RFR 8167065 - Adding support for double precision shifting for x86

[Vladimir Kozlov]

I don't see big benefits to have it in jdk 9 - no numbers for normal applications or big benchmarks. Only numbers for special microbenchmark was shown.

And I am concern about possible regressions so later in jdk 9 cycle.

I would prefer to move it to jdk 10. Unless you think it can give huge benefit to customers as we had with crypto intrinsics.

thanks,
Vladimir

On 10/28/16 1:00 AM, Andrew Haley wrote:
> On 18/10/16 22:19, Vladimir Kozlov wrote:
>> The problem is that it should be RFE(Enchantment) and not Bug. It adds
>> new intrinsics so we have to change its type. We should either target it
>> to JDK 10 or ask FC extension for JDK 9. There are examples for that.
>
> Any more thoughts on this?
>
> Andrew.
>