stdlib: Add SparseHist to PyStats

This is inclusive of tests to ensure they have implemented correctly.

Change-Id: I5c84d5ffdb7b914936cfd86ca012a7b141eeaf42

src/python/m5/ext/pystats/statistic.py

@@ -253,3 +253,31 @@ class Distribution(Vector):
         # These check some basic conditions of a distribution.
         assert self.bin_size >= 0
         assert self.num_bins >= 1
+
+
+class SparseHist(Vector):
+    """A Sparse Histogram of values. A sparse histogram simply counts the "
+    frequency of each value in a sample. Ergo, it is, ineffect an disctionary
+    of values mapped to their count"""
+
+    def __init__(
+        self,
+        value: Dict[float, Scalar],
+        description: Optional[str] = None,
+    ):
+        super().__init__(
+            value=value,
+            type="SparseHist",
+            description=description,
+        )
+
+    def size(self) -> int:
+        """The number of unique sampled values."""
+        return len(self.value)
+
+    def count(self) -> int:
+        """
+        Returns the total number of samples.
+        """
+        assert self.value != None
+        return sum(self.value.values())

src/python/m5/stats/gem5stats.py

@@ -140,6 +140,8 @@ def __get_statistic(statistic: _m5.stats.Info) -> Optional[Statistic]:
         return __get_vector(statistic)
     elif isinstance(statistic, _m5.stats.Vector2dInfo):
         return __get_vector2d(statistic)
+    elif isinstance(statistic, _m5.stats.SparseHistInfo):
+        return __get_sparse_hist(statistic)
 
     return None
 
@@ -268,6 +270,24 @@ def __get_vector2d(statistic: _m5.stats.Vector2dInfo) -> Vector2d:
     return Vector2d(value=vector_rep, type="Vector2d", description=description)
 
 
+def __get_sparse_hist(statistic: _m5.stats.SparseHistInfo) -> SparseHist:
+    description = statistic.desc
+    value = statistic.values
+
+    parsed_values = {}
+    for val in value:
+        parsed_values[val] = Scalar(
+            value=value[val],
+            unit=statistic.unit,
+            datatype=StorageType["f64"],
+        )
+
+    return SparseHist(
+        value=parsed_values,
+        description=description,
+    )
+
+
 def _prepare_stats(group: _m5.stats.Group):
     """
     Prepares the statistics for dumping.

src/python/pybind11/stats.cc

@@ -78,6 +78,7 @@ cast_stat_info(const statistics::Info *info)
     TRY_CAST(statistics::VectorInfo);
     TRY_CAST(statistics::Vector2dInfo);
     TRY_CAST(statistics::DistInfo);
+    TRY_CAST(statistics::SparseHistInfo);
 
     return py::cast(info);
 
@@ -195,6 +196,15 @@ pybind_init_stats(py::module_ &m_native)
         .def_readonly("value", &statistics::Vector2dInfo::cvec)
         ;
 
+    py::class_<statistics::SparseHistInfo, statistics::Info,
+               std::unique_ptr<statistics::SparseHistInfo, py::nodelete>>(
+                    m, "SparseHistInfo")
+        .def_property_readonly("values", //A Dict[float, int] of sample & count
+            [](const statistics::SparseHistInfo &info) {
+                return info.data.cmap;
+            })
+        ;
+
     py::class_<statistics::FormulaInfo, statistics::VectorInfo,
                std::unique_ptr<statistics::FormulaInfo, py::nodelete>>(
                       m, "FormulaInfo")

src/test_objects/SConscript

@@ -32,5 +32,6 @@ if env['CONF']['USE_TEST_OBJECTS']:
         'ScalarStatTester',
         'VectorStatTester',
         'Vector2dStatTester',
+        'SparseHistStatTester',
     ])
     Source('stat_tester.cc')

src/test_objects/StatTester.py

@@ -87,3 +87,14 @@ class Vector2dStatTester(StatTester):
         [],
         "The vector stat's y subdescriptions. If empty, the subdescriptions ",
     )
+
+
+class SparseHistStatTester(StatTester):
+    type = "SparseHistStatTester"
+    cxx_header = "test_objects/stat_tester.hh"
+    cxx_class = "gem5::SparseHistStatTester"
+
+    samples = VectorParam.Float(
+        "The sparse histogram's sampled values, to be inserted into the "
+        "histogram."
+    )

src/test_objects/stat_tester.cc

@@ -28,6 +28,8 @@
 
 #include "test_objects/stat_tester.hh"
 
+#include <set>
+
 #include "base/stats/group.hh"
 
 namespace gem5
@@ -133,4 +135,28 @@ Vector2dStatTester::Vector2dStatTesterStats::Vector2dStatTesterStats(
 
 }
 
+void
+SparseHistStatTester::setStats()
+{
+    for (auto sample : params.samples) {
+        stats.sparse_histogram.sample(sample);
+    }
+}
+
+SparseHistStatTester::SparseHistStatTesterStats::SparseHistStatTesterStats(
+    statistics::Group *parent,
+    const SparseHistStatTesterParams &params
+) : statistics::Group(parent),
+    sparse_histogram(
+        this,
+        params.name.c_str(),
+        statistics::units::Count::get(),
+        params.description.c_str()
+    )
+{
+    sparse_histogram.init(
+        (std::set(params.samples.begin(), params.samples.end())).size()
+    );
+}
+
 } // namespace gem5

src/test_objects/stat_tester.hh

@@ -31,6 +31,7 @@
 
 #include "base/statistics.hh"
 #include "params/ScalarStatTester.hh"
+#include "params/SparseHistStatTester.hh"
 #include "params/StatTester.hh"
 #include "params/Vector2dStatTester.hh"
 #include "params/VectorStatTester.hh"
@@ -158,6 +159,27 @@ class Vector2dStatTester : public StatTester
         } stats;
 };
 
+class SparseHistStatTester : public StatTester
+{
+    private:
+        SparseHistStatTesterParams params;
+
+    public:
+        SparseHistStatTester(const SparseHistStatTesterParams &p) :
+            StatTester(p), params(p), stats(this, p) {}
+
+    protected:
+        void setStats() override;
+        struct SparseHistStatTesterStats : public statistics::Group
+        {
+            SparseHistStatTesterStats(
+                statistics::Group *parent,
+                const SparseHistStatTesterParams &params
+            );
+            statistics::SparseHistogram sparse_histogram;
+        } stats;
+};
+
 
 } // namespace gem5