Datasets:
Ekhlass
/
flutter_docs

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
1
372
and flame chart).
to load all available CPU samples without manually
recording and stopping, you can click load all CPU samples,
which pulls all CPU samples that the VM has recorded and
stored in its ring buffer, and then displays those
CPU samples in the profiler views.
<topic_end>
<topic_start>
bottom up
this table provides a bottom-up representation
of a CPU profile. this means that each top-level method,
or root, in the bottom up table is actually the
top method in the call stack for one or more CPU samples.
in other words, each top-level method in a bottom up
table is a leaf node from the top down table
(the call tree).
in this table, a method can be expanded to show its callers.
this view is useful for identifying expensive methods
in a CPU profile. when a root node in this table
has a high self time, that means that many CPU samples
in this profile ended with that method on top of the call stack.
see the guidelines section below to learn how to
enable the blue and green vertical lines seen in this image.
tooltips can help you understand the values in each column:
for top-level methods in the bottom-up tree
(stack frames that were at the top of at least one
CPU sample), this is the time the method spent executing
its own code, as well as the code for any methods that
it called.
for top-level methods in the bottom-up tree
(stack frames that were at the top of at least one CPU
sample), this is the time the method spent executing only
its own code.
for children methods in the bottom-up tree (the callers),
this is the self time of the top-level method (the callee)
when called through the child method (the caller).
table element (self time)
<topic_end>
<topic_start>
call tree
this table provides a top-down representation of a CPU profile.
this means that each top-level method in the call tree is a root
of one or more CPU samples. in this table,
a method can be expanded to show its callees.
this view is useful for identifying expensive paths in a CPU profile.
when a root node in this table has a high total time,
that means that many CPU samples in this profile started
with that method on the bottom of the call stack.
see the guidelines section below to learn how to
enable the blue and green vertical lines seen in this image.
tooltips can help you understand the values in each column:
<topic_end>
<topic_start>
method table
the method table provides CPU statistics for each method
contained in a CPU profile. in the table on the left,
all available methods are listed with their total and
self time.
total time is the combined time that a method spent
anywhere on the call stack, or in other words,
the time a method spent executing its own code and
any code for methods that it called.
self time is the combined time that a method spent
on top of the call stack, or in other words,
the time a method spent executing only its own code.
selecting a method from the table on the left shows
the call graph for that method. the call graph shows
a method’s callers and callees and their respective
caller / callee percentages.
<topic_end>
<topic_start>
flame chart
the flame chart view is a graphical representation of
the call tree. this is a top-down view
of a CPU profile, so in this chart,
the top-most method calls the one below it.
the width of each flame chart element represents the
amount of time that a method spent on the call stack.
like the call tree, this view is useful for identifying
expensive paths in a CPU profile.
the help menu, which can be opened by clicking the ? icon
next to the search bar, provides information about how to
navigate and zoom within the chart and a color-coded legend.
<topic_end>
<topic_start>
CPU sampling rate
DevTools sets a default rate at which the VM collects CPU samples:
1 sample / 250 μs (microseconds). this is selected by default on
the CPU profiler page as “cpu sampling rate: medium”.
this rate can be modified using the selector at the top
of the page.
the low, medium, and high sampling rates are
1,000 hz, 4,000 hz, and 20,000 hz, respectively.
it’s important to know the trade-offs
of modifying this setting.
a profile that was recorded with a higher sampling rate
yields a more fine-grained CPU profile with more samples.
this might affect performance of your app since the VM
is being interrupted more often to collect samples.
this also causes the VM’s CPU sample buffer to overflow more quickly.