Valhalla/MVT microbenchmarks (first benchmarks and first results)

Maurizio Cimadamore maurizio.cimadamore at oracle.com
Thu Jul 27 00:15:06 UTC 2017


Great work - thanks for putting so much effort on this Sergey, I found 
all your comments very easy to read/understand and I'm not too deep in 
jmh benchmarks.

I'd say those results look good - especially considered the early stage 
of the implementation. Basically, with invokeExact, MVT performs more or 
less identical to the baseline in all benchmarks. The only exception 
seems to be those multIJKTotalLoop/multIKJTotalLoop benchmarks, in 
which, as you point out, something is preventing full MH compilation. 
I'm sure our C2 gurus will look into that soon.

Thanks
Maurizio


On 27/07/17 00:58, Sergey Kuksenko wrote:
> Hi All,
>
> On the link http://cr.openjdk.java.net/~skuksenko/valhalla/benchmarks/
> you could find results  of the first round MVT benchmarking.
>
> On the page you could find benchmarks description, archive with source 
> files and first results.
>
> Please, review my benchmarks especially with MethodHandles.
>



More information about the valhalla-dev mailing list