ft-compute-stats: add histogram & CDF support

1 files changed, 84 insertions, 10 deletions

diff --git a/ft-compute-stats b/ft-compute-stats
index 45e08cf..6399df4 100755
--- a/ft-compute-stats
+++ b/ft-compute-stats
@@ -8,6 +8,8 @@ import optparse
 import sys
 import os
 
+from math import ceil
+
 from os.path import splitext
 
 import itertools as it
@@ -23,7 +25,20 @@ def decode_key_value_filename(name):
         params[k] = v
     return params
 
-def print_rows(rows, first_row_prefix='#', other_rows_prefix=' '):
+def print_cols(cols, first_row_prefix='# ', other_rows_prefix='  '):
+    col_widths = [max((len(str(x)) for x in col)) for col in cols]
+
+    prefs = it.chain([first_row_prefix], it.repeat(other_rows_prefix))
+
+    for prefix, row in it.izip(prefs, range(0, max((len(c) for c in cols)))):
+        reached_end = True
+        data = [col[row] if row < len(col) else '' for col in cols]
+        print '%s%s' % (prefix,
+                        ", ".join([' ' * (c - len(str(f))) + str(f)
+                                   for (c, f) in it.izip(col_widths, data)]))
+
+
+def print_rows(rows, first_row_prefix='# ', other_rows_prefix='  '):
     col = 0
     col_widths = []
     field_widths = True
@@ -79,10 +94,25 @@ o = optparse.make_option
 opts = [
     o('-p', '--cycles-per-usec', action='store', dest='cycles', type='float',
       help='how many cycles per usec'),
+    o(None, '--hist', action='store_true', dest='want_hist',
+      help='generate a histogram'),
+    o('-b', '--bin-size', action='store', dest='bin_size', type='float',
+      help='size of each bin in histogram'),
+    o('-n', '--normalize-counts', action='store_true', dest='normalize',
+      help='give relative frequency, not absolute sample counts'),
+    o('-c', '--cumulative', action='store_true', dest='cumulative',
+      help='report cumulative counts (i.e., CDF)'),
+    o(None, '--percent', action='store_true', dest='want_percent',
+      help='give relative frequency as a percentage'),
     ]
 
 defaults = {
-    'cycles'  : None,
+    'cycles'    : None,
+    'want_hist' : False,
+    'bin_size'  : 1000,
+    'normalize' : False,
+    'want_percent' : False,
+    'cumulative' : False,
     }
 
 options = None
@@ -135,6 +165,32 @@ def get_stats(fname):
     finfo = [fname]
     return [to_str(x) for x in  info + stats + finfo]
 
+def make_bins(max_val):
+    num_bins = int(ceil(max_val / options.bin_size))
+    return [x * options.bin_size for x in range(0, num_bins)]
+
+def hist_file(fname, scale):
+    max_val = 0
+    hist = []
+
+    size = os.stat(fname).st_size
+    if size:
+        samples = numpy.memmap(fname, dtype='float32', mode='c')
+
+        n = len(samples)
+        if n > 0:
+            samples *= scale
+            max_val = numpy.amax(samples)
+            bins = make_bins(max_val)
+            hist, _ = numpy.histogram(samples, bins)
+            if options.cumulative:
+                hist = numpy.cumsum(hist)
+            if options.normalize:
+                hist = [h/n * (100 if options.want_percent else 1) for h in hist]
+
+    return (max_val, hist)
+
+
 if __name__ == '__main__':
     # FIXME: would be nicer with argparse
     parser = optparse.OptionParser(option_list=opts)
@@ -142,13 +198,31 @@ if __name__ == '__main__':
     (options, files) = parser.parse_args()
 
     try:
-        rows = []
-        rows.append(STATS_HEADERS)
-        for f in files:
-            try:
-                rows.append(get_stats(f))
-            except IOError, msg:
-                print >> sys.stderr, msg
-        print_rows(rows)
+        if options.want_hist:
+            cols = []
+            max_val = 0
+            col_names = []
+            for i, f in enumerate(files):
+                try:
+                    (max_in_file, hist) = hist_file(f, 1)
+                    cols.append([i + 1] + [to_str(x) for x in hist])
+                    max_val = max(max_val, max_in_file)
+                    col_names.append(f)
+                except IOError, msg:
+                    print >> sys.stderr, msg
+            bins = ['Bin'] + make_bins(max_val)
+            print_cols([bins] + cols)
+            print '# Columns:'
+            for i, f in enumerate(col_names):
+                print '# (%d) %s' % (i + 1, f)
+        else:
+            rows = []
+            rows.append(STATS_HEADERS)
+            for f in files:
+                try:
+                    rows.append(get_stats(f))
+                except IOError, msg:
+                    print >> sys.stderr, msg
+            print_rows(rows)
     except KeyboardInterrupt:
         pass