implement cache in aggrj for qdp

qdp_project/src/Benchmark.cpp

@@ -40,7 +40,6 @@ void caching(size_t gid, size_t tid) {
     constexpr bool CACHE_SUBCHUNKING = SUBCHUNK_THREAD_RATIO > 1;
     constexpr bool CACHE_OVERCHUNKING = VIRT_TID_INCREMENT > 1;
 
-    if constexpr (PERFORM_CACHING) {
     if constexpr (CACHE_SUBCHUNKING) {
         constexpr size_t SUBCHUNK_COUNT = SUBCHUNK_THREAD_RATIO > 0 ? SUBCHUNK_THREAD_RATIO : 1;
         constexpr size_t SUBCHUNK_SIZE_B = CHUNK_SIZE_B / SUBCHUNK_COUNT;
@@ -74,7 +73,6 @@ void caching(size_t gid, size_t tid) {
             CACHE_.Access(reinterpret_cast<uint8_t*>(chunk_ptr), CHUNK_SIZE_B);
         }
     }
-    }
 }
 
 void scan_b(size_t gid, size_t tid) {
@@ -85,7 +83,7 @@ void scan_b(size_t gid, size_t tid) {
 
     THREAD_TIMING_[SCANB_TIMING_INDEX][UniqueIndex(gid,tid)][0][TIME_STAMP_BEGIN] = std::chrono::steady_clock::now();
 
-    if constexpr (!PERFORM_CACHING_IN_AGGREGATION) {
+    if constexpr (PERFORM_CACHING && !PERFORM_CACHING_IN_AGGREGATION) {
         caching<TC_SCANB>(gid, tid);
     }
 
@@ -135,6 +133,10 @@ void aggr_j(size_t gid, size_t tid) {
 
     LAUNCH_.wait();
 
+    if constexpr (PERFORM_CACHING && PERFORM_CACHING_IN_AGGREGATION) {
+        caching<TC_AGGRJ>(gid, tid);
+    }
+
     for (size_t i = 0; i < RUN_COUNT; i++) {
         THREAD_TIMING_[AGGRJ_TIMING_INDEX][UniqueIndex(gid,tid)][i][TIME_STAMP_BEGIN] = std::chrono::steady_clock::now();