CodeReclaimers / neat-python / 23215631907

85%

push

github

Clean up GPU backend: remove dead matmul, redundant branches, and duplicate logic

- Remove wasted cp.matmul call in CTRNN loop that computed and discarded result;
  use pre-allocated s_buf via out= parameter instead
- Remove per_genome_inputs if/else branches where both paths were identical
  (CuPy broadcasting handles both input shapes)
- Collapse duplicate input-building if/else in both GPU evaluator classes using
  np.zeros((num_steps, *first_input.shape))
- Eliminate N = len(genomes) that could diverge from input_fn's actual shape

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

0 of 14 new or added lines in 2 files covered. (0.0%)

2028 of 2387 relevant lines covered (84.96%)

0.85 hits per line