Add updated DIS post-processing scripts from (rejected) RTSS'20 paper

These are compatible with the outputs of all the DIS stressmarks
as instumented with `extra.h`.

4 files changed, 69 insertions, 23 deletions

diff --git a/.gitignore b/.gitignore
index af908cb..24a4e1f 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,16 +2,44 @@
 *.txt
 *.txt.clean
 
-# Ignore DIS benchmarks
+# Ignore DIS benchmark binaries
 **/field
 **/matrix
 **/neighborhood
 **/pointer
 **/transitive
 **/update
+# Ignore DIS all-pairs binaries
+**/field_all
+**/matrix_all
+**/neighborhood_all
+**/pointer_all
+**/transitive_all
+**/update_all
 
 # Ignore Tacle Benchmarks
-**/bin/
+**/adpcm_dec
+**/adpcm_enc
+**/ammunition
+**/cjpeg_transupp
+**/cjpeg_wrbmp
+**/dijkstra
+**/epic
+**/fmref
+**/gsm_dec
+**/gsm_enc
+**/h264_dec
+**/huff_enc
+**/mpeg2
+**/ndes
+**/petrinet
+**/rijndael_dec
+**/rijndael_enc
+**/statemate
+**/susan
 
 # Ignore random walk benchmark
 **/random_walk
+
+# Ignore misc debug and build outputs
+**/.gdb_history
diff --git a/dis/clean.sh b/dis/clean.sh
index 7c58295..1b6d2f5 100755
--- a/dis/clean.sh
+++ b/dis/clean.sh
@@ -1 +1,6 @@
-tail -n +2 $1 | tr "-" " " | sed "s/L3:0=0000;1=0000;2=0000;3=0000/0/" | sed "s/L3:0=0000;1=0000;2=0000;3=0001/1/" | sed "s/L3:0=0000;1=0000;2=0000;3=0003/2/" | sed "s/L3:0=0000;1=0000;2=0000;3=0007/3/" | sed "s/L3:0=0000;1=0000;2=0000;3=000f/4/" | sed "s/L3:0=0000;1=0000;2=0000;3=003f/6/" | sed "s/L3:0=0000;1=0000;2=0000;3=00ff/8/" | sed "s/L3:0=0000;1=0000;2=0000;3=03ff/10/" | sed "s/L3:0=0000;1=0000;2=0000;3=0fff/12/" | sed "s/L3:0=0000;1=0000;2=0000;3=3fff/14/" | sed "s/L3:0=0000;1=0000;2=0000;3=ffff/16/" > $1.clean
+#!/bin/bash
+# Input: Output of ./summarize.py
+# This script primarially converts an LLC assignment mask to a decimal number
+# of assigned cache ways. It also removes the first line (column headers) and the
+# last three lines (summary statistics)
+tail -n +2 $1 | head -n -3 | tr "-" " " | tr "+" " " | sed "s/L3:0=0000;1=0000;2=0000;3=0000/0/g" | sed "s/L3:0=0000;1=0000;2=0000;3=0001/1/g" | sed "s/L3:0=0000;1=0000;2=0000;3=0003/2/g" | sed "s/L3:0=0000;1=0000;2=0000;3=0007/3/g" | sed "s/L3:0=0000;1=0000;2=0000;3=000f/4/g" | sed "s/L3:0=0000;1=0000;2=0000;3=001f/5/g" | sed "s/L3:0=0000;1=0000;2=0000;3=003f/6/g" | sed "s/L3:0=0000;1=0000;2=0000;3=007f/7/g" | sed "s/L3:0=0000;1=0000;2=0000;3=00ff/8/g" | sed "s/L3:0=0000;1=0000;2=0000;3=01ff/9/g" | sed "s/L3:0=0000;1=0000;2=0000;3=03ff/10/g" | sed "s/L3:0=0000;1=0000;2=0000;3=07ff/11/g" | sed "s/L3:0=0000;1=0000;2=0000;3=0fff/12/g" | sed "s/L3:0=0000;1=0000;2=0000;3=1fff/13/g" | sed "s/L3:0=0000;1=0000;2=0000;3=3fff/14/g" | sed "s/L3:0=0000;1=0000;2=0000;3=7fff/15/g" | sed "s/L3:0=0000;1=0000;2=0000;3=ffff/16/g" > $1.clean
diff --git a/dis/postproc.sh b/dis/postproc.sh
index c5dd1c9..efe6774 100755
--- a/dis/postproc.sh
+++ b/dis/postproc.sh
@@ -1,3 +1,18 @@
 #!/bin/bash
-./summarize.py $1.txt > $1-pp.txt
+# This takes the saved output from the timing code in `extra.h`, summarizes
+# it, and converts LLC masks to decimals.
+# This is compatible with the output of all the current DIS runner scripts.
+if echo $1 | grep -q ".txt"; then
+    echo "Don't include .txt! Exiting..."
+    exit
+fi
+if echo $1 | grep -q "\-pp"; then
+    echo "Don't include -pp! Exiting..."
+    exit
+fi
+if [[ ! -f $1.txt ]]; then
+    echo "Input does not exist. Exiting..."
+    exit
+fi
+./summarize.py $1.txt --inf-precision > $1-pp.txt
 ./clean.sh $1-pp.txt
diff --git a/dis/summarize.py b/dis/summarize.py
index f8f6929..31dcccc 100755
--- a/dis/summarize.py
+++ b/dis/summarize.py
@@ -2,6 +2,8 @@
 import sys
 
 f = sys.argv[1]
+if len(sys.argv) < 2:
+    print("Usage "+sys.argv[0]+" <filename> --inf-precision")
 res = {}
 mem_res = {}
 memw_res = {}
@@ -48,24 +50,20 @@ def percentile(N, percent, key=lambda x:x):
     d1 = key(N[int(c)]) * (k-f)
     return d0+d1
 ## end of http://code.activestate.com/recipes/511478/ }}}
-"""
-print("Average times:")
-for r in res.keys():
-    print(res[r]/samples[r])
-
-print("Average memory read:")
-for r in mem_res.keys():
-    print(mem_res[r]/samples[r])
-
-print("Average memory write:")
-for r in memw_res.keys():
-    print(memw_res[r]/samples[r])
-
-print("Max times:")
-for r in max_res.keys():
-    print(max_res[r])
-"""
+avg_max = 0
+avg_99th = 0
+avg_avg = 0
 print("Name\t\tAverage\t\tMax\t\t99th %\t\tAvg Mem Read\tAvg Mem Write\t")
-for r in res.keys():
+for r in sorted(res.keys()):
 #    print(r + "\t\t" + str(res[r]/samples[r]) + "\t\t" + str(max_res[r]))
-    print("{:12}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}".format(r, sum(res[r])/len(res[r]), max(res[r]), percentile(sorted(res[r]), 0.99), mem_res[r]/samples[r], memw_res[r]/samples[r]))
+    avg_avg += sum(res[r])/len(res[r])
+    avg_max += max(res[r])
+    avg_99th += percentile(sorted(res[r]), 0.99)
+    if len(sys.argv) > 2 and sys.argv[2] == "--inf-precision":
+        print("{:12}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}".format(r, sum(res[r])/len(res[r]), max(res[r]), percentile(sorted(res[r]), 0.99), mem_res[r]/samples[r], memw_res[r]/samples[r]))
+    else:
+        print("{:12.12}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}\t{:>12.0f}".format(r, sum(res[r])/len(res[r]), max(res[r]), percentile(sorted(res[r]), 0.99), mem_res[r]/samples[r], memw_res[r]/samples[r]))
+
+print("Average average:", avg_avg/len(res.keys()))
+print("Average max:", avg_max/len(res.keys()))
+print("Average 99th:", avg_99th/len(res.keys()))