decode one 64-entry block--

get some unsigned bits

returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow

returns 1 if the product of two ints fits in a signed short, 0 on overflow.

set up the kernels

read 16bit value and convert to 24bit RGB

* @brief Store lowest N (vector width) bytes into an unaligned address. */

See header for documentation. */

* @brief Try to quantize an LDR A color using delta encoding. * * At decode time we move one bit from the offset to the base and seize another bit a

* @brief Try to quantize an LDR RGB color using blue-contraction. * * Blue-contraction is only usable if encoded color 1 is larger than color 0. *

* @brief Try to quantize an LDR RGB color using delta encoding. * * At decode time we move one bit from the offset to the base and seize another bit

* @brief Try to quantize an LDR RGB color using delta encoding and blue-contraction. * * Blue-contraction is only usable if encoded color 1 RGB is l

* @brief Validate block size. * * @param block_x The block x dimensions. * @param block_y The block y dimensions. * @param block_z The block

* @brief Validate CPU floating point meets assumptions made in the codec. * * The codec is written with the assumption that float bit patterns are v

* @brief Validate flags. * * @param profile The profile to check. * @param flags The flags to check. * * @return Return @c ASTCENC_SUCCESS

* @brief Wait for stage processing to complete. */

Shift and round for B44 pack (matches OpenEXR's shiftAndRound)

Assume pixel format is FLOAT for all channels.

PXR24 stores 32-bit floats as 24-bit values (truncates 8 mantissa bits) HALF and UINT are stored without modification

Assume pixel format is FLOAT for all channels.

Initialize B44 exp/log lookup tables (matches OpenEXR algorithm)

Parse EXR version header using Reader class

marker output is also

Unpack one 4x4 block from B44 compressed 14 bytes (matches OpenEXR unpack14)

Unpack a 3-byte flat block (all pixels same value)

Utility function to decode color format from file name. Args: file_name: The base file name to parse. Return:

Utility function to decode color profile from file name. Args: file_name: The base file name to parse. Return:

Utility function to decode the flags from file name. Args: file_name: The base file name to parse. Return:

Utility function to decode human readable name from file name. Args: file_name: The base file name to parse. Re

Write the header to a result CSV file. Args: writer: The CSV writer.

Write a record to the CSV file. Args: writer (csv.writer): The CSV writer. record (Record): The record to wr

Add a record to this summary. Args: record (Record): The Record to add.

See header for documentation. */

* @brief The main entry point. * * @param argc The number of arguments. * @param argv The vector of arguments. * * @return 0 on success, non-

See header for documentation. */

See header for documentation. */

* @brief Utility function to create all of the block size descriptors needed. * * This is triggered once via a static initializer. * * Triggering

Build the command line needed for the given test. Args: image: The test image to compress. block_size: The b

* @brief Return the clamped value between min and max. * * It is assumed that neither @c min nor @c max are NaN values. If @c a is NaN * then @c mi

* @brief Return the clamped value between min and max. * * It is assumed that neither @c min nor @c max are NaN values. If @c a is NaN * then @c mi

Idiomatic count-leading zeros, generates native instruction on modern compilers. */

* @brief Compare two canonical patterns to see if they are the same. * * @param part1 The first canonical bit pattern to check. * @param part2

See header for documentation. */

* @brief Compress an image, after any preflight has completed. * * @param[out] ctxo The compressor context. * @param thread_index

* @brief Compress a block using a chosen partitioning and 2 planes of weights. * * @param config The compressor configuratio

See header for documentation. */

See header for documentation. */