[hardware] fix integer reduction commit logic

[Navaneeth-KunhiPurayil]

This PR fixes an issue observed when we have back-to-back reduction operations. lane0 goes to the SIMD_REDUCTION state and set's result_queue_valid_d to 1'b1 and sets commit_cnt_d = '0. This creates a problem if we have back-to-back reductions, since the alu_state_d switches from SIMD_REDUCTION to INTRA_LANE_REDUCTION without staying in SIMD_REDUCTION or NO_REDUCTION state for one more cycle for the commit to happen.

Now, the commit_cnt_d is set to '0 only after a successful commit to the VRF.

Changelog

Fixed

• Commit logic after SIMD_REDUCTION for lane0

Checklist

• Automated tests pass
• Changelog updated
• Code style guideline is observed

Please check our contributing guidelines before opening a Pull Request.

[hopang-0221]

Hi navaneeth, this fix looks good to me. Thanks!