RFR: 8325991: Accelerate Poly1305 on x86_64 using AVX2 instructions [v4]

[Srinivas Vamsi Parasa]

On Thu, 22 Feb 2024 23:56:01 GMT, Sandhya Viswanathan <sviswanathan at openjdk.org> wrote:

>> Srinivas Vamsi Parasa has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   remove unused uniions and fix uses_vl
>
> src/hotspot/cpu/x86/assembler_x86.cpp line 5195:
> 
>> 5193:   assert(VM_Version::supports_avxifma() && (vector_len == AVX_128bit || vector_len == AVX_256bit), "");
>> 5194:   InstructionMark im(this);
>> 5195:   InstructionAttr attributes(vector_len, /* rex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
> 
> uses_vl should be true for all vpmadd52 instructions. 
> The assert could be coded as:
> assert(vector_len == AVX_512bit ? VM_Version::supports_evex() : VM_Version::supports_avxifma(), "");

Please see the updated checks as suggested.

> src/hotspot/cpu/x86/vm_version_x86.cpp line 1360:
> 
>> 1358: 
>> 1359: #ifdef _LP64
>> 1360:   if ((supports_avx512ifma() && supports_avx512vlbw() && MaxVectorSize >= 64) || (supports_avxifma() && MaxVectorSize >= 32))  {
> 
> This could be:
> if ((supports_avx512ifma() && supports_avx512vlbw()) || (supports_avxifma()))  {

Please see the code updated as suggested.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/17881#discussion_r1501108764
PR Review Comment: https://git.openjdk.org/jdk/pull/17881#discussion_r1501109029