Switch branches/tags
aj/revert-sunrpc aj/shared-linux-fcntl allan/config-files andros/avx512f-mem andros/libmvec andros/pr19654 archlinux/2.18/master azanella/bz12683 carlos/dlmopen carlos/master cmetcalf/tile-2.15 cvs/fedora-2_3-branch cvs/fedora-2_5-branch cvs/fedora-branch cvs/glibc-2-1-branch cvs/glibc-2-2-branch cvs/glibc-2_0_x cvs/glibc-2_3-branch cvs/glibc-2_5-branch cvs/glibc-2_6-branch cvs/glibc-2_7-branch cvs/glibc-2_8-branch cvs/glibc-2_9-branch cvs/glibc-2_10-branch cvs/master cvs/sparc-2_0_x-branch cvs/thomas-posix1996 davem/sparc dj/malloc fedora/master fedora/2.10/master fedora/2.11/master fedora/2.12/master fedora/2.13/master fedora/2.14/master fedora/2.22/master fw/extend_alloca gentoo/2.18 gentoo/2.19 gentoo/2.20 gentoo/2.21 gentoo/2.22 gentoo/2.23 hjl/avx128 hjl/avx512f-mem/master hjl/avx512f-mem/old hjl/avx512f-mem/prefetcht1 hjl/cpuid/master hjl/cpuid/2.21 hjl/cpuid/2.22 hjl/ctor/release/2.11 hjl/ctor/release/2.12 hjl/erms/i386 hjl/erms/ifunc hjl/erms/master hjl/erms/nt hjl/erms/2.22 hjl/erms/2.23 hjl/fma/master hjl/global hjl/gold hjl/i386/master hjl/i486/multiarch-old hjl/i486/multiarch hjl/implies hjl/init hjl/ld.so/master hjl/memcpy/dpdk/master hjl/mempcpy hjl/plt/master hjl/plt/2.21 hjl/plt/2.22 hjl/pr13862 hjl/pr14370 hjl/pr14562/master hjl/pr14562/2.16 hjl/pr14654 hjl/pr14716 hjl/pr14831 hjl/pr14937 hjl/pr14941 hjl/pr14955 hjl/pr14995 hjl/pr17711/2.18 hjl/pr17711/2.19 hjl/pr17711/2.20 hjl/pr17711/2.21 hjl/pr17841/master hjl/pr17841/2.21 hjl/pr18078 hjl/pr18422 hjl/pr18661 hjl/pr18696 hjl/pr18858/master hjl/pr19122 hjl/pr19178/master hjl/pr19363/clobber hjl/pr19363/master hjl/pr19363/2.22 hjl/pr19371/master hjl/pr19463 hjl/pr19583 hjl/pr19590 hjl/pr19679/2.23 hjl/pr19776/master hjl/pthread/2.21 hjl/pthread/2.22 hjl/release/2.20/master hjl/secondary/master hjl/secondary/2.21 hjl/size/master hjl/tst-plt hjl/unaligned hjl/x32/master hjl/x32/release/2.12 hjl/x32/release/2.15 hjl/x86 hjl/32bit/master hjl/32bit/2.22 ibm/master ibm/2.8/master ibm/2.10/master ibm/2.11/master ibm/2.12/master ibm/2.13/master ibm/2.16/master ibm/2.18/master ibm/2.19/master ibm/2.20/master ibm/2.22/master ldmitrie/intel_mpx linaro/master lxoliva/getaddrinfo master neleai/string-x64 neleai/strlen origin/ibm/2.8/master pasky/fixes-overdue pasky/fixes release/2.10/master release/2.11/master release/2.12/master release/2.13/master release/2.14/master release/2.15/master release/2.16/master release/2.17/master release/2.18/master release/2.19/master release/2.20/master release/2.21/master release/2.22/master release/2.23/master roland/Wshadow roland/add-on-abi-tags roland/backtrace-syms roland/cancelhandling roland/disable-nis roland/getpid roland/gold-vs-libc roland/hwcap_mask roland/manual-check roland/nacl-debug-hack roland/nacl-exit-stacks roland/nacl-port/master roland/nptl_db roland/pthread_attr_getstack roland/stat64 roland/sysconf-clocks roland/tempname roland/x86_64-crt1-cfi rsa/hwcap2_v3 rsa/hwcap2_v4 rsa/hwcap2_v5 rsa/hwcap2_v6 rsa/power8_partial rsa/power8 rsa/stdint_headers rsa/stdint_noheaders rsa/stdint rsa/2.17_backports_v2 rsa/2.17_backports_v3 rsa/2.17_backports rth/aa-memset rth/aa-opt rth/execl rth/tramp siddhesh/is_in_module siddhesh/mmap-fallback siddhesh/posix-wundef siddhesh/sem_timedwait siddhesh/tunables tuliom/multilib
Nothing to show
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
scripts
strcoll-inputs
Makefile
README
acos-inputs
acosh-inputs
asin-inputs
asinh-inputs
atan-inputs
atanh-inputs
bench-bcopy.c
bench-bzero.c
bench-malloc-thread.c
bench-math-inlines.c
bench-memccpy.c
bench-memchr.c
bench-memcmp.c
bench-memcpy-large.c
bench-memcpy.c
bench-memmem.c
bench-memmove-large.c
bench-memmove.c
bench-mempcpy.c
bench-memrchr.c
bench-memset-large.c
bench-memset.c
bench-rawmemchr.c
bench-skeleton.c
bench-stpcpy.c
bench-stpcpy_chk.c
bench-stpncpy.c
bench-strcasecmp.c
bench-strcasestr.c
bench-strcat.c
bench-strchr.c
bench-strchrnul.c
bench-strcmp.c
bench-strcoll.c
bench-strcpy.c
bench-strcpy_chk.c
bench-strcspn.c
bench-string.h
bench-strlen.c
bench-strncasecmp.c
bench-strncat.c
bench-strncmp.c
bench-strncpy.c
bench-strnlen.c
bench-strpbrk.c
bench-strrchr.c
bench-strsep.c
bench-strspn.c
bench-strstr.c
bench-strtod.c
bench-strtok.c
bench-timing-type.c
bench-timing.h
bench-util.c
bench-util.h
bench-wcpcpy.c
bench-wcpncpy.c
bench-wcscat.c
bench-wcschr.c
bench-wcschrnul.c
bench-wcscmp.c
bench-wcscpy.c
bench-wcscspn.c
bench-wcslen.c
bench-wcsncat.c
bench-wcsncmp.c
bench-wcsncpy.c
bench-wcsnlen.c
bench-wcspbrk.c
bench-wcsrchr.c
bench-wcsspn.c
bench-wmemchr.c
bench-wmemcmp.c
bench-wmemset.c
cos-inputs
cosh-inputs
exp-inputs
exp2-inputs
ffs-inputs
ffsll-inputs
json-lib.c
json-lib.h
log-inputs
log2-inputs
modf-inputs
pow-inputs
pthread_once-inputs
pthread_once-source.c
rint-inputs
sin-inputs
sincos-inputs
sinh-inputs
sprintf-inputs
sprintf-source.c
sqrt-inputs
tan-inputs
tanh-inputs

README

Using the glibc microbenchmark suite
====================================

The glibc microbenchmark suite automatically generates code for specified
functions, builds and calls them repeatedly for given inputs to give some
basic performance properties of the function.

Running the benchmark:
=====================

The benchmark needs python 2.7 or later in addition to the
dependencies required to build the GNU C Library.  One may run the
benchmark by invoking make as follows:

  $ make bench

This runs each function for 10 seconds and appends its output to
benchtests/bench.out.  To ensure that the tests are rebuilt, one could run:

  $ make bench-clean

The duration of each test can be configured setting the BENCH_DURATION variable
in the call to make.  One should run `make bench-clean' before changing
BENCH_DURATION.

  $ make BENCH_DURATION=1 bench

The benchmark suite does function call measurements using architecture-specific
high precision timing instructions whenever available.  When such support is
not available, it uses clock_gettime (CLOCK_PROCESS_CPUTIME_ID).  One can force
the benchmark to use clock_gettime by invoking make as follows:

  $ make USE_CLOCK_GETTIME=1 bench

Again, one must run `make bench-clean' before changing the measurement method.

Adding a function to benchtests:
===============================

If the name of the function is `foo', then the following procedure should allow
one to add `foo' to the bench tests:

- Append the function name to the bench variable in the Makefile.

- Make a file called `foo-inputs` to provide the definition and input for the
  function.  The file should have some directives telling the parser script
  about the function and then one input per line.  Directives are lines that
  have a special meaning for the parser and they begin with two hashes '##'.
  The following directives are recognized:

  - args: This should be assigned a colon separated list of types of the input
    arguments.  This directive may be skipped if the function does not take any
    inputs.  One may identify output arguments by nesting them in <>.  The
    generator will create variables to get outputs from the calling function.
  - ret: This should be assigned the type that the function returns.  This
    directive may be skipped if the function does not return a value.
  - includes: This should be assigned a comma-separated list of headers that
    need to be included to provide declarations for the function and types it
    may need (specifically, this includes using "#include <header>").
  - include-sources: This should be assigned a comma-separated list of source
    files that need to be included to provide definitions of global variables
    and functions (specifically, this includes using "#include "source").
    See pthread_once-inputs and pthreads_once-source.c for an example of how
    to use this to benchmark a function that needs state across several calls.
  - init: Name of an initializer function to call to initialize the benchtest.
  - name: See following section for instructions on how to use this directive.

  Lines beginning with a single hash '#' are treated as comments.  See
  pow-inputs for an example of an input file.

Multiple execution units per function:
=====================================

Some functions have distinct performance characteristics for different input
domains and it may be necessary to measure those separately.  For example, some
math functions perform computations at different levels of precision (64-bit vs
240-bit vs 768-bit) and mixing them does not give a very useful picture of the
performance of these functions.  One could separate inputs for these domains in
the same file by using the `name' directive that looks something like this:

  ##name: 240bit

See the pow-inputs file for an example of what such a partitioned input file
would look like.

Benchmark Sets:
==============

In addition to standard benchmarking of functions, one may also generate
custom outputs for a set of functions.  This is currently used by string
function benchmarks where the aim is to compare performance between
implementations at various alignments and for various sizes.

To add a benchset for `foo':

- Add `foo' to the benchset variable.
- Write your bench-foo.c that prints out the measurements to stdout.
- On execution, a bench-foo.out is created in $(objpfx) with the contents of
  stdout.