2.8 reports a bytecode rate almost 50% higher than before for
Integer>>tinyBenchmark, and I'm guessing that this is caused by the
LargeIntegers plugin.

In any case, if this is true I'm not well enough initiated to fix this.

Henrik