From 02d86d3e5216e51661974d069e1a7ed12fc12bad Mon Sep 17 00:00:00 2001
From: Ludwig Fueracker <ludwig.fueracker@fau.de>
Date: Sun, 29 Jan 2017 20:46:05 +0100
Subject: [PATCH] fixed axis units, exclude broken commits from build graphs
---
evaluate_data_with_stop.py | 68 +++++++++++++++++++++++++++-----------
validate_hashes.py | 56 ++++++++++++++++---------------
2 files changed, 77 insertions(+), 47 deletions(-)
diff --git a/evaluate_data_with_stop.py b/evaluate_data_with_stop.py
index 622eecd..53a5e11 100755
--- a/evaluate_data_with_stop.py
+++ b/evaluate_data_with_stop.py
@@ -150,7 +150,7 @@ def write_to_csv(data, column_names, filename):
def print_avg(data, name):
- print 'avg %s: %f' % (name, sum(data)/float(len(data)))
+ print 'avg %s: %d' % (name, int(sum(data)/float(len(data))))
################################################################################
@@ -160,18 +160,18 @@ def plot_build_time_graph1(data):
plot_build_time_composition_graph(data[0], data[1], data[2], data[3])
-def plot_build_time_composition_graph(parse_times, hash_times, compile_times, diff_to_build_time): # times in ms
+def plot_build_time_composition_graph(parse_times, hash_times, compile_times, diff_to_build_time): # times in ns
fig, ax = plt.subplots()
-
+#[i/1e6 for i in parse_times],
ax.stackplot(np.arange(1, len(parse_times)+1), # x axis
- [parse_times, hash_times, compile_times,
+ [[i/1e6 for i in parse_times], [i/1e6 for i in hash_times],[i/1e6 for i in compile_times], # ns to ms
#diff_to_build_time
], colors=[parse_color,hash_color,compile_color,
# remain_color
], edgecolor='none')
plt.xlim(1,len(parse_times))
plt.xlabel('commits')
- plt.ylabel('time [s]')
+ plt.ylabel('time [ms]')
ax.set_yscale('log')
lgd = ax.legend([#mpatches.Patch(color=remain_color),
mpatches.Patch(color=compile_color),
@@ -182,16 +182,6 @@ def plot_build_time_composition_graph(parse_times, hash_times, compile_times, di
loc='center left', bbox_to_anchor=(1, 0.5))
fig.savefig(abs_path(BUILD_TIME_FILENAME), bbox_extra_artists=(lgd,), bbox_inches='tight')
- print "\n-----------------"
- print "average total times per build:"
- print_avg(parse_times, 'parse')
- print_avg(hash_times, 'hash')
- print_avg(compile_times, 'compile')
- print_avg(diff_to_build_time, 'remainder')
- print ""
- print "average times if header/source file touched"
- print "-----------------\n"
-
################################################################################
@@ -205,7 +195,12 @@ def make_graphs(full_record):
diff_to_build_times = []
parse_times_header_touched = []
+ hash_times_header_touched = []
+ compile_times_header_touched = []
+
parse_times_source_touched = []
+ hash_times_source_touched = []
+ compile_times_source_touched = []
# freshBuildRecord = full_record[0]
@@ -231,26 +226,59 @@ def make_graphs(full_record):
total_hash_duration += current_file_record[tr('hash-duration')]
total_compile_duration += current_file_record[tr('compile-duration')]
+ if current_record[tr('filename')].endswith('.h'):
+ parse_times_header_touched.append(total_parse_duration)
+ hash_times_header_touched.append(total_hash_duration)
+ compile_times_header_touched.append(total_compile_duration)
+ elif current_record[tr('filename')].endswith('.c'):
+ parse_times_source_touched.append(total_parse_duration)
+ hash_times_source_touched.append(total_hash_duration)
+ compile_times_source_touched.append(total_compile_duration)
+ else:
+ print "unknown file extension: " + filename
+
# if total_parse_duration == 0:# or (total_compile_duration/1e6) > 500000:
# continue
- total_parse_times.append(total_parse_duration / 1e6) # nano to milli
- total_hash_times.append(total_hash_duration / 1e6)
- total_compile_times.append(total_compile_duration / 1e6)
+ total_parse_times.append(total_parse_duration)
+ total_hash_times.append(total_hash_duration)
+ total_compile_times.append(total_compile_duration)
build_time = current_record[tr('build-time')]
- total_build_times.append(build_time / 1e6)
- diff_to_build_times.append((build_time - total_parse_duration - total_hash_duration - total_compile_duration) / 1e6)
+ total_build_times.append(build_time)
+ diff_to_build_times.append((build_time - total_parse_duration - total_hash_duration - total_compile_duration))
print 'run_id %d, #files_changed: %d' % (run_id, files_changed)
+
+ print "\n---- Results ----"
+ print "avg total build times [ns]"
print_avg(total_build_times, 'total')
+ print "-----------------"
+ print "average times if header file touched [ns]"
+ print_avg(parse_times_header_touched, 'parse')
+ print_avg(hash_times_header_touched, 'hash')
+ print_avg(compile_times_header_touched, 'compile')
+ print "-----------------"
+ print "average times if header source touched [ns]"
+ print_avg(parse_times_source_touched, 'parse')
+ print_avg(hash_times_source_touched, 'hash')
+ print_avg(compile_times_source_touched, 'compile')
+ print "-----------------"
+ print "average total times per build [ns]:"
+ print_avg(total_parse_times, 'parse')
+ print_avg(total_hash_times, 'hash')
+ print_avg(total_compile_times, 'compile')
+ print_avg(diff_to_build_times, 'remainder')
+ print "-----------------\n"
# save data to csv files
build_time_data = np.column_stack((total_parse_times, total_hash_times, total_compile_times, diff_to_build_times, total_build_times))
write_to_csv(build_time_data, BUILD_TIME_DATA_HEADER, abs_path(BUILD_TIME_DATA_FILENAME))
plot_build_time_composition_graph(total_parse_times, total_hash_times, total_compile_times, diff_to_build_times)
+
+
################################################################################
"""functions for reading data from the csv files to skip full record building"""
diff --git a/validate_hashes.py b/validate_hashes.py
index 5c23bac..370a7a4 100755
--- a/validate_hashes.py
+++ b/validate_hashes.py
@@ -119,7 +119,7 @@ def validate_records():
print ""
print "avg times:"
for k,v in sum_of_times.items():
- print "%s: %d" % (k, v/nr_of_records)
+ print "%s: %d ns" % (k, v/nr_of_records)
print "-----------------\n"
write_to_csv([ [k,len(v)] for k,v in ast_hashes_dict.items() ], ['filename', 'nr of different hashes'], abs_path('different_ast_hashes_per_file.csv'))
@@ -332,15 +332,15 @@ def plot_build_time_graph1(data):
def plot_build_time_graph(measuredBuildTimes, realClangHashBuildTimes, optimalClangHashBuildTimes, optimalBuildTimes): # times in s
fig, ax = plt.subplots()
- ax.plot(measuredBuildTimes, label='measured build time')
- ax.plot(realClangHashBuildTimes, label='real clang-hash build time')
- ax.plot(optimalClangHashBuildTimes, label='optimal clang-hash build time')
- ax.plot(optimalBuildTimes, label='optimal build time')
+ ax.plot([i/60 for i in measuredBuildTimes], label='measured build time')
+ ax.plot([i/60 for i in realClangHashBuildTimes], label='real clang-hash build time')
+ ax.plot([i/60 for i in optimalClangHashBuildTimes], label='optimal clang-hash build time')
+ ax.plot([i/60 for i in optimalBuildTimes], label='optimal build time')
lgd = ax.legend(loc='center left', bbox_to_anchor=(1, 0.5)) # legend on the right
-
+ ax.set_ylim([0,5])
plt.xlabel('commits')
- plt.ylabel('time [ms]')
+ plt.ylabel('time [min]')
fig.savefig(abs_path(BUILD_TIMES_GRAPH_FILENAME), bbox_extra_artists=(lgd,), bbox_inches='tight')
@@ -375,19 +375,19 @@ def plot_build_time_composition_graph(parseTimes, hashTimes, compileTimes, diffT
print_avg(diffToBuildTime, 'remainder')
-def plotTimeHistogram(times, filename):
+def plotTimeHistogram(times, filename): # times in ms
#TODO: understand params and vars
- hist, bins = np.histogram(times, bins=50)
+ hist, bins = np.histogram([i/1000 for i in times], bins=50) # times to s
width = 0.7 * (bins[1] - bins[0])
center = (bins[:-1] + bins[1:]) / 2
fig, ax = plt.subplots()
- plt.xlabel('time [ms]')
+ plt.xlabel('time [s]')
plt.ylabel('#files')
ax.bar(center, hist, align='center', width=width)
fig.savefig(filename)
-def plotTimeMultiHistogram(parseTimes, hashTimes, compileTimes, filename):
+def plotTimeMultiHistogram(parseTimes, hashTimes, compileTimes, filename): # times in ms
bins = np.linspace(0, 5000, 50)
data = np.vstack([parseTimes, hashTimes, compileTimes]).T
fig, ax = plt.subplots()
@@ -398,9 +398,9 @@ def plotTimeMultiHistogram(parseTimes, hashTimes, compileTimes, filename):
fig.savefig(filename)
fig, ax = plt.subplots()
- data = [parseTimes, hashTimes, compileTimes]
- plt.boxplot(data, 0, 'rs', 0, [5, 95])
- plt.xlabel('time [ms]')
+ boxplot_data = [[i/1000 for i in parseTimes], [i/1000 for i in hashTimes], [i/1000 for i in compileTimes]] # times to s
+ plt.boxplot(boxplot_data, 0, 'rs', 0, [5, 95])
+ plt.xlabel('time [s]')
plt.yticks([1, 2, 3], ['parsing', 'hashing', 'compiling'])
#lgd = ax.legend(loc='center left', bbox_to_anchor=(1, 0.5)) # legend on the right
fig.savefig(filename[:-4] + '_boxplots' + GRAPH_EXTENSION)
@@ -570,22 +570,24 @@ def make_graphs(full_record):
optimalAstHashBuildTime = buildTime - totalOptimalRedundantCompileTime
- measuredBuildTimes.append(buildTime / 1e9) # nano to seconds
- optimalBuildTimes.append(optimalBuildTime / 1e9)
- optimalClangHashBuildTimes.append(optimalAstHashBuildTime / 1e9)
- realClangHashBuildTimes.append(realAstHashBuildTime / 1e9)
+ #TODO: remove broken commits; ok?
+ if buildTime > 3e12 and totalParseDuration/1e9 > 300:
+ measuredBuildTimes.append(buildTime / 16e9) # nano to seconds; also /16 to account for make -j16
+ optimalBuildTimes.append(optimalBuildTime / 16e9)
+ optimalClangHashBuildTimes.append(optimalAstHashBuildTime / 16e9)
+ realClangHashBuildTimes.append(realAstHashBuildTime / 16e9)
- totalParseTimes.append(totalParseDuration / 1e9) # nano to seconds
- totalHashTimes.append(totalHashDuration / 1e9)
- totalCompileTimes.append(totalCompileDuration / 1e9)
- diffToBuildTime.append((buildTime - totalParseDuration - totalHashDuration - totalCompileDuration) / 1e9)
+ totalParseTimes.append(totalParseDuration / 16e9) # nano to seconds
+ totalHashTimes.append(totalHashDuration / 16e9)
+ totalCompileTimes.append(totalCompileDuration / 16e9)
+ diffToBuildTime.append((buildTime - totalParseDuration - totalHashDuration - totalCompileDuration) / 16e9)
- # changes graph
- differentAstHashes.append(differentAstHash)
- differentObjHashes.append(differentObjHash)
- sameHashes.append(same)
- fileCounts.append(fileCount)
+ # changes graph
+ differentAstHashes.append(differentAstHash)
+ differentObjHashes.append(differentObjHash)
+ sameHashes.append(same)
+ fileCounts.append(fileCount)
prevCommit = currentCommit
prevCommitID = commitID
--
GitLab