When pipeline is used -P numreq (where numreq > 1), the number of generated requests may be actually higher than configured with -n requests and printed in benchmark report.
$ ./redis-benchmark -t set -n 100 -P 1
====== SET ======
100 requests completed in 0.00 seconds
50 parallel clients
3 bytes payload
keep alive: 1
100.00% <= 0 milliseconds
50000.00 requests per second
$ ./redis-cli info stats
# Stats
total_connections_received:52
total_commands_processed:101
instantaneous_ops_per_sec:62
rejected_connections:0
sync_full:0
sync_partial_ok:0
sync_partial_err:0
expired_keys:0
evicted_keys:0
keyspace_hits:0
keyspace_misses:0
pubsub_channels:0
pubsub_patterns:0
latest_fork_usec:0
migrate_cached_sockets:0
Same test with pipeline enabled (after restarting the server):
$ ./redis-benchmark -t set -n 100 -P 10
====== SET ======
100 requests completed in 0.00 seconds
50 parallel clients
3 bytes payload
keep alive: 1
100.00% <= 0 milliseconds
inf requests per second
$ ./redis-cli info stats
# Stats
total_connections_received:52
total_commands_processed:501
instantaneous_ops_per_sec:312
rejected_connections:0
sync_full:0
sync_partial_ok:0
sync_partial_err:0
expired_keys:0
evicted_keys:0
keyspace_hits:0
keyspace_misses:0
pubsub_channels:0
pubsub_patterns:0
latest_fork_usec:0
migrate_cached_sockets:0
Comment From: filipecosta90
@krzycz closing this one given #7947 addresses it.