Maksim Kutakov

### Details: - Memory reuse for back edges ### Tickets: - 90171

### Details: oneDNN has improved the `inner_product` primitive performance since 1x1 convolution was employed to perform the fully connected operation. Now it makes sense to remove 1x1 convolution primitive usage...

### Details: Gemm convolution supports binary post ops only in the `jit_pp_kernel`, while `ref_pp_kernel` doesn't support such functionality. More over, even supporting binary post ops, the `jit_pp_kernel` doesn't perform all...

[CPU] Support nested internal states

1

### Details: This patch contains all the necessary changes to support internal states in the composite nodes (i.e. If, Loop, TensorIterator operations). Main changes: 1. Move the internal states registry...

[FORK][x64] Remove custom binary PReLU post op

3

# Description OV PR: https://github.com/openvinotoolkit/openvino/pull/23963

[CPU] TBB auto partitioner with muliplier

1

### Details: Try to dynamically balance the workload on hybrid platforms ### Tickets: - *ticket-id*

### Details: - *item1* - *...* ### Tickets: - *ticket-id*

### Details: Support releasing memory allocated for the intermediate tensors each infer request. ### Tickets: - CVS-145873

[DO_NOT_REVIEW][CPU] Memory refactoring

3

### Details: The main purpose of this PR is to remove dnnl dependency from the CPU plugin memory implementation. This is a necessary step toward better modularity that will help...

### Details: In LLMs, certain patterns in subgraphs, such as rotary embeddings calculation, involve math operations over integer numbers (position indices) with results stored in floating-point precision. For such operations,...