baseline

mars-research · Mar 15, 2024 · 42672ba · 42672ba
1 parent 729eee5
commit 42672ba
Show file tree

Hide file tree

Showing 5 changed files with 282 additions and 230 deletions.
diff --git a/include/tests/KmerTest.hpp b/include/tests/KmerTest.hpp
@@ -25,6 +25,10 @@ class KmerTest {
     void count_kmer_radix_partition_global(Shard* sh, const Configuration& config,
                             std::barrier<VoidFn>* barrier,
                             RadixContext* context, BaseHashTable* ht);
+
+    void count_kmer_baseline(Shard* sh, const Configuration& config,
+                                std::barrier<VoidFn>* barrier,
+                                BaseHashTable* ht);
 };
 
 }  // namespace kmercounter

diff --git a/include/types.hpp b/include/types.hpp
@@ -83,6 +83,7 @@ typedef enum {
   RW_RATIO = 12,
   HASHJOIN = 13,
   FASTQ_WITH_INSERT_RADIX = 14,
+  BASELINE = 15,
 } run_mode_t;
 
 // XXX: If you add/modify a mode, update the `ht_type_strings` in
@@ -355,6 +356,10 @@ struct Configuration {
   // controls distribution of threads across numa nodes
   uint32_t numa_split;
 
+  //baseline
+  uint32_t data_size;
+  uint32_t workload_size;
+
   // hashtable configuration
   // different hashtable types
   uint32_t ht_type;

diff --git a/jerry-benchmark.sh b/jerry-benchmark.sh
@@ -1,7 +1,7 @@
 #!/bin/bash
 
 GENOME="fvesca"
-LOG_DIR=./log
+LOG_DIR=.
 DATASET_DIR=/opt/dramhit/kmer_dataset
 declare -A DATASET_ARRAY
 DATASET_ARRAY["dmela"]=${DATASET_DIR}/ERR4846928.fastq
@@ -19,10 +19,35 @@ function command_regular()
     --mode 4 \
     --ht-type 3 \
     --numa-split 1 \
-    --num-threads $2 \
+    --num-threads 64 \
     --ht-size $HT_SIZE \
     --in-file $FASTA_FILE \
-    --k $1 > $LOG_DIR/kmer_k${1}.log
+    --k 10 > $LOG_DIR/kmer_k${1}.log
+}
+
+function command_baseline() 
+{
+    sudo ./build/dramhit \
+    --mode 15 \
+    --ht-type 3 \
+    --numa-split 1 \
+    --num-threads 1 \
+    --ht-size $1 \
+    --datasize $2 \
+    --workload $3 > logs/baseline_d${2}_ht${1}.log
+}
+
+
+function bench_baseline() 
+{
+    #ht=300
+    #d=100
+    w=100000
+    for (( i=5; i<=25; i+=1 )); do
+        d=$((1<<i))
+        ht=$((1*d))
+        command_baseline $ht $d $w
+    done    
 }
 
 function command_radix() 
@@ -31,7 +56,6 @@ function command_radix()
     --mode 14 \
     --ht-type 3 \
     --numa-split 1 \
-    --hw-pref on \
     --num-threads $3 \
     --ht-size ${HT_SIZE} \
     --in-file ${FASTA_FILE} \
@@ -41,46 +65,24 @@ function command_radix()
 function bench_radix()
 {
     echo ">>> Input File: ${FASTA_FILE} Hashtable Size: ${HT_SIZE}"
-    t=64 # thread number
+    t=1 # thread number
     k=15
-    for d in $(seq 6 6); do
+    for d in $(seq 10 10); do
         echo "-> Starting to run with d ${d} k ${k}"
         command_radix $k $d $t
     done
 }
 
 function debug() 
 {
-    sudo gdb --args \ 
-    ./build/dramhit \ 
-    "--mode" "14" \ 
-    "--k" "6" \
-    "--d" "2" \
-    "--ht-type" "3" \ 
-    "--numa-split" "1" \ 
-    "--num-threads" "1" \
-     "--ht-size" "100" \
-    "--in-file" $DATASET_ARRAY[$ACTIVE_DATASET]    
+    sudo gdb --args ./build/dramhit --mode 15 --ht-type 3 --numa-split 1 --num-threads 1 --ht-size 300 --datasize 100 --workload 10000 
 }
 
-# perf report
-function perf() 
-{
-    sudo perf record \
-    -e cache-misses, cycles, faults \ 
-    ./build/dramhit \
-    --mode 14 \
-    --ht-type 3 \
-    --numa-split 1 \
-    --num-threads 64 \
-    --ht-size $HT_SIZE \
-    --in-file $FASTA_FILE \
-    --k 8 --d 8    
-}
 
 function bench() {
-    bench_radix
-    #bench_regular
+    #bench_radix
+    #command_regular
+    bench_baseline
 }
 
 function build() {

diff --git a/src/Application.cpp b/src/Application.cpp
@@ -174,8 +174,7 @@ void Application::shard_thread(int tid,
 
   switch (config.mode) {
     case FASTQ_WITH_INSERT:
-      kmer_ht = init_ht(config.ht_size, sh->shard_idx);
-      break;
+    case BASELINE:
     case FASTQ_WITH_INSERT_RADIX:
       kmer_ht = init_ht(config.ht_size, sh->shard_idx);
       break;
@@ -242,6 +241,9 @@ void Application::shard_thread(int tid,
     case FASTQ_WITH_INSERT:
       this->test.kmer.count_kmer(sh, config, kmer_ht, barrier);
       break;
+    case BASELINE:
+      this->test.kmer.count_kmer_baseline(sh, config, barrier, kmer_ht);
+      break;
     default:
       break;
   }
@@ -258,7 +260,7 @@ void Application::shard_thread(int tid,
   //   kmer_ht->print_to_file(outfile);
   // }
 
-  free_ht(kmer_ht);
+  // free_ht(kmer_ht);
 
 done:
   --num_entered;
@@ -278,7 +280,7 @@ int Application::spawn_shard_threads() {
 
   if ((config.mode != SYNTH) && (config.mode != ZIPFIAN) &&
       (config.mode != PREFETCH) && (config.mode != CACHE_MISS) &&
-      (config.mode != RW_RATIO) && (config.mode != HASHJOIN)) {
+      (config.mode != RW_RATIO) && (config.mode != HASHJOIN) && (config.mode != BASELINE)) {
     config.in_file_sz = get_file_size(config.in_file.c_str());
     PLOG_INFO.printf("File size: %" PRIu64 " bytes", config.in_file_sz);
     seg_sz = config.in_file_sz / config.num_threads;
@@ -400,7 +402,8 @@ int Application::process(int argc, char *argv[]) {
         "11: Zipfian non-bqueue test\n"
         "12: RW-ratio test\n"
         "13: Hashjoin\n"
-        "14: Fastq insert with Radix Parition")("base",
+        "14: Fastq insert with Radix Parition\n"
+        "15: BaseLine")("base",
                         po::value<uint64_t>(&config.kmer_create_data_base)
                             ->default_value(def.kmer_create_data_base),
                         "Number of base K-mers")(
@@ -503,7 +506,15 @@ int Application::process(int argc, char *argv[]) {
         "Enable R/W tests for queues tests")(
         "pollute-ratio",
         po::value(&config.pollute_ratio)->default_value(def.pollute_ratio),
-        "Ratio of pollution events to ops (>1)");
+        "Ratio of pollution events to ops (>1)")
+        (
+        "workload",
+        po::value<uint32_t>(&config.workload_size)->default_value(10000),
+        "Baseline workload size")
+        (
+        "datasize",
+        po::value<uint32_t>(&config.data_size)->default_value(100),
+        "Baseline data size");
 
     papi_init();
 
@@ -586,6 +597,8 @@ int Application::process(int argc, char *argv[]) {
         // config.ht_fill;
       }
       PLOGI.printf("Setting ht size to %llu for hashjoin test", config.ht_size);
+    }else if(config.mode == BASELINE)
+    {
     }
 
     switch (config.ht_type) {
@@ -641,7 +654,7 @@ int Application::process(int argc, char *argv[]) {
     bq_load = BQUEUE_LOAD::HtInsert;
   }
 
-  if ((config.mode == BQ_TESTS_YES_BQ) ||
+  if ((config.mode == BASELINE) ||(config.mode == BQ_TESTS_YES_BQ) ||
       ((config.mode == FASTQ_WITH_INSERT ||
         (config.mode == FASTQ_WITH_INSERT_RADIX)) &&
        (config.ht_type == PARTITIONED_HT))) {