From 3a538f4b002d98864c3e33baca6768488e6d4278 Mon Sep 17 00:00:00 2001
From: Jacob Leverich <leverich@cs.stanford.edu>
Date: Wed, 26 Dec 2012 17:08:00 -0700
Subject: [PATCH] Implement a log-based histogram sampler to increase fidelity
 for small samples and maximum range for large samples.

---
 ConnectionStats.h     | 34 ++++++++++++++--
 LogHistogramSampler.h | 93 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 123 insertions(+), 4 deletions(-)
 create mode 100644 LogHistogramSampler.h

diff --git a/ConnectionStats.h b/ConnectionStats.h
index 2b90920..27dfeb6 100644
--- a/ConnectionStats.h
+++ b/ConnectionStats.h
@@ -8,8 +8,10 @@
 
 #ifdef USE_ADAPTIVE_SAMPLER
 #include "AdaptiveSampler.h"
-#else
+#elif defined(USE_HISTOGRAM_SAMPLER)
 #include "HistogramSampler.h"
+#else
+#include "LogHistogramSampler.h"
 #endif
 #include "AgentStats.h"
 #include "Operation.h"
@@ -21,8 +23,10 @@ class ConnectionStats {
  ConnectionStats(bool _sampling = true) :
 #ifdef USE_ADAPTIVE_SAMPLER
    get_sampler(100000), set_sampler(100000), op_sampler(100000),
-#else
+#elif defined(USE_HISTOGRAM_SAMPLER)
    get_sampler(10000,1), set_sampler(10000,1), op_sampler(1000,1),
+#else
+   get_sampler(200), set_sampler(200), op_sampler(100),
 #endif
    rx_bytes(0), tx_bytes(0), gets(0), sets(0),
    get_misses(0), sampling(_sampling) {}
@@ -31,10 +35,14 @@ class ConnectionStats {
   AdaptiveSampler<Operation> get_sampler;
   AdaptiveSampler<Operation> set_sampler;
   AdaptiveSampler<double> op_sampler;
-#else
+#elif defined(USE_HISTOGRAM_SAMPLER)
   HistogramSampler get_sampler;
   HistogramSampler set_sampler;
   HistogramSampler op_sampler;
+#else
+  LogHistogramSampler get_sampler;
+  LogHistogramSampler set_sampler;
+  LogHistogramSampler op_sampler;
 #endif
 
   uint64_t rx_bytes, tx_bytes;
@@ -160,7 +168,7 @@ class ConnectionStats {
            );
     if (newline) printf("\n");
   }
-#else
+#elif defined(USE_HISTOGRAM_SAMPLER)
   void print_stats(const char *tag, HistogramSampler &sampler,
                    bool newline = true) {
     if (sampler.total() == 0) {
@@ -170,6 +178,24 @@ class ConnectionStats {
       return;
     }
 
+    printf("%-7s %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f",
+           tag, sampler.average(),
+           sampler.get_nth(0), sampler.get_nth(1), sampler.get_nth(5),
+           sampler.get_nth(10), sampler.get_nth(90),
+           sampler.get_nth(95), sampler.get_nth(99));
+
+    if (newline) printf("\n");
+  }
+#else
+  void print_stats(const char *tag, LogHistogramSampler &sampler,
+                   bool newline = true) {
+    if (sampler.total() == 0) {
+      printf("%-7s %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f",
+             tag, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0);
+      if (newline) printf("\n");
+      return;
+    }
+
     printf("%-7s %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f %7.1f",
            tag, sampler.average(),
            sampler.get_nth(0), sampler.get_nth(1), sampler.get_nth(5),
diff --git a/LogHistogramSampler.h b/LogHistogramSampler.h
new file mode 100644
index 0000000..7f218a4
--- /dev/null
+++ b/LogHistogramSampler.h
@@ -0,0 +1,93 @@
+/* -*- c++ -*- */
+#ifndef LOGHISTOGRAMSAMPLER_H
+#define LOGHISTOGRAMSAMPLER_H
+
+#include <inttypes.h>
+
+#include <assert.h>
+#include <vector>
+
+#include "Operation.h"
+
+#define _POW 1.1
+
+class LogHistogramSampler {
+public:
+  std::vector<uint64_t> bins;
+
+  double sum;
+
+  LogHistogramSampler() = delete;
+  LogHistogramSampler(int _bins) : sum(0.0) {
+    assert(_bins > 0);
+
+    bins.resize(_bins + 1, 0);
+  }
+
+  void sample(const Operation &op) {
+    sample(op.time());
+  }
+
+  void sample(double s) {
+    assert(s >= 0);
+    size_t bin = log(s)/log(_POW);
+
+    sum += s;
+
+    //    I("%f", sum);
+
+    if ((int64_t) bin < 0) {
+      bin = 0;
+    } else if (bin >= bins.size()) {
+      bin = bins.size() - 1;
+    }
+
+    bins[bin]++;
+  }
+
+  double average() {
+    //    I("%f %d", sum, total());
+    return sum / total();
+  }
+
+  double minimum() {
+    for (size_t i = 0; i < bins.size(); i++)
+      if (bins[i] > 0) return pow(_POW, (double) i + 0.5);
+  }
+
+  double get_nth(double nth) {
+    uint64_t count = total();
+    uint64_t n = 0;
+    double target = count * nth/100;
+
+    for (size_t i = 0; i < bins.size(); i++) {
+      n += bins[i];
+
+      if (n > target) { // The nth is inside bins[i].
+        double left = target - (n - bins[i]);
+        return pow(_POW, (double) i) +
+          left / bins[i] * (pow(_POW, (double) (i+1)) - pow(_POW, (double) i));
+      }
+    }
+
+    return pow(_POW, bins.size());
+  } 
+
+  uint64_t total() {
+    uint64_t sum = 0.0;
+
+    for (auto i: bins) sum += i;
+
+    return sum;
+  }
+
+  void accumulate(const LogHistogramSampler &h) {
+    assert(bins.size() == h.bins.size());
+
+    for (size_t i = 0; i < bins.size(); i++) bins[i] += h.bins[i];
+
+    sum += h.sum;
+  }
+};
+
+#endif // LOGHISTOGRAMSAMPLER_H