Chm: Add auto-scaled benchmark and set num_threads to 64

benchpress/config/jobs_ai.yml

@@ -172,7 +172,7 @@
 
   vars:
     - 'distribution_file=benchmarks/chm/model_a.dist'
-    - 'num_threads=80'
+    - 'num_threads=64'
     - 'duration_seconds=360'
     - 'batch_size=10000000'
     - 'num_batch_threads=4'
@@ -190,11 +190,46 @@
 
   vars:
     - 'distribution_file=benchmarks/chm/model_b.dist'
-    - 'num_threads=80'
+    - 'num_threads=64'
     - 'duration_seconds=360'
     - 'batch_size=10000000'
     - 'num_batch_threads=4'
 
+- benchmark: chm
+  name: chm_autoscale_a
+  description: Concurrent hash map benchmark for Model A.
+  args:
+    - '--distribution_file={distribution_file}'
+    - '--num_threads_for_one_socket={num_threads_for_one_socket}'
+    - '--duration_seconds={duration_seconds}'
+    - '--batch_size={batch_size}'
+    - '--num_batch_threads={num_batch_threads}'
+
+  vars:
+    - 'distribution_file=benchmarks/chm/model_a.dist'
+    - 'duration_seconds=360'
+    - 'batch_size=10000000'
+    - 'num_batch_threads=4'
+    - 'num_threads_for_one_socket=true'
+
+
+- benchmark: chm
+  name: chm_autoscale_b
+  description: Concurrent hash map benchmark for Model B.
+  args:
+    - '--distribution_file={distribution_file}'
+    - '--num_threads_for_one_socket={num_threads_for_one_socket}'
+    - '--duration_seconds={duration_seconds}'
+    - '--batch_size={batch_size}'
+    - '--num_batch_threads={num_batch_threads}'
+
+  vars:
+    - 'distribution_file=benchmarks/chm/model_b.dist'
+    - 'duration_seconds=360'
+    - 'batch_size=10000000'
+    - 'num_batch_threads=4'
+    - 'num_threads_for_one_socket=true'
+
 
 - benchmark: deser
   name: deser_a

packages/chm/ChmBenchmark.cpp

@@ -36,6 +36,8 @@
 // Command line flags
 DEFINE_string(distribution_file, "", "Path to the distribution CSV file");
 DEFINE_int32(num_threads, 4, "Number of worker threads per batch");
+// If num_threads_for_one_socket is true then it will override num_threads
+DEFINE_bool(num_threads_for_one_socket, false, "Use number of threads available in socket 0");
 DEFINE_int32(num_batch_threads, 2, "Number of parallel batch threads");
 DEFINE_int32(duration_seconds, 10, "Benchmark duration in seconds");
 DEFINE_int32(initial_capacity, 0, "Initial hash map capacity hint");
@@ -622,6 +624,34 @@ class ChmBenchmark {
 
 } // namespace chm_benchmark
 
+/**
+ * Get the cpu core count on single socket
+ */
+int get_cpu_core_count_for_one_socket() {
+    FILE* pipe = popen("lscpu -p=CPU,SOCKET", "r");
+    if (!pipe) {
+        std::cerr << "Failed to run lscpu\n";
+        return -1;
+    }
+
+    char line[128];
+    int cpu, socket;
+    int target_socket = 0; // Assume socket 0
+    int count = 0;
+    while (fgets(line, sizeof(line), pipe)) {
+        if (line[0] == '#') continue;
+        if (sscanf(line, "%d,%d", &cpu, &socket) == 2) {
+            if (socket == target_socket) {
+                count++;
+            }
+        }
+    }
+
+    pclose(pipe);
+    return count;
+}
+
+
 /**
  * Main entry point for the ConcurrentHashMap benchmark
  * Parses command line arguments, loads distribution data, and executes
@@ -631,6 +661,17 @@ int main(int argc, char* argv[]) {
   // Parse command line flags
   gflags::ParseCommandLineFlags(&argc, &argv, true);
 
+  // Check autoscale
+  if(FLAGS_num_threads_for_one_socket){
+    int num_threads = get_cpu_core_count_for_one_socket();
+      if (0 >= num_threads) {
+        std::cerr << "Failed to get cpu core count!\n";
+        return 1;
+      }
+      FLAGS_num_threads = num_threads;
+      std::cout << "Autoscaled: " << FLAGS_num_threads << std::endl;
+  }
+
   if (FLAGS_distribution_file.empty()) {
     std::cerr
         << "Error: Distribution file path is required. Use --distribution_file=<path>"

packages/chm/README.md

@@ -2,20 +2,23 @@
 `chm` is a benchmark that simulates a concurrent hashmap workload representative of a widely used model. The benchmark closely mimics the implementation of a production-grade concurrent hashmap and utilizes workload distributions collected from real-world production environments.
 
 ## Installation
-To install Adsim, execute the following command:
+To install Chm, execute the following command:
 ```bash
 ./benchpress -b ai install chm_a
 ./benchpress -b ai install chm_b
 ```
 
-## Run Adsim
+## Run Chm
 ### Job - `chm_a` and `chm_b`
 `chm_a` and `chm_b` correspondingly simuate the workload for Model A and Model B.
+`chm_autoscale_a` and `chm_autoscale_b` sets num_threads to the threads of single socket.
 
 To run `chm` benchmark, please use following command
 ```bash
 ./benchpress -b ai run chm_a
-./benchpress -b ai run chm_a
+./benchpress -b ai run chm_b
+./benchpress -b ai run chm_autoscale_a
+./benchpress -b ai run chm_autoscale_b
 ```
 
 ## Reporting and Measurement