benchmarks: Improve readme and convert to Markdown
authorTimo Tijhof <krinklemail@gmail.com>
Sat, 27 Jan 2018 00:31:54 +0000 (16:31 -0800)
committerTimo Tijhof <krinklemail@gmail.com>
Sat, 27 Jan 2018 00:31:54 +0000 (16:31 -0800)
* Add sections.
* Rename to *.md.
* Document that taskset is Linux-specific.

Change-Id: If6c5e54eda4dd1d6ea02445efa73e356dbc49016

maintenance/benchmarks/README [deleted file]
maintenance/benchmarks/README.md [new file with mode: 0644]

diff --git a/maintenance/benchmarks/README b/maintenance/benchmarks/README
deleted file mode 100644 (file)
index 27da9de..0000000
+++ /dev/null
@@ -1,10 +0,0 @@
-This directory hold several benchmarking scripts used as a proof of speed
-or to track PHP performances over time.
-
-To get somehow accurate result, you might want to bound the PHP process
-to a specific CPU with `taskset` and raise its priority with `nice`. Example:
-
- $ taskset 1 nice -n-10 php bench_wfIsWindows.php
-
-australia-untidy.html.gz contains representative input text for
-benchmarkTidy.php. It needs to be decompressed before use.
diff --git a/maintenance/benchmarks/README.md b/maintenance/benchmarks/README.md
new file mode 100644 (file)
index 0000000..b411c52
--- /dev/null
@@ -0,0 +1,15 @@
+This directory hold several benchmarking scripts used track performances of
+MediaWiki and/or PHP.
+
+## Consistency
+
+On Linux, use of `taskset` and `nice` can help get more consistent results.
+
+For example:
+
+ $ taskset 1 nice -n-10 php bench_wfIsWindows.php
+
+## Fixtures
+
+* australia-untidy.html.gz: Representative input text for benchmarkTidy.php.
+  It needs to be decompressed before use.