[vector] Infrastructure for calling math routines in .s files

[Paul Sandoz]

Hi Razvan,

I don’t have specific comments on the patch, i’ll let you and Vladimir hash this one out, but i do want to express my appreciation to Intel, you/team for this and further contributions to Java in this area. I think this is great step forward for Java numerics on x86 platforms.

Paul.

> On May 18, 2018, at 4:34 PM, Lupusoru, Razvan A <razvan.a.lupusoru at intel.com> wrote:
> 
> Hi everyone,
> 
> The Vector API defines multiple math operations for FP including sin, cos, tan, exp. In order to get good performance, it is desirable to have implementations that perform these operations in vectorized manner (instead of what is happening now which is boxing and calling of scalar optimized routine).
> 
> We would like to contribute assembly files which contain these vectorized implementations. These implementations are from Intel Short Vector Math Library (https://software.intel.com/en-us/node/524289).
> 
> So I am sending out the initial patch which adds infrastructure (plus one implementation - exp) in order to be able to call these assembly routines from JIT compiled code. In a subsequent patches, we will contribute the other implementations in form of assembly files.
> http://cr.openjdk.java.net/~rlupusoru/panama/webrev_callsvml_09/index.html
> 
> Please take a look at the patch and let me know if you have any questions. I would like to point out that among changes there are also a few which make sure that compiler does not tag with "NO_FP" any calls that may end up calling methods that are using FP/vector code. This is quite important for Vector API since unlike auto-vectorizer, there may be live vector registers at any point in the code.
> 
> Thanks,
> Razvan
>