I built the project using the U8G_I2C_OPT_FAST option. It is faster now, although not as fast as I thought it would be. With a 128 byte page size, the timer count (using the 1/256 divider) was 5427, which works out to 86.8 msec. So roughly twice as fast as before, but not 4 times. I'll need to hook up the logic analyzer and see what's happening with the signals.