Skip to content
GitLab
Explore
Sign in
Register
Primary navigation
Search or go to…
Project
P
Poulpe
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Package Registry
Model registry
Operate
Environments
Terraform modules
Monitor
Incidents
Service Desk
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Terms and privacy
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
octobus
Poulpe
Commits
47fb6cf1
Commit
47fb6cf1
authored
1 year ago
by
Raphaël Gomès
Browse files
Options
Downloads
Patches
Plain Diff
Update the benchmarks docs
parent
c998a787
Branches
topic/default/pull-out-of-order.pdb
No related tags found
No related merge requests found
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
docs/benchmarks.rst
+50
-44
50 additions, 44 deletions
docs/benchmarks.rst
with
50 additions
and
44 deletions
docs/benchmarks.rst
+
50
−
44
View file @
47fb6cf1
...
...
@@ -2,8 +2,8 @@
Details on benchmarks in Poulpe
===============================
A benchmark describe "something" we want to get performance
data on. It is
defined in a "poulpe-benchmark-definition" file (`.pbd`).
It will use code that
exists in a
poulpe's "
binary-environment
"
to benchmark
that "something" on the
data present in a
poulpe's "binary
-environment
"
.
A benchmark describe
s
"something" we want to get performance
information about.
It is
defined in a "poulpe-benchmark-definition" file (`.pbd`).
It will use code that
exists in a binary-environment to benchmark
that "something" on the
data present in a
data
-environment.
...
...
@@ -9,4 +9,4 @@
The `.pbd` files include
s
multiple metadata and definition.
Some are common to
all type of benchmarks, some are specific
to each type of benchmarks
.
The `.pbd` files include multiple
kinds of
metadata and definition
s
.
Some are common to
all type
s
of benchmarks, some are specific.
...
...
@@ -12,6 +12,6 @@
Generic metata data
===================
Generic meta
da
ta data
===================
==
section: "meta"
---------------
...
...
@@ -19,10 +19,10 @@
format
~~~~~~
The version of this benchmark definition. Currently known version
are:
The version of this benchmark definition. Currently known version
s
are:
:0: The very first, quite volatile, version.
name
~~~~
...
...
@@ -23,12 +23,12 @@
:0: The very first, quite volatile, version.
name
~~~~
identifier of the benchmark. This will be stored in result.
identifier of the benchmark. This will be stored in result
s
.
method
~~~~~~
The way this benchmark is expected to be run. It will also define most of the
...
...
@@ -30,7 +30,7 @@
method
~~~~~~
The way this benchmark is expected to be run. It will also define most of the
other value we expect to find to configure this benchmark.
other value
s
we expect to find to configure this benchmark.
...
...
@@ -36,5 +36,5 @@
The current list of supported method is
:
The current list of supported method
s
is:
:simple-command: benchmarking the run of a single shell command.
:mercurial-perf-extension: using Mercurial perf extensions to run benchmarks
...
...
@@ -51,9 +51,9 @@
section: environment
--------------------
A section that allow to set environment variable to be used when running the
command. This is useful to replace some of the command argument with. Each
sub-key is
a
an environment variable to replace.
A section that allow
s
to set environment variable
s
to be used when running the
command. This is useful to replace some of the command
's
argument
s
with. Each
sub-key is an environment variable to replace.
Value supports the `DATA-VARS:` notation if necessary
...
...
@@ -68,8 +68,8 @@
The command to be run.
It can
w
e extended with variants (see next section)
It can
b
e extended with variants (see next section)
cwd
~~~
...
...
@@ -72,6 +72,6 @@
cwd
~~~
The directory to run the command
in
.
The directory
in which
to run the command.
...
...
@@ -77,5 +77,5 @@
Value can be substitu
d
ed for data-env variable using the
Value
s
can be substitu
t
ed for data-env variable
s
using the
`"DATA-VARS:path.to.variable.in.the.data.env"` syntax.
For example `"DATA-VARS:mercurial.status.large.all"`
...
...
@@ -87,8 +87,11 @@
prepare-run
~~~~~~~~~~~
A command (or series of command) to run before each command run.
A command (or series of commands) to run before each command run.
Note: this command will be run with bash and prefixed with `set -eEuo pipefail`.
You can opt out of the prefix by specifying "simple-command.no-set-builtin=true"
variants
--------
...
...
@@ -91,7 +94,7 @@
variants
--------
Variants allow for slight variation of a benchmark. For example,
they allow to change the
input data, o
f to
turn some feature on and off.
Variants allow for slight variation
s
of a benchmark. For example,
they allow for changing the
input data, o
r for
turn
ing
some feature on and off.
...
...
@@ -97,5 +100,5 @@
Variants are organi
s
ed within independ
a
nt "dimensions".
Variants are organi
z
ed within independ
e
nt "dimensions".
Each dimension has a set of values, each value can modify the benchmark. Each
dimension must have a default value.
...
...
@@ -107,7 +110,7 @@
simple-command.variants.dimensions.<dimension-name>.<variant-key>
The `dimension-name` and `variant-key` are used as identifier in the benchmark
The `dimension-name` and `variant-key` are used as identifier
s
in the benchmark
result and for selecting variant at run time.
variants variables
...
...
@@ -118,9 +121,9 @@
Exactly one variant must have it set to `true` per dimension.
When set and no variant ha
ve
been explicitly selected for this dimension
, this
variant will be used.
When set and no variant ha
s
been explicitly selected for this dimension
during
a benchmark run, this
variant will be used.
extend-command
~~~~~~~~~~~~~~
...
...
@@ -123,9 +126,9 @@
extend-command
~~~~~~~~~~~~~~
A string to append to the comm
e
nd when this variable is selected.
A string to append to the comm
a
nd when this variable is selected.
cwd
~~~
...
...
@@ -128,6 +131,6 @@
cwd
~~~
overwrite the global `simple-command.cwd` variable. Behave the same.
overwrite
s
the global `simple-command.cwd` variable. Behave
s
the same.
...
...
@@ -133,6 +136,6 @@
Mercurial perf extension
s
configuration
======================================
=
Mercurial perf extension configuration
======================================
section: "hg-perf-ext"
----------------------
...
...
@@ -151,8 +154,8 @@
args
~~~~
Arguments to pass to the command (a list of string).
Arguments to pass to the command (a list of string
s
).
setup-script
~~~~~~~~~~~~
...
...
@@ -155,9 +158,12 @@
setup-script
~~~~~~~~~~~~
a shell script to run before running the perf-command.
A shell script to run before running the perf-command.
Note: this command will be run with bash and prefixed with `set -eEuo pipefail`.
You can opt out of the prefix by specifying "hg-perf-ext.no-set-builtin=true"
constraints
~~~~~~~~~~~
...
...
@@ -160,10 +166,10 @@
constraints
~~~~~~~~~~~
These
s
variables control when this benchmark can be run.
These variables control when this benchmark can be run.
Here is the list of currently supported constraints:
- `graph.visible-revision-count.min=<integer>`:
...
...
@@ -165,10 +171,10 @@
Here is the list of currently supported constraints:
- `graph.visible-revision-count.min=<integer>`:
Target repository must have at least that many visible revision for this
Target repository must have at least that many visible revision
s
for this
benchmark to run.
- `compatible-revision-range=<revset>`:
...
...
@@ -171,10 +177,10 @@
benchmark to run.
- `compatible-revision-range=<revset>`:
The range of revision compatible with th
is changes.
The range of revision
s
compatible with th
e command being run.
variants
--------
...
...
@@ -176,8 +182,8 @@
variants
--------
Variants allow for slight variation of a benchmark. For example,
they allow to change the
input data, of
to
turn some feature on and off.
Variants allow for slight variation
s
of a benchmark. For example,
they allow for changing the
input data, of
for
turn
ing
some feature on and off.
...
...
@@ -183,5 +189,5 @@
Variants are organi
s
ed within independ
a
nt "dimensions".
Variants are organi
z
ed within independ
e
nt "dimensions".
Each dimension has a set of values, each value can modify the benchmark. Each
dimension must have a default value.
...
...
@@ -189,7 +195,7 @@
defining a variant
~~~~~~~~~~~~~~~~~~
Variants are defined under keys following this pattern
s
:
Variants are defined under keys following this pattern:
hg-perf-ext.variants.dimensions.<dimension-name>.<variant-key>
...
...
@@ -193,8 +199,8 @@
hg-perf-ext.variants.dimensions.<dimension-name>.<variant-key>
The `dimension-name` and `variant-key` are used as identifier in the benchmark
result and for selecting variant at run time.
The `dimension-name` and `variant-key` are used as identifier
s
in the benchmark
result and for selecting variant
s
at run time.
variants variables
~~~~~~~~~~~~~~~~~~
...
...
@@ -204,7 +210,7 @@
Exactly one variant must have it set to `true` per dimension.
When set and no variant ha
ve
been explicitly selected for this dimension, this
When set and no variant ha
s
been explicitly selected for this dimension, this
variant will be used.
args
...
...
@@ -208,10 +214,10 @@
variant will be used.
args
~~~~
~~~~~~~~~~
~~~~
A list of args to add to the command
cwd
~~~
...
...
@@ -212,11 +218,11 @@
A list of args to add to the command
cwd
~~~
o
verwrite the global `hg-perf-ext.cwd` variable. Behave the same.
O
verwrite
s
the global `hg-perf-ext.cwd` variable. Behave
s
the same.
setup-script
~~~~~~~~~~~~
...
...
@@ -219,5 +225,5 @@
setup-script
~~~~~~~~~~~~
o
verwrite the global `hg-perf-ext.setup-script` variable. Behave the same.
O
verwrite
s
the global `hg-perf-ext.setup-script` variable. Behave
s
the same.
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment