share iterations among threads

rocksdb/main.cpp

@@ -1,10 +1,10 @@
-#include <algorithm>
 #include <atomic>
 #include <cstdint>
 #include <cstdio>
+#include <cstdlib>
+#include <exception>
 #include <fmt/base.h>
 #include <fmt/format.h>
-#include <iostream>
 #include <memory>
 #include <random>
 #include <rocksdb/cache.h>
@@ -21,7 +21,6 @@
 #include <filesystem>
 #include <format>
 #include <string>
-#include <syncstream>
 
 #include "CLI/CLI.hpp"
 #include "instant.h"
@@ -36,8 +35,8 @@ struct Args {
     size_t iterations;
     size_t key_size;
     size_t value_size;
-    size_t insert_ratio;
     size_t blob_size;
+    size_t insert_ratio;
     bool random;
     std::string mode;
     std::string path;
@@ -50,6 +49,7 @@ int main(int argc, char *argv[]) {
             .iterations = 100000,
             .key_size = 16,
             .value_size = 1024,
+            .blob_size = 8192,
             .insert_ratio = 30,
             .mode = "insert",
             .path = "/tmp/rocksdb_tmp",
@@ -148,22 +148,21 @@ int main(int argc, char *argv[]) {
     std::mt19937 gen(rd());
     std::uniform_int_distribution<int> dist(0, 100);
 
-
     std::string val(args.value_size, 'x');
+    auto keys_per_thread = args.iterations / args.threads;
     for (size_t tid = 0; tid < args.threads; ++tid) {
         std::vector<std::string> key{};
-        for (size_t i = 0; i < args.iterations; ++i) {
+        for (size_t i = 0; i < keys_per_thread; ++i) {
             auto tmp = std::format("key_{}_{}", tid, i);
             tmp.resize(args.key_size, 'x');
             key.emplace_back(std::move(tmp));
         }
-        if (args.mode == "get" || args.random || args.mode == "scan") {
+        if (args.mode == "get" || args.random) {
             std::shuffle(keys.begin(), keys.end(), gen);
         }
         keys.emplace_back(std::move(key));
     }
 
-
     auto *handle = handles[0];
 
     if (args.mode == "get" || args.mode == "scan") {
@@ -176,17 +175,28 @@ int main(int argc, char *argv[]) {
         }
         kv->Commit();
         delete kv;
-        if (args.mode == "get") {
-            delete handle;
-            delete db;
-            handles.clear();
-            // re-open db
-            s = rocksdb::OptimisticTransactionDB::Open(options, args.path, cfd, &handles, &db);
-            assert(s.ok());
+        delete handle;
+        delete db;
+        handles.clear();
+        // re-open db
+        s = rocksdb::OptimisticTransactionDB::Open(options, args.path, cfd, &handles, &db);
+        assert(s.ok());
+
+        handle = handles[0];
+
+        // simulate common use cases
+        std::uniform_int_distribution<int> dist(0, args.threads - 1);
+        for (size_t i = 0; i < keys_per_thread; ++i) {
+            auto tid = dist(gen);
+            auto k = std::format("key_{}_{}", tid, i);
+            k.resize(args.key_size, 'x');
+            auto s = db->Get(rocksdb::ReadOptions(), k, &val);
+            if (!s.ok()) {
+                std::terminate();
+            }
         }
     }
 
-    handle = handles[0];
     auto *snapshot = db->GetSnapshot();
     for (size_t tid = 0; tid < args.threads; ++tid) {
         wg.emplace_back([&, tid] {

scripts/mace.sh

@@ -16,7 +16,7 @@ function samples() {
         # set -x
         db_root=$1
 
-        cnt=10000
+        cnt=100000
         for ((i = 1; i <= $(nproc); i *= 2))
         do
                 for ((j = 0; j < ${#kv_sz[@]}; j += 2))

scripts/rocksdb.sh

@@ -15,7 +15,7 @@ function samples() {
         kv_sz=(16 16 100 1024 1024 1024 16 10240)
         # set -x
         db_root=$1
-        cnt=10000
+        cnt=100000
         for ((i = 1; i <= $(nproc); i *= 2))
         do
                 for ((j = 0; j < ${#kv_sz[@]}; j += 2))

src/main.rs

@@ -81,7 +81,7 @@ fn main() {
     opt.sync_on_write = false;
     opt.over_provision = true; // large value will use lots of memeory
     opt.inline_size = args.blob_size;
-    opt.tmp_store = args.mode != "get";
+    opt.tmp_store = args.mode != "get" && args.mode != "scan";
     let mut saved = opt.clone();
     saved.tmp_store = false;
     let mut db = Mace::new(opt.validate().unwrap()).unwrap();
@@ -89,14 +89,15 @@ fn main() {
 
     let mut rng = rand::rng();
     let value = Arc::new(vec![b'0'; args.value_size]);
+    let keys_per_thread = args.iterations / args.threads;
     for tid in 0..args.threads {
-        let mut tk = Vec::with_capacity(args.iterations);
-        for i in 0..args.iterations {
+        let mut tk = Vec::with_capacity(keys_per_thread);
+        for i in 0..keys_per_thread {
             let mut key = format!("key_{tid}_{i}").into_bytes();
             key.resize(args.key_size, b'x');
             tk.push(key);
         }
-        if args.random || args.mode == "get" || args.mode == "scan" {
+        if args.random || args.mode == "get" {
             tk.shuffle(&mut rng);
         }
         keys.push(tk);
@@ -110,11 +111,18 @@ fn main() {
             }
         });
         pre_tx.commit().unwrap();
-        if args.mode == "get" {
-            drop(db);
-            // re-open db
-            saved.tmp_store = true;
-            db = Mace::new(saved.validate().unwrap()).unwrap();
+        drop(db);
+        // re-open db
+        saved.tmp_store = true;
+        db = Mace::new(saved.validate().unwrap()).unwrap();
+
+        // simulate common use cases
+        for i in 0..keys_per_thread {
+            let tid = rng.random_range(0..args.threads);
+            let mut k = format!("key_{tid}_{i}").into_bytes();
+            k.resize(args.key_size, b'x');
+            let view = db.view().unwrap();
+            view.get(&k).unwrap();
         }
     }
 
@@ -182,10 +190,7 @@ fn main() {
                         let iter = view.seek(prefix);
                         for x in iter {
                             round += 1;
-                            let k = x.key();
-                            let v = x.val();
-                            std::hint::black_box(k);
-                            std::hint::black_box(v);
+                            std::hint::black_box(x);
                         }
                     }
                     _ => panic!("Invalid mode"),
@@ -205,16 +210,6 @@ fn main() {
     let total = total_ops.load(std::sync::atomic::Ordering::Relaxed);
     let ops = (total as f64 / duration.as_secs_f64()) as usize;
 
-    // println!("{:<20} {}", "Test Mode:", args.mode);
-    // println!("{:<20} {}", "Threads:", args.threads);
-    // println!("{:<20} {}", "Iterations", args.iterations);
-    // println!("{:<20} {}B", "Key Size:", args.key_size);
-    // println!("{:<20} {}B", "Value Size:", args.value_size);
-    // println!("{:<20} {ops}", "Total Ops:");
-    // if args.mode == "mixed" {
-    //     println!("{:<20} {}%", "Insert Ratio:", args.insert_ratio);
-    // }
-
     let ratio = if args.mode == "mixed" {
         args.insert_ratio
     } else if args.mode == "insert" {
@@ -222,7 +217,6 @@ fn main() {
     } else {
         0
     };
-    // eprintln!("mode,threads,key_size,value_size,insert_ratio,ops");
     eprintln!(
         "{},{},{},{},{},{},{}",
         args.mode,