[00:02]
when chatting with someone on Jupiter's
[00:03]
thirdd largest moon you will notice a
[00:06]
distinct lag in Communications that is
[00:08]
due to async
[00:14]
IO when you try to perform a task that
[00:16]
requires a lot of waiting you may notice
[00:19]
an annoying lag in execution you can
[00:21]
solve this with async IO this module
[00:25]
brings the await and async syntax to
[00:27]
python this pair of felet words empowers
[00:30]
you with the tools to easily manage a
[00:33]
collection of laggy
[00:34]
tasks sometimes impatience is a
[00:40]
virtue code can be categorized as
[00:43]
synchronous or
[00:44]
asynchronous in synchronous programming
[00:47]
tasks are performed one at a time in the
[00:49]
order they are called each task must
[00:52]
finish before the next one can
[00:54]
begin with asynchronous programming
[00:56]
however different tasks can start
[00:59]
process and finish in overlapping
[01:01]
periods of time for example imagine a
[01:04]
task that submits many HTTP requests in
[01:08]
the synchronous World these HTTP
[01:11]
requests must be completed one at a time
[01:14]
in the asynchronous world each request
[01:17]
can start and be set aside while waiting
[01:19]
for a response during this waiting
[01:21]
period different tasks can step forward
[01:24]
and use the Computing resources this
[01:27]
nonlinear approach saves one of the most
[01:29]
Val resources around time to better
[01:34]
understand the mechanics of synchronous
[01:35]
and asynchronous execution you need to
[01:37]
learn about sub routines and co-routines
[01:40]
a sub routine is a block of code that
[01:42]
can be called as needed when a sub
[01:44]
routine is called control of the program
[01:47]
is transferred to that sub routine when
[01:49]
its work is done control is returned
[01:52]
back to the main program and execution
[01:54]
continues from where it left off sub
[01:57]
routines cannot be paused and resumed
[02:00]
they run until
[02:02]
done a co- routine is a special kind of
[02:05]
function that can have its executions
[02:07]
paused and resumed this is possible
[02:10]
because they maintain their state
[02:12]
between pauses co- routines are perfect
[02:15]
for tasks that need to wait for
[02:17]
something like IO operations database
[02:20]
calls and HTTP requests the terms sub
[02:24]
routine and co-routine are derived from
[02:27]
the roles they play and how they
[02:28]
function sub routine is a delightful
[02:31]
blend of sub and routine as the name
[02:35]
suggests it is a subset of a larger
[02:37]
program co- routine is a ju toos of Co
[02:41]
and routine Co for together and routine
[02:46]
for
[02:47]
routine this name was chosen by someone
[02:49]
to describe routines that can run
[02:52]
cooperatively co-routines are the key
[02:54]
ingredient to make asynchronous
[02:56]
programming or concurrency possible in
[02:59]
in a typical single-threaded app all of
[03:02]
the code and sub routines run
[03:04]
sequentially this is simple but can be
[03:07]
wasteful there are two different
[03:08]
approaches that make better use of
[03:10]
available Computing resources
[03:12]
concurrency and
[03:14]
parallelism using concurrency the start
[03:17]
and stop times of multiple co- routines
[03:20]
can overlap with parallelism different
[03:23]
threads can execute at the same time is
[03:27]
there going to be any python in this
[03:28]
python lesson
[03:30]
yes yes there is but before we can enter
[03:34]
python mode I must import an important
[03:37]
message socratica is now in the course
[03:39]
business visit socratic.com to see the
[03:42]
excitement for yourself and for an
[03:45]
asynchronous update about future courses
[03:47]
feverishly type your email address in a
[03:50]
text box somewhere on the site now back
[03:54]
to python we are now going to show you
[03:56]
how to use concurrency in Python with
[03:58]
the async and await keywords and the
[04:01]
async io
[04:03]
module let us create two functions brew
[04:08]
coffee and toast
[04:12]
Bagel The Brew Coffee function will use
[04:14]
the sleep function to simulate a command
[04:17]
that takes 3 seconds to complete the
[04:19]
toast Bagel function will pretend it
[04:21]
takes 2 seconds to complete with this
[04:24]
5-second toasty breakfast the future is
[04:26]
fast and
[04:28]
carbolicious we will record a start time
[04:31]
call both
[04:33]
functions record a stop
[04:35]
time compute the elaps time and print
[04:38]
the
[04:44]
results
[04:50]
run this code took 5 seconds and a bit
[04:53]
more to execute but there is no reason
[04:56]
to wait for the coffee to finish before
[04:58]
you start toasting the B Bagel such
[05:00]
inefficiency gives the secretly embedded
[05:03]
AI systems far too much time for
[05:05]
Mischief let us modify this code so it
[05:08]
runs
[05:09]
concurrently with
[05:11]
co-routines to turn each function into a
[05:14]
co- routine add the async keyword this
[05:18]
is the first step but not the last as is
[05:21]
evident if you run using the async
[05:24]
keyword to make a co- routine is
[05:26]
necessary but not sufficient you also
[05:29]
need to to specify where in the co-
[05:30]
routine it is safe to pause and yield
[05:33]
control to other co- routines you do
[05:35]
this with the await
[05:38]
keyword this brings us to another
[05:40]
essential concept you can only put await
[05:43]
in front of commands that are awaitable
[05:46]
the sleep function is not
[05:48]
awaitable luckily the async io module
[05:51]
gives you an awaitable version of the
[05:53]
sleep
[05:54]
function you can call the co- routines
[05:57]
individually or in a batch we will first
[06:00]
demonstrate the batch approach the async
[06:03]
io module has a gather function to group
[06:05]
co- routines for concurrent execution
[06:08]
the arguments determine which co-
[06:10]
routines will run
[06:11]
concurrently the arguments may appear to
[06:13]
be regular function calls but they are
[06:15]
not calling a co-routine returns a
[06:18]
co-routine object not a typical return
[06:21]
value these co-routine objects give
[06:24]
async IO the ability to start and stop
[06:27]
their
[06:28]
execution to get the return values from
[06:30]
the co- routines you need to await them
[06:34]
and note that you await the return
[06:35]
values in the same order you pass the
[06:37]
co- routines to The Gather function
[06:40]
order matters and the use of a weight
[06:43]
here is not optional after all we have
[06:47]
to wait for both co- routines to finish
[06:50]
so it only makes sense that we have to
[06:51]
wait on the batch too we are so close to
[06:54]
the finish line but there is one final
[06:57]
step see how we await for the batch
[07:00]
results any function that has an await
[07:02]
keyword must be declared async this
[07:06]
means we must put the async keyword
[07:08]
before the main
[07:09]
function because main has now become a
[07:12]
co-routine we have to call it in a
[07:14]
slightly different way the async io
[07:17]
module has you covered just call the Run
[07:20]
function and pass in the main co-
[07:22]
routine now inhale widen your eyes and
[07:28]
run
[07:32]
Joy of Joys our breakfast was completed
[07:35]
in 3 seconds instead of five time saved
[07:39]
me
[07:41]
happy there is an alternative to using
[07:43]
the gather function you can use the
[07:46]
create task function in the async io
[07:48]
module to create a task out of each co-
[07:52]
routine then you can await each co-
[07:54]
routine
[07:57]
individually run
[08:03]
just as fast just as happy so which is
[08:08]
better working in batches or with
[08:10]
individual tasks it is your choice so
[08:14]
Choose Wisely computer chips want to
[08:17]
compute that is their singular purpose
[08:19]
in life the async io module was built to
[08:23]
help chips realize their full potential
[08:26]
by adding a pause button to objects you
[08:28]
enable them to take turns sharing is
[08:31]
caring after all and speaking of sharing
[08:34]
many human people ask you to share their
[08:36]
videos with as many human people as
[08:38]
possible this is not an efficient use of
[08:41]
resources instead await the opportunity
[08:44]
to yield this video to someone who has
[08:47]
finished a task and is thinking about
[08:50]
the
[08:54]
[Music]
[08:58]
future
[09:02]
[Music]