PowerMizer Is Not Likely The Problem In 256.xx Issue
One of the articles published on Phoronix this week was NVIDIA's 256.25 Beta Linux Driver Slows Things Down? With NVIDIA's first 256.xx Linux beta driver we encountered significant performance drops from a ZaReason notebook with a NVIDIA GeForce GTS 250M graphics processor. Some thought that PowerMizer was to blame, but this does not appear to be the case.
Since running that article we've received a few emails and forum posts from many not encountering any performance shortcomings with this NVIDIA Linux beta drivers, a few who are affected, and some blaming this problem on PowerMizer -- NVIDIA's technology for dynamically adjusting the GPU's clock frequencies based upon load and other power savings techniques. We're still going to be conducting more tests on other GPUs, but right now it doesn't appear that PowerMizer is to blame for this problem.
With the Phoronix Test Suite we monitored various system sensors (including the NVIDIA GT215 core clock frequency and power consumption) while running tests. As you can see from the graph below, there ended up being no change in the GPU clock frequency between the two drivers when running OpenArena.
We also re-ran the test when the Intel Core i7 notebook was running off its battery (the earlier tests were run on AC power).
The clock frequencies were also the same and there was no significant difference in the power consumption when the GPU was active between the NVIDIA 195.36.15 and 256.25 Beta drivers.
The investigation continues.
Since running that article we've received a few emails and forum posts from many not encountering any performance shortcomings with this NVIDIA Linux beta drivers, a few who are affected, and some blaming this problem on PowerMizer -- NVIDIA's technology for dynamically adjusting the GPU's clock frequencies based upon load and other power savings techniques. We're still going to be conducting more tests on other GPUs, but right now it doesn't appear that PowerMizer is to blame for this problem.
With the Phoronix Test Suite we monitored various system sensors (including the NVIDIA GT215 core clock frequency and power consumption) while running tests. As you can see from the graph below, there ended up being no change in the GPU clock frequency between the two drivers when running OpenArena.
We also re-ran the test when the Intel Core i7 notebook was running off its battery (the earlier tests were run on AC power).
The clock frequencies were also the same and there was no significant difference in the power consumption when the GPU was active between the NVIDIA 195.36.15 and 256.25 Beta drivers.
The investigation continues.
14 Comments