Phoronix, Benchmarking At XDC2012
At XDC2012 I gave a quick presentation at XDC2012 about Phoronix and Linux benchmarking.
There wasn't much to this brief after-lunch talk beyond saying that more improvements to the Phoronix Test Suite and OpenBenchmarking.org are coming. I also shared my main feature request to graphics driver developers: better standardize on the sysfs/debugfs interfaces for exposing run-time values for the GPU core/memory clock frequencies, available vRAM capacity, and other attributes that are useful for system monitoring applications -- such as the Phoronix Test Suite. Right now the Intel, Radeon, and Nouveau drivers (plus the various other drivers and the binary blobs) are all exposing their clock information and other useful monitoring values in different ways. This lack of standardization just creates unnecessary headaches and could be easily resolved and is a feature request I've been after for years.
The rest of the time was trying to get the X.Org developers to voice their complaints and constructive feedback about Phoronix.com articles with any feedback they want to give, but while they're happy to flame in the forums and blogs, in person no one really wanted to provide much feedback or flames.
Below are the brief slides that were hastily thrown together.
There wasn't much to this brief after-lunch talk beyond saying that more improvements to the Phoronix Test Suite and OpenBenchmarking.org are coming. I also shared my main feature request to graphics driver developers: better standardize on the sysfs/debugfs interfaces for exposing run-time values for the GPU core/memory clock frequencies, available vRAM capacity, and other attributes that are useful for system monitoring applications -- such as the Phoronix Test Suite. Right now the Intel, Radeon, and Nouveau drivers (plus the various other drivers and the binary blobs) are all exposing their clock information and other useful monitoring values in different ways. This lack of standardization just creates unnecessary headaches and could be easily resolved and is a feature request I've been after for years.
The rest of the time was trying to get the X.Org developers to voice their complaints and constructive feedback about Phoronix.com articles with any feedback they want to give, but while they're happy to flame in the forums and blogs, in person no one really wanted to provide much feedback or flames.
Below are the brief slides that were hastily thrown together.
1 Comment