lmbench - system benchmarks
lmbench is a series of micro benchmarks intended to measure basic
operating system and hardware system metrics. The benchmarks fall into
three general classes: bandwidth, latency, and ‘‘other’’.
Most of the lmbench benchmarks use a standard timing harness described
in timing(3) and have a few standard options: parallelism, warmup, and
repetitions. Parallelism specifies the number of benchmark processes
to run in parallel. This is primarily useful when measuring the
performance of SMP or distributed computers and can be used to evaluate
the system’s performance scalability. Warmup is the number of minimum
number of microseconds the benchmark should execute the benchmarked
capability before it begins measuring performance. Again this is
primarily useful for SMP or distributed systems and it is intended to
give the process scheduler time to "settle" and migrate processes to
other processors. By measuring performance over various warmup
periods, users may evaulate the scheduler’s responsiveness.
Repetitions is the number of measurements that the benchmark should
take. This allows lmbench to provide greater or lesser statistical
strength to the results it reports. The default number of repetitions
Data movement is fundamental to the performance on most computer
systems. The bandwidth measurements are intended to show how the
system can move data. The results of the bandwidth metrics can be
compared but care must be taken to understand what it is that is being
compared. The bandwidth benchmarks can be reduced to two main
components: operating system overhead and memory speeds. The bandwidth
benchmarks report their results as megabytes moved per second but
please note that the data moved is not necessarily the same as the
memory bandwidth used to move the data. Consult the individual man
pages for more information.
Each of the bandwidth benchmarks is listed below with a brief overview
of the intent of the benchmark.
bw_file_rd reading and summing of a file via the read(2) interface.
bw_mem_cp memory copy.
bw_mem_rd memory reading and summing.
bw_mem_wr memory writing.
bw_mmap_rd reading and summing of a file via the memory mapping
bw_pipe reading of data via a pipe.
bw_tcp reading of data via a TCP/IP socket.
bw_unix reading data from a UNIX socket.
Control messages are also fundamental to the performance on most
computer systems. The latency measurements are intended to show how
fast a system can be told to do some operation. The results of the
latency metrics can be compared to each other for the most part. In
particular, the pipe, rpc, tcp, and udp transactions are all identical
benchmarks carried out over different system abstractions.
Latency numbers here should mostly be in microseconds per operation.
lat_connect the time it takes to establish a TCP/IP connection.
lat_ctx context switching; the number and size of processes is
lat_fcntl fcntl file locking.
lat_fifo ‘‘hot potato’’ transaction through a UNIX FIFO.
lat_fs creating and deleting small files.
lat_pagefault the time it takes to fault in a page from a file.
lat_mem_rd memory read latency (accurate to the ~2-5 nanosecond
range, reported in nanoseconds).
lat_mmap time to set up a memory mapping.
lat_ops basic processor operations, such as integer XOR, ADD,
SUB, MUL, DIV, and MOD, and float ADD, MUL, DIV, and
double ADD, MUL, DIV.
lat_pipe ‘‘hot potato’’ transaction through a Unix pipe.
lat_proc process creation times (various sorts).
lat_rpc ‘‘hot potato’’ transaction through Sun RPC over UDP or
lat_select select latency
lat_sig signal installation and catch latencies. Also protection
fault signal latency.
lat_syscall non trivial entry into the system.
lat_tcp ‘‘hot potato’’ transaction through TCP.
lat_udp ‘‘hot potato’’ transaction through UDP.
lat_unix ‘‘hot potato’’ transaction through UNIX sockets.
the time it takes to establish a UNIX socket connection.
mhz processor cycle time
tlb TLB size and TLB miss latency
line cache line size (in bytes)
cache cache statistics, such as line size, cache sizes, memory
stream John McCalpin’s stream benchmark
par_mem memory subsystem parallelism. How many requests can the
memory subsystem service in parallel, which may depend on
the location of the data in the memory hierarchy.
par_ops basic processor operation parallelism.
bargraph(1), graph(1), lmbench(3), results(3), timing(3),
bw_file_rd(8), bw_mem_cp(8), bw_mem_wr(8), bw_mmap_rd(8), bw_pipe(8),
bw_tcp(8), bw_unix(8), lat_connect(8), lat_ctx(8), lat_fcntl(8),
lat_fifo(8), lat_fs(8), lat_http(8), lat_mem_rd(8), lat_mmap(8),
lat_ops(8), lat_pagefault(8), lat_pipe(8), lat_proc(8), lat_rpc(8),
lat_select(8), lat_sig(8), lat_syscall(8), lat_tcp(8), lat_udp(8),
lmdd(8), par_ops(8), par_mem(8), mhz(8), tlb(8), line(8), cache(8),
Funding for the development of these tools was provided by Sun
Microsystems Computer Corporation.
A large number of people have contributed to the testing and
development of lmbench.
The benchmarking code is distributed under the GPL with additional
restrictions, see the COPYING file.
Carl Staelin and Larry McVoy
Comments, suggestions, and bug reports are always welcome.
(c)1994-2000 Larry McVoy and Carl St$Date$