From 1a60faf70e303ab9d0aa5dc1a993616a220313b6 Mon Sep 17 00:00:00 2001
From: Philippe PITTOLI <karchnu@karchnu.fr>
Date: Sat, 1 Jun 2024 03:22:22 +0200
Subject: [PATCH] bin/

---
 bin/extract-data-benchmark-cars.sh | 16 ++++++++++++
 bin/rsum2line.awk                  | 39 +++++++++++++++++++++++++++
 bin/stats.sh                       | 42 ++++++++++++++++++++++++++++++
 bin/summary-to-truncated-data.sh   | 16 ++++++++++++
 bin/summary.r                      | 14 ++++++++++
 5 files changed, 127 insertions(+)
 create mode 100755 bin/extract-data-benchmark-cars.sh
 create mode 100755 bin/rsum2line.awk
 create mode 100755 bin/stats.sh
 create mode 100755 bin/summary-to-truncated-data.sh
 create mode 100644 bin/summary.r

diff --git a/bin/extract-data-benchmark-cars.sh b/bin/extract-data-benchmark-cars.sh
new file mode 100755
index 0000000..b976d61
--- /dev/null
+++ b/bin/extract-data-benchmark-cars.sh
@@ -0,0 +1,16 @@
+#!/bin/sh
+if [ $# -ne 1 ]
+then
+	echo "usage: $0 result-directory"
+	exit 0
+fi
+
+d=$1
+
+awk '{ print $1 }' < $d/ram_index.d | sort -n | uniq > it
+mkdir data
+
+echo "from truncated data (.t) to graphed data data/XXX.d"
+paste it $d/ram_index.t      $d/cached_index.t      $d/fifo_index.t      $d/semi_index.t      $d/uncached_index.t      > ./data/index.d
+paste it $d/ram_partitions.t $d/cached_partitions.t $d/fifo_partitions.t $d/semi_partitions.t $d/uncached_partitions.t > ./data/partitions.d
+paste it $d/ram_tags.t       $d/cached_tags.t       $d/fifo_tags.t       $d/semi_tags.t       $d/uncached_tags.t       > ./data/tags.d
diff --git a/bin/rsum2line.awk b/bin/rsum2line.awk
new file mode 100755
index 0000000..e62528d
--- /dev/null
+++ b/bin/rsum2line.awk
@@ -0,0 +1,39 @@
+#!/usr/bin/awk -f
+
+BEGIN {
+	FOUND_95pct = 0
+	FOUND_mean  = 0
+}
+
+FOUND_95pct == 1 {
+	pct_min = $1
+	pct_max = $2
+	FOUND_95pct = 0
+}
+
+FOUND_mean == 1 {
+	mean = $1
+	print pct_min, median, mean, pct_max, t, df, pvalue
+	FOUND_mean = 0
+}
+
+/^t = / {
+	gsub(",", "", $3)
+	t = $3
+	gsub(",", "", $6)
+	df = $6
+	pvalue = $9
+}
+
+/mean of x/ {
+	FOUND_mean = 1
+}
+
+/Median/ {
+	gsub(":", "")
+	median = $2
+}
+
+/95 percent confidence/ {
+	FOUND_95pct = 1
+}
diff --git a/bin/stats.sh b/bin/stats.sh
new file mode 100755
index 0000000..4a660dd
--- /dev/null
+++ b/bin/stats.sh
@@ -0,0 +1,42 @@
+#!/bin/sh
+
+# .raw -> bad format -> .summary (great format)
+raw2sum="./bin/raw-to-summary.sh"
+# .summary (with too much data) -> truncated data (.t)
+truncate_data="./bin/summary-to-truncated-data.sh"
+# ./bin/extract-data-*.sh: .t -> data/XXX.d (paste an index + *.t)
+
+if [ $# -ne 1 ]
+then
+	echo "usage: $0 result-directory"
+	exit 0
+fi
+
+dir="$1"
+
+$raw2sum "${dir}"
+
+# List raw files with the number of iterations as a prefix so they can then be sorted.
+sort_summary_files() {
+	for i in $dir/*.summary ; do f $i ; done | sort -n
+}
+
+f() {
+	echo $* | sed "s/[_./]/ /g" | xargs echo "$* " | awk '{ printf "%s %s/%s_%s %s\n", $4, $2, $3, $5, $1 }'
+}
+
+fill() {
+	# Remove previous computations.
+	rm ${dir}/*.d
+	while read LINE; do
+		nb_it=$(echo $LINE | awk '{ print $1 }')
+		target=$(echo $LINE | awk '{ print $2 }')
+		fname=$(echo $LINE | awk '{ print $3 }')
+
+		cat $fname | xargs echo "$nb_it " >> $target.d
+	done
+}
+
+sort_summary_files | fill
+
+$truncate_data "${dir}"
diff --git a/bin/summary-to-truncated-data.sh b/bin/summary-to-truncated-data.sh
new file mode 100755
index 0000000..f01885e
--- /dev/null
+++ b/bin/summary-to-truncated-data.sh
@@ -0,0 +1,16 @@
+#!/bin/sh
+
+if [ $# -ne 1 ]
+then
+	echo "usage: $0 result-directory"
+	exit 0
+fi
+
+dir=$1
+
+echo "from data (.d) to truncated data (.t)"
+for i in $dir/*.d
+do
+	fname=$(echo $i | sed "s/[.]d$/.t/")
+	awk '{ print $2, $3, $5 }' < $i > $fname
+done
diff --git a/bin/summary.r b/bin/summary.r
new file mode 100644
index 0000000..a3f18a3
--- /dev/null
+++ b/bin/summary.r
@@ -0,0 +1,14 @@
+# Rscript summary handshake-duration.txt
+
+require(grDevices) # for colours
+
+tbl <- read.table(file=commandArgs(TRUE)[1])
+val <- tbl[1]
+
+summary(val)
+# standarddeviation=sd(unlist(val))
+sd(unlist(val))
+# print (standarddeviation, zero.print="standard deviation: ")
+
+# confint.default (val)
+t.test (val)