CPU Frequency Modeling Augmentation (#1428)

Author: 17zhangw

script/model/info/data_info.py

@@ -1,8 +1,8 @@
-from type import OpUnit, ArithmeticFeature, Target, ExecutionFeature
+from type import OpUnit, Target, ExecutionFeature
 
 # Boundary from input feature -> outputs
 # The field >= boundary is the inputs and outputs
-INPUT_OUTPUT_BOUNDARY = ExecutionFeature.EXEC_MODE
+INPUT_OUTPUT_BOUNDARY = ExecutionFeature.CPU_FREQ
 
 # End of the input features
 INPUT_END_BOUNDARY = Target.START_TIME
@@ -19,12 +19,6 @@
 # The index for different targets in the output
 TARGET_CSV_INDEX = {}
 
-# The mini model features for arithmetic operations
-ARITHMETIC_FEATURE_INDEX = {
-    ArithmeticFeature.EXEC_NUMBER: 0,
-    ArithmeticFeature.EXEC_MODE: 1,
-}
-
 # The position of the name of the operating unit in the execution engine csv metrics file
 EXECUTION_CSV_NAME_POSITION = 0
 

script/model/type.py

@@ -80,28 +80,22 @@ class ExecutionFeature(enum.IntEnum):
     FEATURES = 3,
 
     # input features
-    EXEC_MODE = 4,
-    NUM_ROWS = 5,
-    KEY_SIZES = 6,
-    NUM_KEYS = 7,
-    EST_CARDINALITIES = 8,
-    MEM_FACTOR = 9,
-    NUM_LOOPS = 10,
-    NUM_CONCURRENT = 11,
+    CPU_FREQ = 4,
+    EXEC_MODE = 5,
+    NUM_ROWS = 6,
+    KEY_SIZES = 7,
+    NUM_KEYS = 8,
+    EST_CARDINALITIES = 9,
+    MEM_FACTOR = 10,
+    NUM_LOOPS = 11,
+    NUM_CONCURRENT = 12,
 
     # interval input features
-    TXNS_DEALLOCATED = 12,
-    TXNS_UNLINKED = 13,
-    BUFFER_UNLINKED = 14,
-    READONLY_UNLINKED = 15,
-    INTERVAL = 16,
-
-
-class ArithmeticFeature(enum.Enum):
-    """The input fields of the arithmetic operating units
-    """
-    EXEC_NUMBER = 0,
-    EXEC_MODE = 1,
+    TXNS_DEALLOCATED = 13,
+    TXNS_UNLINKED = 14,
+    BUFFER_UNLINKED = 15,
+    READONLY_UNLINKED = 16,
+    INTERVAL = 17,
 
 
 class ConcurrentCountingMode(enum.Enum):

src/include/execution/util/cpu_info.h

@@ -88,6 +88,11 @@ y   * @return The size of a cache line at level @em level.
    */
   uint64_t GetRefCyclesUs() const { return ref_cycles_us_; }
 
+  /**
+   * @return The CPU Frequency in MHz recorded at initialization
+   */
+  double GetCpuFreq() const { return cpu_mhz_; }
+
   /**
    * @return True if the CPU has the input hardware feature @em feature; false otherwise;
    */

src/include/metrics/metrics_util.h

@@ -2,8 +2,19 @@
 
 #include <chrono>  // NOLINT
 
+#include "execution/util/cpu_info.h"
+
 namespace noisepage::metrics {
 
+/**
+ * Struct used to represent the hardware context that we are interested in capturing.
+ * We currently capture only the CPU frequency.
+ */
+struct HardwareContext {
+  /** CPU Frequency (MHz) */
+  double cpu_mhz_;
+};
+
 /**
  * Static utility methods for the metrics component
  */
@@ -19,5 +30,10 @@ struct MetricsUtil {
                std::chrono::high_resolution_clock::now().time_since_epoch())
         .count();
   }
+
+  /**
+   * @return The hardware context to record
+   */
+  static HardwareContext GetHardwareContext() { return HardwareContext{execution::CpuInfo::Instance()->GetCpuFreq()}; }
 };
 }  // namespace noisepage::metrics

src/include/metrics/pipeline_metric.h

@@ -50,12 +50,14 @@ class PipelineMetricRawData : public AbstractRawData {
                      "Not all files are open.");
 
     auto &outfile = (*outfiles)[0];
+    auto context = MetricsUtil::GetHardwareContext();
 
     for (auto &data : pipeline_data_) {
       outfile << data.query_id_.UnderlyingValue() << ", ";
       outfile << data.pipeline_id_.UnderlyingValue() << ", ";
       outfile << data.features_.size() << ", ";
       outfile << data.GetFeatureVectorString() << ", ";
+      outfile << context.cpu_mhz_ << ", ";
       outfile << static_cast<uint32_t>(data.execution_mode_) << ", ";
       outfile << data.GetEstRowsVectorString() << ", ";
       outfile << data.GetKeySizeVectorString() << ", ";
@@ -81,7 +83,7 @@ class PipelineMetricRawData : public AbstractRawData {
    * Note: This includes the columns for the input feature, but not the output (resource counters)
    */
   static constexpr std::array<std::string_view, 1> FEATURE_COLUMNS = {
-      "query_id, pipeline_id, num_features, features, exec_mode, num_rows, key_sizes, num_keys, "
+      "query_id, pipeline_id, num_features, features, cpu_freq, exec_mode, num_rows, key_sizes, num_keys, "
       "est_cardinalities, mem_factor, num_loops, num_concurrent"};
 
  private:

test/self_driving/modeling_test.cpp

@@ -69,10 +69,10 @@ TEST_F(ModelServerTest, PipelineTest) {
   }
 
   std::vector<std::vector<double>> features{
-      {0, 10000, 4, 1, 10000, 1, 0, 0},
-      {0, 10000, 4, 1, 10000, 1, 0, 0},
-      {0, 10000, 4, 1, 10000, 1, 0, 0},
-      {0, 10000, 4, 1, 10000, 1, 0, 0},
+      {0, 0, 10000, 4, 1, 10000, 1, 0, 0},
+      {0, 0, 10000, 4, 1, 10000, 1, 0, 0},
+      {0, 0, 10000, 4, 1, 10000, 1, 0, 0},
+      {0, 0, 10000, 4, 1, 10000, 1, 0, 0},
   };
 
   // Send a message