I've just been looking at some kernel options and noticed CONFIG_MTRR.
The Documentation says:
On Intel P6 family processors (Pentium Pro, Pentium II and later)
the Memory Type Range Registers (MTRRs) may be used to control
processor access to memory ranges. This is
No, VirtualGL does everything through the X server. This sounds like more of a
driver thing. Not sure what it would speed up in our case. Pixel readback is
already hundreds or even 1000+ Megapixels/sec on modern hardware. That's not
even close to being a bottleneck when compared to compression