Stefan Marr
Stefan Marr
It is demonstrated by configurator_test.py test_number_of_experiments_testconf.
Make it possible to track info of multiple source repositories. For instance, for the TruffleRuby project, benchmarks and language implementation my be in different repositories. So, it would be necessary...
See: - https://notes.tweakblogs.net/blog/15291/temporarily-disabling-address-space-layout-randomization-on-linux - https://askubuntu.com/questions/318315/how-can-i-temporarily-disable-aslr-address-space-layout-randomization Roughly, for disabling it: ```bash echo 0 | sudo tee /proc/sys/kernel/randomize_va_space ``` Should probably capture the current value. But likely should be something like: ```bash...
I suspect this might be a different pstate driver or something like that.
Traceback: ``` Minimizing noise with rebench-denoise failed possibly causing benchmark results to vary more. Error: {"scaling_governor": "failed", "no_turbo": "failed", "perf_event_max_sample_rate": 1, "can_set_nice": true, "shielding": "3-39"} Traceback (most recent call last):...
Currently, it seems like the benchmark processes might survive denoise when it is terminated because of a timeout. Needs to be further investigated. There might be multiple options: 1. make...
This PR prepares the 1.1 release. This release focuses on reducing the noise from the system. For this purpse, it introduces the `rebench-denoise` tool, which will adapt system parameters to:...
The AMD driver seems to use another path: `/sys/devices/system/cpu/cpufreq/boost` https://www.kernel.org/doc/Documentation/cpu-freq/boost.txt Fedora also does not seem to have a `cpuset` package.
ReBench should steer the user more aggressively to a setup that is more likely to avoid measuring noise. Much of the [currently observed noise](https://stefan-marr.de/2020/07/is-this-noise-or-does-this-mean-something-benchmarking/) seems a mistake on my part....
Currently, `parallel_interference_factor` is documented but not actually implemented anymore. See #124.