[Feat] Support LHD

examples/basic_usage.py

@@ -30,4 +30,4 @@
     move_to_main_threshold=2,
 )
 req_miss_ratio, byte_miss_ratio = cache.process_trace(reader, start_req=0, max_req=1000)
-print(f"Request miss ratio: {req_miss_ratio:.4f}, Byte miss ratio: {byte_miss_ratio:.4f}")
+print(f"Request miss ratio: {req_miss_ratio:.4f}, Byte miss ratio: {byte_miss_ratio:.4f}")

libcachesim/__init__.py

@@ -19,6 +19,7 @@
 from .cache import (
     CacheBase,
     # Core algorithms
+    LHD,
     LRU,
     FIFO,
     LFU,
@@ -74,6 +75,7 @@
     # Cache base class
     "CacheBase",
     # Core cache algorithms
+    "LHD",
     "LRU",
     "FIFO",
     "LFU",

libcachesim/cache.py

@@ -6,6 +6,7 @@
     CacheObject,
     Cache,
     # Core cache algorithms
+    LHD_init,
     LRU_init,
     FIFO_init,
     LFU_init,
@@ -152,6 +153,17 @@ def _create_common_params(
 # ------------------------------------------------------------------------------------------------
 # Core cache algorithms
 # ------------------------------------------------------------------------------------------------
+class LHD(CacheBase):
+    """Least Hit Density cache (no special parameters)"""
+
+    def __init__(
+        self, cache_size: int, default_ttl: int = 86400 * 300, hashpower: int = 24, consider_obj_metadata: bool = False
+    ):
+        super().__init__(
+            _cache=LHD_init(_create_common_params(cache_size, default_ttl, hashpower, consider_obj_metadata))
+        )
+
+
 class LRU(CacheBase):
     """Least Recently Used cache (no special parameters)"""
 

src/export_cache.cpp

@@ -28,6 +28,8 @@ namespace libcachesim {
 
 namespace py = pybind11;
 
+const cache_obj_t* LHD_HIT_MARKER = reinterpret_cast<cache_obj_t *>(0x1);
+
 // Custom deleters for smart pointers
 struct CacheDeleter {
   void operator()(cache_t* ptr) const {
@@ -279,9 +281,20 @@ void export_cache(py::module& m) {
           "req"_a)
       .def(
           "find",
-          [](cache_t& self, const request_t& req, const bool update_cache) {
+          [](cache_t& self, const request_t& req, const bool update_cache) -> py::object {
             cache_obj_t* obj = self.find(&self, &req, update_cache);
-            return obj ? py::cast(obj, py::return_value_policy::reference) : py::none();
+            // Return None if obj is null (not found)
+            if (obj == nullptr) {
+                return py::none();
+            }
+            // NOTE(haocheng): For LHD only, return a dummy object for hit
+            if (obj == LHD_HIT_MARKER) {
+                cache_obj_t* dummy_obj = static_cast<cache_obj_t*>(calloc(1, sizeof(cache_obj_t)));
+                dummy_obj->obj_id = req.obj_id;
+                dummy_obj->obj_size = req.obj_size;
+                return py::cast(std::unique_ptr<cache_obj_t, CacheObjectDeleter>(dummy_obj));
+            }
+            return py::cast(obj, py::return_value_policy::reference);
           },
           "req"_a, "update_cache"_a = true)
       .def(

tests/test_cache.py

@@ -9,6 +9,7 @@
 import os
 from libcachesim import (
     # Basic algorithms
+    LHD,
     LRU,
     FIFO,
     LFU,
@@ -71,6 +72,7 @@ class TestCacheBasicFunctionality:
     @pytest.mark.parametrize(
         "cache_class",
         [
+            LHD,
             LRU,
             FIFO,
             LFU,
@@ -109,10 +111,14 @@ def test_cache_initialization(self, cache_class):
                 pytest.skip(f"Cache {cache_class.__name__} failed to initialize: {e}")
 
     @pytest.mark.parametrize(
-        "cache_class", [LRU, FIFO, LFU, ARC, Clock, Random, S3FIFO, Sieve, LIRS, TwoQ, SLRU, WTinyLFU]
+        "cache_class", [LHD, LRU, FIFO, LFU, ARC, Clock, Random, S3FIFO, Sieve, LIRS, TwoQ, SLRU, WTinyLFU]
     )
     def test_basic_get_and_insert(self, cache_class):
         """Test basic get and insert operations"""
+        if cache_class == LHD:
+            pytest.skip("LHD's insert always returns None")
+
+
         cache = cache_class(1024 * 1024)  # 1MB cache
 
         # Create a request
@@ -141,6 +147,7 @@ def test_basic_get_and_insert(self, cache_class):
     @pytest.mark.parametrize(
         "cache_class",
         [
+            LHD,
             LRU,
             FIFO,
             LFU,
@@ -192,6 +199,7 @@ def test_cache_eviction(self, cache_class):
     @pytest.mark.parametrize(
         "cache_class",
         [
+            LHD,
             LRU,
             FIFO,
             LFU,
@@ -239,6 +247,7 @@ def test_cache_find_method(self, cache_class):
     @pytest.mark.parametrize(
         "cache_class",
         [
+            LHD,
             LRU,
             FIFO,
             LFU,