diff --git a/benchmarks/ring_buffer_bench.cpp b/benchmarks/ring_buffer_bench.cpp
index 442268c..8283fbe 100644
--- a/benchmarks/ring_buffer_bench.cpp
+++ b/benchmarks/ring_buffer_bench.cpp
@@ -1,13 +1,64 @@
 #include <benchmark/benchmark.h>
 #include <spsc/ring_buffer.hpp>
+#include <thread>
+#include <atomic>
 
-static void BM_BufferCreation(benchmark::State& state){
-    for(auto _ : state){
-        shovy::RingBuffer<int> buffer(1024);
-        benchmark::DoNotOptimize(buffer); 
+template<typename Queue>
+static void BM_SPSC_Throughput(benchmark::State& state) {
+    const size_t capacity = state.range(0);
+    constexpr size_t total_operations = 10'000'000;
+
+    for (auto _ : state) {
+        state.PauseTiming();
+
+        Queue buffer{capacity};
+        std::atomic<bool> start{false};
+
+        std::thread producer([&]() {
+            while (!start.load(std::memory_order_acquire)) {
+
+            }
+
+            for (size_t i = 0; i < total_operations; ++i) {
+                while (!buffer.push(i)) {
+                    std::this_thread::yield();
+                }
+            }
+        });
+
+        std::thread consumer([&]() {
+            while (!start.load(std::memory_order_acquire)) {
+
+            }
+
+            size_t val = 0;
+            for (size_t i = 0; i < total_operations; ++i) {
+                while (!buffer.pop(val)) {
+                    std::this_thread::yield();
+                }
+                benchmark::DoNotOptimize(val);
+            }
+        });
+
+        state.ResumeTiming();
+
+        start.store(true, std::memory_order_release);
+
+        producer.join();
+        consumer.join();
+
+        state.PauseTiming();
     }
+
+    state.SetItemsProcessed(state.iterations() * total_operations * 2);
 }
 
-BENCHMARK(BM_BufferCreation);
+BENCHMARK_TEMPLATE(BM_SPSC_Throughput, shovy::RingBuffer<size_t>)
+        ->Arg(64)
+        ->Arg(1024)
+        ->Arg(4096)
+        ->UseRealTime()
+        ->Repetitions(5)
+        ->ReportAggregatesOnly(true);
 
 BENCHMARK_MAIN();
\ No newline at end of file
diff --git a/include/spsc/ring_buffer.hpp b/include/spsc/ring_buffer.hpp
index 4d41d7f..77bc924 100644
--- a/include/spsc/ring_buffer.hpp
+++ b/include/spsc/ring_buffer.hpp
@@ -1,6 +1,7 @@
 #pragma once
 #include <cstddef>
 #include <stdexcept>
+#include <atomic>
 
 namespace shovy{
 
@@ -21,29 +22,29 @@ class RingBuffer{
     // producer
     bool push(const T& item) { 
         if (full())  return false;
-        buffer_[write_idx % capacity_] = item;
-        write_idx++;
+        buffer_[write_idx.load() % capacity_] = item;
+        write_idx.fetch_add(1);     // write_idx++; is okay
         return true; 
     }
 
     // consumer
     bool pop(T& item) { 
         if (empty()) return false;
-        item = buffer_[read_idx % capacity_];
-        read_idx++;
+        item = buffer_[read_idx.load() % capacity_];
+        read_idx.fetch_add(1);
         return true;
     }
 
-    bool empty() const { return write_idx == read_idx; }
+    bool empty() const { return write_idx.load() == read_idx.load(); }
     bool full() const { return size() == capacity_; }
-    size_t size() const { return write_idx - read_idx; }
+    size_t size() const { return write_idx.load() - read_idx.load(); }
     size_t capacity() const { return capacity_; }
 
 private:
     T* buffer_;
     size_t capacity_;
-    size_t write_idx{0};
-    size_t read_idx{0};
+    std::atomic<size_t> write_idx{0};
+    std::atomic<size_t> read_idx{0};
 };
 
 }   // namespace shovy
\ No newline at end of file