RFR: 8341137: Optimize long vector multiplication using x86 VPMUL[U]DQ instruction

[Quan Anh Mai]

On Fri, 18 Oct 2024 05:42:21 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>>> The issue is that a node is not immutable.
>> 
>> I don't see any issues with mutability here. `MulVLNode::_mult_lower_double_word` is constant, so you have to allocate new node if you want to change its value. (And that's exactly what `MulVLNode::Ideal()` does.) But I agree with you that  a dedicated ideal node type (e.g., `MulVI2L`) is much cleaner than `MulVLNode::_mult_lower_double_word`. Still, I'd prefer to see the logic confined in matcher-related code instead.
>
> Hi @iwanowww , @merykitty , I am in process of addressing all your concerns.
> 
> I still feel idealization is the right place to execute this pattern detection, we just need to re-wire the effective inputs bypassing doubleword clearing logic to newly annotated MulVL node and allow clearing IR to sweepout during successive passes, moving it to final graph reshaping just before instruction selection will prevent dead IR cleanups.

@jatin-bhateja I think you can do it at the same place as `Compile::optimize_logic_cones`, we do perform IGVN there. Unless you think this information is needed early in the compiling process, currently I see it is used during matching only, which makes it unnecessary to repeatedly checking it in `Node::Ideal`

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

PR Comment: https://git.openjdk.org/jdk/pull/21244#issuecomment-2421519087