Acer Aspire One (Intel Atom N270, 945 GSE, i686)
kernel 3.8-rc1
xf86-video-intel 2.20.17
libdrm 2.4.40
mesa 9.0.1
xserver 1.13.1

LVDS Downclocking enabled:
[root@lamtop 0 0 tests]$ ./gem_gtt_speed 
Time to read 16k through a CPU map:             123.742µs
Time to write 16k through a CPU map:             73.699µs
Time to clear 16k through a CPU map:             59.700µs
Time to clear 16k through a cached CPU map:       1.417µs
Time to pwrite 16k through the CPU:              79.910µs
Time to pread 16k through the CPU:               41.149µs
Time to read 16k through a GTT map:             814.364µs
Time to write 16k through a GTT map:            122.917µs
Time to clear 16k through a GTT map:             62.967µs
Time to clear 16k through a cached GTT map:      42.385µs
Time to read 16k (again) through a GTT map:     769.990µs
Time to pwrite 16k through the GTT:              41.343µs
Time to pread 16k through the GTT:               51.208µs
Time to pwrite 16k through the GTT (clflush):    80.866µs
Time to pread 16k through the GTT (clflush):     56.937µs
Now partial writes.
Time to pwrite 4k through the GTT (clflush):     50.945µs
Time to pread 4k through the GTT (clflush):      48.360µs

Setting tiling mode to X
Time to read 16k through a GTT map:             800.716µs
Time to write 16k through a GTT map:             60.374µs
Time to clear 16k through a GTT map:             68.367µs
Time to clear 16k through a cached GTT map:      38.627µs
Time to read 16k (again) through a GTT map:     798.588µs

Setting tiling mode to Y
Time to read 16k through a GTT map:             820.807µs
Time to write 16k through a GTT map:             81.033µs
Time to clear 16k through a GTT map:             65.275µs
Time to clear 16k through a cached GTT map:      40.555µs
Time to read 16k (again) through a GTT map:     901.823µs


*********************************************************
LVDS Downclocking enabled:
[root@lamtop 0 0 tests]$ ./gem_gtt_speed 
Time to read 16k through a CPU map:              34.387µs
Time to write 16k through a CPU map:             31.050µs
Time to clear 16k through a CPU map:             19.059µs
Time to clear 16k through a cached CPU map:       1.195µs
Time to pwrite 16k through the CPU:              11.425µs
Time to pread 16k through the CPU:               11.075µs
Time to read 16k through a GTT map:             754.961µs
Time to write 16k through a GTT map:             33.068µs
Time to clear 16k through a GTT map:             25.763µs
Time to clear 16k through a cached GTT map:       6.420µs
Time to read 16k (again) through a GTT map:     756.845µs
Time to pwrite 16k through the GTT:               9.400µs
Time to pread 16k through the GTT:               22.195µs
Time to pwrite 16k through the GTT (clflush):    10.170µs
Time to pread 16k through the GTT (clflush):     23.178µs
Now partial writes.
Time to pwrite 4k through the GTT (clflush):      4.248µs
Time to pread 4k through the GTT (clflush):       7.156µs

Setting tiling mode to X
Time to read 16k through a GTT map:             755.179µs
Time to write 16k through a GTT map:             33.134µs
Time to clear 16k through a GTT map:             25.639µs
Time to clear 16k through a cached GTT map:       6.406µs
Time to read 16k (again) through a GTT map:     754.988µs

Setting tiling mode to Y
Time to read 16k through a GTT map:             826.666µs
Time to write 16k through a GTT map:             33.163µs
Time to clear 16k through a GTT map:             29.155µs
Time to clear 16k through a cached GTT map:      18.160µs
Time to read 16k (again) through a GTT map:     826.913µs