Expand examples/cpu_only.py

Benchmark function that sleeps for 1 seconda on the host using CPU-only
timer, as well as CPU/GPU timer that does/doesn't use blocking kernel.

All three methods must report consistent values close to 1 second.

python/examples/cpu_only.py

@@ -4,15 +4,31 @@ import time
 import cuda.nvbench as nvbench
 
 
-def throughput_bench(state: nvbench.State) -> None:
+def sleep_bench(state: nvbench.State) -> None:
     def launcher(launch: nvbench.Launch):
         time.sleep(1)
 
     state.exec(launcher)
 
 
+def sleep_bench_sync(state: nvbench.State) -> None:
+    sync = state.get_string("Sync")
+    sync_flag = sync == "Do sync"
+
+    def launcher(launch: nvbench.Launch):
+        time.sleep(1)
+
+    state.exec(launcher, sync=sync_flag)
+
+
 if __name__ == "__main__":
-    b = nvbench.register(throughput_bench)
+    # time function sleeping on the host
+    # using CPU timer only
+    b = nvbench.register(sleep_bench)
     b.set_is_cpu_only(True)
 
+    # time the same function using both CPU/GPU timers
+    b2 = nvbench.register(sleep_bench_sync)
+    b2.add_string_axis("Sync", ["Do not sync", "Do sync"])
+
     nvbench.run_all_benchmarks(sys.argv)