RFR (14) 8235837: Memory access API refinements

[Maurizio Cimadamore]

I think that's just the tip - which is unrelated from which branch you 
are on.

If hg branch says "foreign-memaccess" then I think it should be ok.

I think it's not about code differences, there's probably something 
different in how the various inline/EA thresholds are being setup in our 
machines.

But I do find odd that the unsafe loop doesn't vectorize - so perhaps 
that's where we should look first :-)

Maurizio

On 21/01/2020 14:54, Remi Forax wrote:
> Hi Andrew,
> should be branch "foreign-memaccess"
>
> I also see vectorisation of segment and bytebuffer loop.
>
> Rémi
>
> ----- Mail original -----
>> De: "Andrew Haley" <aph at redhat.com>
>> À: "Maurizio Cimadamore" <maurizio.cimadamore at oracle.com>, "panama-dev at openjdk.java.net'" <panama-dev at openjdk.java.net>
>> Envoyé: Mardi 21 Janvier 2020 15:28:54
>> Objet: Re: RFR (14) 8235837: Memory access API refinements
>> On 1/21/20 11:34 AM, Maurizio Cimadamore wrote:
>>> Copied and pasted your benchmark - still see vectorization (on both
>>> segment and buffer loop):
>> OK, so I'll perhaps dig a bit more. My HEAD is
>>
>> changeset:   59338:4cabdd7fcba8
>> branch:      foreign-jextract
>>
>> $ hg branch
>> foreign-memaccess
>>
>> --
>> 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