Add some more 'hot functions'

section_ordering.txt

-.text.intLtInt
 .text.div_i64_i64
 .text.boxInt
+.text.boxBool
 .text._ZN6pyston2gc10SmallArena6_allocEmi
 .text._ZNSt17_Function_handlerIFvPN6pyston2gc10SmallArena16ThreadBlockCacheEEZNS2_12freeUnmarkedERSt6vectorIPNS0_3BoxESaIS8_EEE3$_0E9_M_invokeERKSt9_Any_dataS4_
+.text.memset
+.text.gc_alloc
+.text.gc_realloc
+.text._ZN6pyston9BoxedList6ensureEi
+.text.PyNumber_AsSsize_t
+.text._ZN6pyston15objectNewNoArgsEPNS_10BoxedClassE
+.text._PyIndex_Check
+.text._ZN6pyston9threading21allowGLReadPreemptionEv
+.text._ZN6pyston9getOpNameEi
+.text._ZN6pyston8callFuncEPNS_17BoxedFunctionBaseEPNS_15CallRewriteArgsENS_11ArgPassSpecEPNS_3BoxES6_S6_PS6_PKSt6vectorIPKSsSaISA_EE
+.text._ZN6pyston2gc9GCVisitor5visitEPv
+.text.intLtInt
+.text.intModInt
+.text.intAddInt
+.text.intMulInt

tools/annotate.py

@@ -106,6 +106,7 @@ This will typically look like:
 Benchmark that was run.  '--heap-map-target BENCHMARK' is
 equivalent to '--heap-map-args ./pyston_release -i BENCHMARK'.
     """.strip())
+    parser.add_argument("--perf-data", default="perf.data")
     args = parser.parse_args()
 
     func = args.func_name
@@ -116,7 +117,7 @@ equivalent to '--heap-map-args ./pyston_release -i BENCHMARK'.
 
     objdump = get_objdump(func)
 
-    p = subprocess.Popen(["perf", "annotate", "-v", func], stdout=subprocess.PIPE, stderr=open("/dev/null", "w"))
+    p = subprocess.Popen(["perf", "annotate", "-i", args.perf_data, "-v", func], stdout=subprocess.PIPE, stderr=open("/dev/null", "w"))
     annotate = p.communicate()[0]
     assert p.wait() == 0
 

tools/perf_diff.py

@@ -5,7 +5,7 @@ def tally(fn):
 
     for l in open(fn):
         samples = int(l.split()[3])
-        func = l.split()[-1]
+        func = l.rsplit(']', 1)[1].strip()
         counts[func] = counts.get(func, 0) + samples
 
     return counts
@@ -23,12 +23,11 @@ def main():
 
     diff_thresh = (total1 + total2) / 2 / 100
 
-    names.sort(key=lambda n: counts1.get(n, 0) + counts2.get(n, 0), reverse=True)
-    for n in names:
+    names.sort(key=lambda n: abs(counts1.get(n, 0) - counts2.get(n, 0)), reverse=True)
+    for n in names[:10]:
         c1 = counts1.get(n, 0)
         c2 = counts2.get(n, 0)
-        if abs(c1 - c2) >= diff_thresh:
-            print n, counts1.get(n, 0), counts2.get(n, 0)
+        print n, counts1.get(n, 0), counts2.get(n, 0)
 
 if __name__ == "__main__":
     main()