[CPU] [Snippets] Snippets for ARM

[xuchen-intel]

Details:

• Feature

  • POC scope of Snippets on ARM architecture supports simple elementwise (combinations of Add, Multiply, Divide, Relu, Exp and Tanh ops) subgraphs for FP32 precision.

• Implementations

  • aarch64 CPU Generator and TargetMachine
  • aarch64 load/store emitters
  • aarch64 control-flow emitters (Kernel, LoopBegin, LoopEnd, BroadcastMove, Scalar, Fill)

Tickets:

• 134078

[xuchen-intel]

@dmitry-gorokhov Hi Dmitry, could you please take a look?

[dmitry-gorokhov]

@eshoguli @a-sidorova Guys, could you please review?

[wenjiew]

@xuchen-intel Anything further work to be done before merge? Thanks!

[a-sidorova]

Review status:

• Need to finish SnippetsMarkSkipped impl and add tests that cover this pass.
• Need to resolve the problem with incorrect layout choice. In the progress. Waiting for the execution graphs offline.