| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
* Use a larger sample size for sysbench 1.0+ and 64-bit
* Store pointer size with bench results
* Re-enable single thread memory benchmark
Use a larger sample size where available. The small size was
chosen because the 32-bit ARM sysbench 0.4x in raspbian
can't do more than about ~4G, but the problem is that much more
powerful machines burn through that very quickly. The result is
in MiB/s so it should still be comparable, but the results should
be more stable.
Noticed with Ryzen that multi-thread varies significantly based on
what threads are used, but benchmark doesn't really have control
over that.
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
* fix zlib display order
* fix cryptohash MiB/s calculation
* revision and params for other benchmarks
* allow attaching user note to bench result with -u
* don't inlcude the new result value bits if they are empty/invalid
in bench_value_to_str(). bench_value_from_str() doesn't need to be modified.
* bench_results: clean old result cpu name for x86
* use problem_marker() from dmi_memory to mark old version bench results
* benchmark: verifiable test data size and content
- The test data benchmark.data is stored in a file
that could be edited to change the size or content.
/* to guarantee size */
gchar *get_test_data(gsize min_size);
/* to checksum content */
char *md5_digest_str(const char *data, unsigned int len);
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
* add an integer revision field
* add an extra information string field
* zlib benchmark: new revision[1] is 2, so that old results are obviously old,
and now the zlib version string will be stored in extra information.
[1] as of 6a8e19a14305079b03e45eeb0580a45104f300dd
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
The new version uses a fixed time and provides variants for
single-thread, multi-thread, and multi-core.
A few results are included.
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
If (end - start) / threads left a remainder, benchmark_parallel_for()
would start an extra thread for the leftover elements. Now, it has the
last thread process the few extra elements.
Added a note:
benchmark_parallel_for(): element [start] included, but [end] is excluded.
callback(): expected to processes elements [start] through [end] inclusive.
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
New benchmark types:
* benchmark_crunch_for(): call function repeatedly for a number
of seconds; result is number of completions.
* benchmark_parallel(): one call for each available thread
up to n_threads; result is sum of return values.
Tweaks:
* Store return values from callback via
benchmark_parallel_for_dispatcher()
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
Benchmark results store actual number of threads used by benchmark
when it was run. Previously, results assumed all available threads
were used.
Examples:
* CPU Fib only uses one
* FPU FFT uses 4, 2, or 1
* N-Queens uses 10, 5, 2, or 1
Signed-off-by: Burt P <pburt0@gmail.com>
|
|
|
|
gcc was (correctly) assuming that benchmark_parallel_for() returned
an integer, which was being to a double and messing up results. Added
the missing prototype.
Also, populate bench_results array with sane values on module init.
Some other minor cleanups in benchmark code.
|
|
|
|
|
