With talk of a massive power regression in the recently released Linux 3.5 kernel, yesterday I began benchmarking some different systems with varying versions of the Linux kernel looking for any new kernel power regressions on different hardware.
The good news is that this new Linux 3.5 kernel power regression, which James Bottomley classifies himself as a "massive power regression", doesn't appear to be affecting a widespread number of systems. The first system I tested was an Intel Core i7 "Ivy Bridge" desktop, which is doing fine. I was quite curious to see if there are any recent changes in its power consumption since it is the latest-generation platform from Intel and the Linux support continues to be refined.
This testing was with an Intel Core i7 3770K processor with HD 4000 graphics, an ECS-Z77H2 motherboard (Intel Z77 chipset), 8GB of RAM, and a 240GB OCZ Vertex 3 SSD. A development snapshot of Ubuntu 12.10 x86_64 was running on the Intel Ivy Bridge system while the vanilla versions of the Linux 3.2, 3.4, and 3.5 kernels were tested. The Linux 3.3 kernel was skipped due to problems booting in this hardware configuration. These results are namely to look for any changes between the Linux 3.2 and 3.5 kernel releases.
The Phoronix Test Suite performed the benchmarking and automatically measured the system's AC power consumption (along with calculating the power efficiency / performance-per-Watt) from a WattsUp USB-based power meter.