Phoronix Test Suite 9.2 Milestone 2 Released
The second development release of Phoronix Test Suite 9.2-Hurdal is now available for open-source, cross-platform and fully-automated benchmarking.
In the few weeks since Phoronix Test Suite 9.2 Milestone 1 that brought Apple/macOS support improvements and other work, the latest changes to find with 9.2 Milestone 2 include:
FORCE_MIN_DURATION_PER_TEST - The Phoronix Test Suite has long offered FORCE_MIN_RUN_TIMES and other environment variables to force the minimum number of times to run a given test and to override the default run count specified in each individual test profile (benchmark). With the new FORCE_MIN_DURATION_PER_TEST it allows specifying a time (in minutes) that each test should be run. Rather than saying to force each test to run at least 3 or 5 times, the FORCE_MIN_DURATION_PER_TEST can be used for ensuring that all tests say run a minimum of ten minutes if wanting to see if the system encounters thermal throttling or other behavior.
Multiple PERFORMANCE_PER_SENSOR= - The system_monitor module now supports specifying multiple sensors (delimited by comma) for generating performance-per-xxx complementary graphs rather than being limited to a single sensor. This makes it easy to say generate performance-per-Watt and performance-per-MHz complementary graphs off a single test run.
Continued Graphing Improvements - Various visualization improvements to pts_Graph for helping to present the data.
Archiving CPU Microcode Versions - The Phoronix Test Suite for years has reported to the user the CPU microcode revision of the system on supported platforms. That CPU microcode version hasn't been archived as part of the test result meta-data. However, given recent CPU issues having microcode updates that do affect the overall system performance, that CPU microcode data is now archived in the result file. That microcode version for each CPU/system is then presented on the overall system table in the result viewers, OpenBenchmarking.org, etc. This offers greater transparency given new CPU microcode releases for helping to explain some performance differences between systems.
Other Minor Fixes + Improvements - Various other small things are being tweaked ahead of the official Phoronix Test Suite 9.2.0 release later this quarter.
The latest Phoronix Test Suite is available via GitHub.
In the few weeks since Phoronix Test Suite 9.2 Milestone 1 that brought Apple/macOS support improvements and other work, the latest changes to find with 9.2 Milestone 2 include:
FORCE_MIN_DURATION_PER_TEST - The Phoronix Test Suite has long offered FORCE_MIN_RUN_TIMES and other environment variables to force the minimum number of times to run a given test and to override the default run count specified in each individual test profile (benchmark). With the new FORCE_MIN_DURATION_PER_TEST it allows specifying a time (in minutes) that each test should be run. Rather than saying to force each test to run at least 3 or 5 times, the FORCE_MIN_DURATION_PER_TEST can be used for ensuring that all tests say run a minimum of ten minutes if wanting to see if the system encounters thermal throttling or other behavior.
Multiple PERFORMANCE_PER_SENSOR= - The system_monitor module now supports specifying multiple sensors (delimited by comma) for generating performance-per-xxx complementary graphs rather than being limited to a single sensor. This makes it easy to say generate performance-per-Watt and performance-per-MHz complementary graphs off a single test run.
Continued Graphing Improvements - Various visualization improvements to pts_Graph for helping to present the data.
Archiving CPU Microcode Versions - The Phoronix Test Suite for years has reported to the user the CPU microcode revision of the system on supported platforms. That CPU microcode version hasn't been archived as part of the test result meta-data. However, given recent CPU issues having microcode updates that do affect the overall system performance, that CPU microcode data is now archived in the result file. That microcode version for each CPU/system is then presented on the overall system table in the result viewers, OpenBenchmarking.org, etc. This offers greater transparency given new CPU microcode releases for helping to explain some performance differences between systems.
Other Minor Fixes + Improvements - Various other small things are being tweaked ahead of the official Phoronix Test Suite 9.2.0 release later this quarter.
The latest Phoronix Test Suite is available via GitHub.
Add A Comment