far
as
I
can
tell
there
is
not
a
there's,
no
competitor
for
stimulink.
In
in
open
source,
there's
nothing
really
that
that
competes
with
simulink
but
matlab
and
octave
are
very
compatible
and
very
close.
So
yeah,
let
me
let
me
try
to
work
on
that
and
see
see
how
far
we
can
get.
A
A
Wow
all
right!
So
we'll
see
we'll
do
this
again
next
week
and
I
will
I'll
edit
up
our
video
and
and
put
it
to
put
it
up
for
people
that
are
that
are
missing
it.
We
have
a
couple
of
different
people
that
are
moving
on
the
team,
several
actually,
so
so
good
luck
with
all
of
your
moves
out
there.
Those
are
moving
households
and
see
you
on
slack.