VolodymyrBg requested wasmtime-wasi-reviewers for a review on PR #12618.
VolodymyrBg opened PR #12618 from VolodymyrBg:fix/onnx-optimize-byte-conversion to bytecodealliance:main:
The functions created unnecessary intermediate Vec allocations -
f32_vec_to_bytesallocated aVec<[u8; 4]>then flattened it, andbytes_to_f32_vechad triple allocation with double collect.
Replaced with single-passflat_mapand simplifiedmapwith singlecollect, matching the pattern already used in winml.rs.
alexcrichton submitted PR review.
alexcrichton has enabled auto merge for PR #12618.
alexcrichton added PR #12618 perf: optimize f32_vec_to_bytes and bytes_to_f32_vec to the merge queue
alexcrichton removed PR #12618 perf: optimize f32_vec_to_bytes and bytes_to_f32_vec from the merge queue
alexcrichton merged PR #12618.
Last updated: Feb 24 2026 at 04:36 UTC