[aarch64-port-dev ] Some benchmarking help required

[Andrew Haley]

On 05/17/2018 12:33 PM, Dmitry Chuyko wrote:
> Sparse-small -1-6% regression looks real. Monte Carlo on T99 looks 
> promising as it may work so for many CAS loops. Other +3-17% numbers are 
> also very nice.

Very cool, thanks for doing this.  I'll investigate some more.

The problem with a lot of these measurements is that JIT compilation
can be very unstable, so that a tiny change can change inlining
behaviour.  It can behave like a chaotic attractor.

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