Intrinsic methods and time to safepoint

[Andrew Haley]

I've been looking at long-running intrinsics, and how long they block
safepoints. At the moment, for example, encrypting a large array might
lead to seconds of safepoint delay: even if you can encrypt a gigabyte
per second, as we do now, a second is a long time for a computer.

The most recent incarnations of concurrent GCs have pushed GC-caused
pauses down into the millisecond range, a superb achievement. However,
the non-GC pauses remain.

I believe we should have a policy to cover how long an intrinsic can
delay without responding to a safepoint, and that it should be in the
millisecond range. It would make almost no difference to the
performance of encryption if chunks handles by a fast intrinsic were,
say, about a megabyte. The difference in performance is so small as to
be immeasurable, and the improvement in the performance of other threads
is vast.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671