Crate jxl_encoder_simd 

SIMD-accelerated primitives for jxl_encoder.

This crate wraps platform-specific SIMD intrinsics behind safe public functions. The main encoder crate (jxl_encoder) maintains #![forbid(unsafe_code)] and calls into these safe wrappers.

Uses archmage for token-based SIMD dispatch and magetypes for cross-platform vector types.

Direct variant access

Each kernel is available in three forms:

• A dispatching function (e.g. dct_8x8) that picks the best at runtime
• Concrete _avx2(token, ...) / _neon(token, ...) / _scalar(...) variants

For hot loops, callers should summon a token once, then call the concrete variant directly from an #[arcane] function so LLVM can inline across the target-feature boundary.

Structs

EntropyCoeffResult

Results from vectorized entropy coefficient processing.

NeonToken

Proof that NEON is available.

Constants

NEWTON_EPS_DEFAULT

Newton’s method constants.

NEWTON_MAX_ITERS_DEFAULT

Traits

SimdToken

Marker trait for SIMD capability tokens.

Functions

cfl_find_best_multiplier

Find the best integer CfL multiplier via regularized least-squares.

cfl_find_best_multiplier_neon

cfl_find_best_multiplier_newton

Find the best integer CfL multiplier via Newton’s method with perceptual cost.

cfl_find_best_multiplier_newton_scalar

Scalar Newton’s method for CfL multiplier.

cfl_find_best_multiplier_scalar

compute_block_l2_errors

Compute per-block masked weighted L2 error between original and reconstructed XYB planes.

compute_block_l2_errors_neon

compute_block_l2_errors_scalar

compute_mask1x1

Compute per-pixel masking field from XYB Y channel.

compute_mask1x1_neon

compute_mask1x1_scalar

compute_pre_erosion

Compute pre-erosion map from Y channel: per-pixel stencil with 4× downsampling.

compute_pre_erosion_neon

compute_pre_erosion_scalar

dct_4x4_full

Compute full DCT4X4 transform for 8x8 pixel block.

dct_4x4_full_scalar

dct_4x8_full

Compute full DCT4X8 transform for 8x8 pixel block.

dct_4x8_full_scalar

dct_8x8

Compute scaled 8x8 forward DCT with SIMD acceleration.

dct_8x4_full

Compute full DCT8X4 transform for 8x8 pixel block.

dct_8x4_full_scalar

dct_8x8_neon

NEON 8x8 forward DCT: two-pass (4 columns at a time), in-register transpose.

dct_8x8_scalar

dct_8x16

Compute scaled 8x16 forward DCT with SIMD acceleration.

dct_8x16_neon

NEON 8x16 forward DCT.

dct_8x16_scalar

dct_16x8

Compute scaled 16x8 forward DCT with SIMD acceleration.

dct_16x8_neon

NEON 16x8 forward DCT.

dct_16x8_scalar

dct_16x16

Compute 16x16 forward DCT with SIMD acceleration.

dct_16x32

Compute 16×32 forward DCT with SIMD acceleration.

dct_16x16_neon

NEON 16x16 forward DCT: process 4 rows at a time.

dct_16x16_scalar

dct_16x32_scalar

Scalar 16×32 forward DCT.

dct_32x16

Compute 32×16 forward DCT with SIMD acceleration.

dct_32x32

Compute 32×32 forward DCT with SIMD acceleration.

dct_32x64

Compute 32×64 forward DCT with SIMD acceleration.

dct_32x16_scalar

Scalar 32×16 forward DCT.

dct_32x32_scalar

Scalar 32×32 forward DCT.

dct_32x64_scalar

Scalar 32×64 forward DCT.

dct_64x32

Compute 64×32 forward DCT with SIMD acceleration.

dct_64x64

Compute 64×64 forward DCT with SIMD acceleration.

dct_64x32_scalar

Scalar 64×32 forward DCT.

dct_64x64_scalar

Scalar 64×64 forward DCT.

denoise_channel

Wiener filter for a single channel (runtime dispatch).

denoise_channel_neon

denoise_channel_scalar

Wiener filter for a single channel (scalar).

dequant_block_dct8

Dequantize a DCT8 block and apply CfL (chroma-from-luma) in one pass.

dequant_dct8_neon

dequant_dct8_scalar

entropy_coeffs_neon

entropy_coeffs_scalar

entropy_estimate_coeffs

Vectorized entropy coefficient processing.

epf_step1

Apply EPF Step 1 to 3-channel XYB planes.

epf_step2

Apply EPF Step 2 to 3-channel XYB planes.

epf_step1_neon

epf_step1_scalar

epf_step2_neon

epf_step2_scalar

fast_log2f

Fast log2 approximation. Max relative error ~3e-7. Input must be > 0.

fast_pow2f

Fast base-2 exponentiation. Max relative error ~3e-7.

fast_powf

Fast power function: base^exponent. Max relative error ~3e-5.

forward_xyb_neon

forward_xyb_scalar

fused_dct8_entropy

Fused DCT8 + entropy estimation.

fused_dct8_entropy_fallback

Fallback: extract + separate DCT + entropy using dispatching functions.

gab_smooth_channel

Apply 3x3 weighted gaborish smooth to a single channel in-place.

gab_smooth_neon

NEON gab smooth: processes 4 pixels per iteration in interior rows.

gab_smooth_scalar

gaborish_5x5_channel

Apply the 5x5 gaborish inverse kernel to a single channel.

gaborish_5x5_neon

NEON gaborish 5x5: processes 4 pixels per iteration in interior region.

gaborish_5x5_scalar

idct_4x4_full

Compute full inverse DCT4X4 transform for 8x8 coefficient block.

idct_4x4_full_scalar

idct_4x8_full

Compute full inverse DCT4X8 transform for 8x8 coefficient block.

idct_4x8_full_scalar

idct_8x8

Compute scaled 8x8 inverse DCT with SIMD acceleration.

idct_8x4_full

Compute full inverse DCT8X4 transform for 8x8 coefficient block.

idct_8x4_full_scalar

idct_8x8_neon

NEON 8x8 inverse DCT.

idct_8x8_scalar

idct_8x16

Compute 8x16 inverse DCT with SIMD acceleration.

idct_8x16_neon

NEON 8x16 inverse DCT.

idct_8x16_scalar

idct_16x8

Compute 16x8 inverse DCT with SIMD acceleration.

idct_16x8_neon

NEON 16x8 inverse DCT.

idct_16x8_scalar

idct_16x16

Compute 16x16 inverse DCT with SIMD acceleration.

idct_16x32

Compute 16×32 inverse DCT with SIMD acceleration.

idct_16x16_neon

NEON 16x16 inverse DCT: process 4 rows at a time.

idct_16x16_scalar

idct_16x32_scalar

Scalar 16×32 inverse DCT.

idct_32x16

Compute 32×16 inverse DCT with SIMD acceleration.

idct_32x32

Compute 32×32 inverse DCT with SIMD acceleration.

idct_32x64

Compute 32×64 inverse DCT with SIMD acceleration.

idct_32x16_scalar

Scalar 32×16 inverse DCT.

idct_32x32_scalar

Scalar 32×32 inverse DCT.

idct_32x64_scalar

Scalar 32×64 inverse DCT.

idct_64x32

Compute 64×32 inverse DCT with SIMD acceleration.

idct_64x64

Compute 64×64 inverse DCT with SIMD acceleration.

idct_64x32_scalar

Scalar 64×32 inverse DCT.

idct_64x64_scalar

Scalar 64×64 inverse DCT.

inverse_xyb_neon

inverse_xyb_planar_neon

inverse_xyb_planar_scalar

inverse_xyb_scalar

linear_rgb_to_xyb_batch

Convert separate R, G, B channel buffers to separate X, Y, B channel buffers.

pad_plane

Pad a single channel plane with edge replication.

per_block_modulations

Apply per-block modulations to aq_map in-place.

per_block_modulations_neon

per_block_modulations_scalar

pixel_domain_loss

Compute pixel-domain loss for one channel of a block.

pixel_domain_loss_neon

pixel_domain_loss_scalar

quantize_block_dct8

Quantize a DCT8 block (64 coefficients) with dead-zone thresholding.

quantize_block_large

Quantize AC coefficients for a large block (DCT16+) to a flat output buffer.

quantize_dct8_neon

quantize_dct8_scalar

quantize_large_neon

quantize_large_scalar

shannon_entropy_bits

Compute Shannon entropy of a histogram: -sum(count * log2(count / total)).

shannon_entropy_neon

shannon_entropy_scalar

Scalar Shannon entropy using fast_log2f.

transpose_8x8

Transpose an 8x8 f32 matrix.

transpose_8x8_neon

NEON 8x8 transpose using four 4x4 sub-transposes.

vec_f32_dirty

Allocate a zero-initialized Vec<f32> of length n (safe default path).

xyb_to_linear_rgb_batch

Convert separate X, Y, B channel buffers to interleaved linear RGB.

xyb_to_linear_rgb_planar

Convert separate X, Y, B channel buffers to planar linear RGB.