I laid out more SteamOS technical details last night and with the SteamOS Debian package of Valve's heavily-patched Linux 3.10 kernel, there's other dependencies that also need to be changed -- even on an Ubuntu 14.04 development installation -- for the SteamOS kernel to properly install. For those that really want to do so, the easiest way to go about doing this is adding the SteamOS Alchemist beta to the APT sources file on the Ubuntu installation and then install Valve's 3.10 modified kernel package.
While SteamOS is all about the gaming experience and performance, right now this won't get you too much in the way of performance over a stock Linux kernel. As I had written in the earlier article, the SteamOS kernel is disabling the Intel P-State driver over video playback issues when it uses this Intel CPUfreq driver for modern processors. This is likely the cause of the SteamOS kernel being slower than a "vanilla" Linux kernel or Canonical's kernel package on Ubuntu.
I have plenty more SteamOS benchmarks coming out in a matter of hours, but on a separate system I had ran some tests comparing Valve's SteamOS-modified kernel compared to the upstream Linux 3.10.11 kernel for which it's based and then the Linux 3.13 Git kernel (the very latest upstream code). This testing was done from an Ubuntu 14.04 x86_64 installation on a system with a Radeon GPU using the open-source R600 Gallium3D driver. The other tests that should be out later today of the full SteamOS stack are using the NVIDIA binary driver.
As you can see, the SteamOS kernel can actually lead to slower results right now than a pure vanilla kernel. This is likely due to the P-State driver not being used on the SteamOS kernel but used by upstream Linux 3.10 and 3.13 for the Sandy Bridge Extreme system. Or maybe the Valve Linux developers have just been too busy drinking all the Phoronix beers they received? ;)
For more details on this CPUfreq/P-State matter, see this semi-related article. Using the binary NVIDIA driver though you shouldn't as see as big of fluctuation as when using the open-source graphics drivers. The P-State negativity also only affects newer generations of Intel CPUs.
The other initial SteamOS kernel benchmark results can be found in the 1312140-SO-STEAMOSKE25 result file from OpenBenchmarking.org, including the full system specs and logs. You will see that the reported CPU frequency is different for the i7-3960X EE and that is simply due to the different CPUfreq driver and what it's reporting over sysfs, but the CPU frequency as set through the UEFI/BIOS was maintained the same throughout all testing.
More tests forthcoming, but obviously don't expect SteamOS 1.0 Beta to be an instant miracle but it will likely take some months for the SteamOS Linux stack to be finely tuned and perfected.