kernels.hpp File Reference

#include <cstddef>

Include dependency graph for kernels.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
int	postprocess_rg48_to_r12l (void *postprocessor, void *img_custom_data, size_t img_custom_data_size, int size_x, int size_y, struct cmpto_j2k_dec_comp_format *comp_formats, int comp_count, void *input_samples, size_t input_samples_size, void *temp_buffer, size_t temp_buffer_size, void *output_buffer, size_t output_buffer_size, void *stream)
int	preprocess_r12l_to_rg48 (void *preprocessor, void *img_custom_data, size_t img_custom_data_size, int size_x, int size_y, struct cmpto_j2k_enc_comp_format *comp_formats, int comp_count, void *input_samples, size_t input_samples_size, void *output_samples, size_t output_samples_size, void *stream)

Detailed Description

Author

Martin Pulec pulec.nosp@m.@ces.nosp@m.net.c.nosp@m.z

Definition in file kernels.hpp.

Function Documentation

postprocess_rg48_to_r12l()

int postprocess_rg48_to_r12l	(	void *	postprocessor,
		void *	img_custom_data,
		size_t	img_custom_data_size,
		int	size_x,
		int	size_y,
		struct cmpto_j2k_dec_comp_format *	comp_formats,
		int	comp_count,
		void *	input_samples,
		size_t	input_samples_size,
		void *	temp_buffer,
		size_t	temp_buffer_size,
		void *	output_buffer,
		size_t	output_buffer_size,
		void *	vstream )

See also

cmpto_j2k_dec_postprocessor_run_callback_cuda

Definition at line 261 of file kernels.cu.

  {
        cudaStream_t stream = (cudaStream_t) vstream;
        dim3 threads_per_block(256);
        dim3 blocks((((size_x + 7) / 8) + 255) / 256, size_y);
 
        MEASURE_KERNEL_DURATION_START(stream)
 
        if (size_x % 2 == 0) {
                kernel_rg48_to_r12l<uint32_t>
                    <<<blocks, threads_per_block, 0, (cudaStream_t) stream>>>(
                        (uint8_t *) input_samples, (uint8_t *) output_buffer,
                        size_x);
        } else {
                thread_local bool warn_print;
                if (!warn_print) {
                        fprintf(stderr,
                                "%s: Odd width %d px will use slower kernel!\n",
                                __func__, size_x);
                        warn_print = true;
                }
                kernel_rg48_to_r12l<uint16_t>
                    <<<blocks, threads_per_block, 0, (cudaStream_t) stream>>>(
                        (uint8_t *) input_samples, (uint8_t *) output_buffer,
                        size_x);
        }
 
        MEASURE_KERNEL_DURATION_STOP(stream)
 
        return cudaGetLastError();
}

References cudaGetLastError, kernel_rg48_to_r12l(), MEASURE_KERNEL_DURATION_START, and MEASURE_KERNEL_DURATION_STOP.

Here is the call graph for this function:

preprocess_r12l_to_rg48()

int preprocess_r12l_to_rg48	(	void *	preprocessor,
		void *	img_custom_data,
		size_t	img_custom_data_size,
		int	size_x,
		int	size_y,
		struct cmpto_j2k_enc_comp_format *	comp_formats,
		int	comp_count,
		void *	input_samples,
		size_t	input_samples_size,
		void *	output_samples,
		size_t	output_samples_size,
		void *	stream )

Definition at line 453 of file kernels.cu.

{
        (void) preprocessor, (void) img_custom_data,
            (void) img_custom_data_size, (void) comp_formats, (void) comp_count,
            (void) input_samples_size, (void) output_samples_size;
 
        cudaStream_t stream = (cudaStream_t) vstream;
        dim3 threads_per_block(256);
        dim3 blocks((((size_x+ 7) / 8) + 255) / 256, size_y);
 
        MEASURE_KERNEL_DURATION_START(stream)
        if (size_x % 2 == 0) {
                kernel_r12l_to_rg48<uint32_t>
                    <<<blocks, threads_per_block, 0, stream>>>(
                        (uint8_t *) input_samples, (uint8_t *) output_samples,
                        size_x);
        } else {
                thread_local bool warn_print;
                if (!warn_print) {
                        fprintf(stderr,
                                "%s: Odd width %d px will use slower kernel!\n",
                                __func__, size_x);
                        warn_print = true;
                }
                kernel_r12l_to_rg48<uint16_t>
                    <<<blocks, threads_per_block, 0, stream>>>(
                        (uint8_t *) input_samples, (uint8_t *) output_samples,
                        size_x);
        }
        MEASURE_KERNEL_DURATION_STOP(stream)
 
        return cudaGetLastError();
}

References cudaGetLastError, kernel_r12l_to_rg48(), MEASURE_KERNEL_DURATION_START, and MEASURE_KERNEL_DURATION_STOP.

Here is the call graph for this function: