The first nine patches re-factor the mode-setting code that Eric Anholt has been working on. This re-factored code will hopefully result in less breakage across generations of Intel graphics processor and all-around should be an improvement to Intel's Linux support.
The big part of this patch-set is the LLC caching series for Sandy Bridge. As illustrated on Phoronix, the LLC caching is able to significantly boost the Intel open-source graphics performance across the board. It's really quite good and another step in the right direction for the Sandy Bridge Linux support compared to the Windows performance.
The rest of the patches offer up RC6 fixes for Ironlake, Jesse's module parameter for Sandy Bridge, small code re-factoring, pipelined fence fixes, and caching of GT FIFO count. The GT FIFO count caching should result in a performance gain for the Intel X.Org DDX (2D) driver, but the work is likely to be replaced by the proper GT read/write fixes.
What this set of 30 kernel patches doesn't offer though is the G45 VA-API video acceleration support or the initial bits of code for Intel's next-generation "Ivy Bridge" hardware. It hasn't been publicly announced yet, but expect the Ivy Bridge hardware enablement to begin with Linux 2.6.40.
While Sandy Bridge was just launched back in January and its support is still maturing, the initial Ivy Bridge graphics support should land with Linux 2.6.40 when it comes to the kernel-space side of the graphics stack. (Not sure offhand when the DDX and Mesa user-space bits will be pushed, I'd suspect around the same time.)
Ivy Bridge though isn't expected for release until near the end of the calendar year, so there should be a few releases for its support to be ironed out and hopefully with good "out of the box" support compared to Sandy Bridge.
The set of 30 patches can be found on the mailing list.