Phoronix Test Suite 9.0 Milestone 1 Delivers New Result Viewer, Other Features

[tiger276]

Another useful feature for an upcoming release would be the possibility to set environment variables (e.g. cpu.freq, cpu.power) for the test system from the Phoromatic UI before running a test suite.

Currently I'm also looking for a way to rename existing test results in Phoromatic.

[attah]

You'd likely need to post the URLs for the results you're looking at for me to be able to tell if you're hitting a bug or just running some non-popular configuration.

Here is a fresh run of the Gimp benchmark, with all 4 test options, i.e. alternative 5.
Feels like it should be one of the more common productivity benchmarks to me...

https://openbenchmarking.org/result/1907278-HV-GIMPTEST160

No comparisons available.

Here are some of the old ones:

https://openbenchmarking.org/result/1903169-SP-FIO4K311637

https://openbenchmarking.org/result/1907103-HV-DT123997726

This one now has comparable results:

https://openbenchmarking.org/result/1903167-SP-AIO29515372

I guess i'm more of a benchmarking hipster than i thought.

[Michael]

Hmm, i must be extremely unlucky since i keep experiencing nothing to compare with despite just running e.g. all 4 versions of one test. Sorry for assuming comparisons are completely broken, but i do unfortunately have 100% fail rate over the handful attempts i have made.

So the comparability of system configs have a role here too, what does that mean? Is it more narrow than having the same gpu for a gpu test?

You'd likely need to post the URLs for the results you're looking at for me to be able to tell if you're hitting a bug or just running some non-popular configuration.

[attah]

Hmm, i must be extremely unlucky since i keep experiencing nothing to compare with despite just running e.g. all 4 versions of one test. Sorry for assuming comparisons are completely broken, but i do unfortunately have 100% fail rate over the handful attempts i have made.

So the comparability of system configs have a role here too, what does that mean? Is it more narrow than having the same gpu for a gpu test?

[Michael]

Sorry for not finding the reply feature on mobile. It seems i chose an inopportune time for complaining, since i'm not at my workstation now. But my experience has been that the link provided to openbenchmarking after publishing a result doesn't have any options for comparisons. I'll get back on this with a fresh run in the next few days.

When going off results from a Phoronix article, it's often already a comparison. But for finding others in the community with the same set of results, it's often difficult due to dozens of different benchmarks being run generally in Phoronix articles so the likelihood of others in the community in a comparable system configuration or even just having run all those same tests is quite low, compared to Windows benchmarks where there may be just a single or a few different options/points to compare.

As noted in that CNX link, you can append multiple result IDs to the URL to generate an ad-hoc comparison that way regardless of the number of direct matches.

If going to a result file with one (or a few) tests where there is much greater odds of having matches in the database, you then see the "Compare Results" section to generate a direct comparison, e.g. https://openbenchmarking.org/result/...HV-LINUXKERN99 among the "checkbox shopping cart like comparisons" on the other pages of OB.

[attah]

Old, but less than 11 years. https://www.cnx-software.com/2015/04...ix-test-suite/ So i'm not alone in being in the dark of how to compare easily.

[attah]

Sorry for not finding the reply feature on mobile. It seems i chose an inopportune time for complaining, since i'm not at my workstation now. But my experience has been that the link provided to openbenchmarking after publishing a result doesn't have any options for comparisons. I'll get back on this with a fresh run in the next few days.

[Michael]

So there is still no online comparison tool?

The benchmarks are great, but if i can't compare results, it becomes very hard to gain any knowledge from the results.

Uhh there is an online comparison tool for the past 11 years. OpenBenchmarking.org and before that Phoronix Global.

[attah]

So there is still no online comparison tool?

The benchmarks are great, but if i can't compare results, it becomes very hard to gain any knowledge from the results.

[coder]

I wish OpenBenchmarking.org had a way to compare benchmarks from different runs.

I also wish it had a better search feature, that allowed me to specify keywords per-category. For instance, I was trying to search for test results of a certain CPU model, but the results kept coming back with the model number matching to a GPU in one of the test systems.