godot/modules/cvtt/image_compress_cvtt.cpp
Rémi Verschelde 4b8946102b
Merge pull request #91535 from BlueCube3310/betsy-bc6h
Add Betsy to speed up BC6 compression
2024-08-21 06:38:03 +02:00

368 lines
13 KiB
C++

/**************************************************************************/
/* image_compress_cvtt.cpp */
/**************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* https://godotengine.org */
/**************************************************************************/
/* Copyright (c) 2014-present Godot Engine contributors (see AUTHORS.md). */
/* Copyright (c) 2007-2014 Juan Linietsky, Ariel Manzur. */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. */
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/**************************************************************************/
#include "image_compress_cvtt.h"
#include "core/object/worker_thread_pool.h"
#include "core/os/os.h"
#include "core/string/print_string.h"
#include "core/templates/safe_refcount.h"
#include <ConvectionKernels.h>
struct CVTTCompressionJobParams {
bool is_hdr = false;
bool is_signed = false;
int bytes_per_pixel = 0;
cvtt::BC7EncodingPlan bc7_plan;
cvtt::Options options;
};
struct CVTTCompressionRowTask {
const uint8_t *in_mm_bytes = nullptr;
uint8_t *out_mm_bytes = nullptr;
int y_start = 0;
int width = 0;
int height = 0;
};
struct CVTTCompressionJobQueue {
CVTTCompressionJobParams job_params;
const CVTTCompressionRowTask *job_tasks = nullptr;
uint32_t num_tasks = 0;
SafeNumeric<uint32_t> current_task;
};
static void _digest_row_task(const CVTTCompressionJobParams &p_job_params, const CVTTCompressionRowTask &p_row_task) {
const uint8_t *in_bytes = p_row_task.in_mm_bytes;
uint8_t *out_bytes = p_row_task.out_mm_bytes;
int w = p_row_task.width;
int h = p_row_task.height;
int y_start = p_row_task.y_start;
int y_end = y_start + 4;
int bytes_per_pixel = p_job_params.bytes_per_pixel;
bool is_hdr = p_job_params.is_hdr;
bool is_signed = p_job_params.is_signed;
cvtt::PixelBlockU8 input_blocks_ldr[cvtt::NumParallelBlocks];
cvtt::PixelBlockF16 input_blocks_hdr[cvtt::NumParallelBlocks];
for (int x_start = 0; x_start < w; x_start += 4 * cvtt::NumParallelBlocks) {
int x_end = x_start + 4 * cvtt::NumParallelBlocks;
for (int y = y_start; y < y_end; y++) {
int first_input_element = (y - y_start) * 4;
const uint8_t *row_start;
if (y >= h) {
row_start = in_bytes + (h - 1) * (w * bytes_per_pixel);
} else {
row_start = in_bytes + y * (w * bytes_per_pixel);
}
for (int x = x_start; x < x_end; x++) {
const uint8_t *pixel_start;
if (x >= w) {
pixel_start = row_start + (w - 1) * bytes_per_pixel;
} else {
pixel_start = row_start + x * bytes_per_pixel;
}
int block_index = (x - x_start) / 4;
int block_element = (x - x_start) % 4 + first_input_element;
if (is_hdr) {
memcpy(input_blocks_hdr[block_index].m_pixels[block_element], pixel_start, bytes_per_pixel);
input_blocks_hdr[block_index].m_pixels[block_element][3] = 0x3c00; // 1.0 (unused)
} else {
memcpy(input_blocks_ldr[block_index].m_pixels[block_element], pixel_start, bytes_per_pixel);
}
}
}
uint8_t output_blocks[16 * cvtt::NumParallelBlocks];
if (is_hdr) {
if (is_signed) {
cvtt::Kernels::EncodeBC6HS(output_blocks, input_blocks_hdr, p_job_params.options);
} else {
cvtt::Kernels::EncodeBC6HU(output_blocks, input_blocks_hdr, p_job_params.options);
}
} else {
cvtt::Kernels::EncodeBC7(output_blocks, input_blocks_ldr, p_job_params.options, p_job_params.bc7_plan);
}
unsigned int num_real_blocks = ((w - x_start) + 3) / 4;
if (num_real_blocks > cvtt::NumParallelBlocks) {
num_real_blocks = cvtt::NumParallelBlocks;
}
memcpy(out_bytes, output_blocks, 16 * num_real_blocks);
out_bytes += 16 * num_real_blocks;
}
}
static void _digest_job_queue(void *p_job_queue, uint32_t p_index) {
CVTTCompressionJobQueue *job_queue = static_cast<CVTTCompressionJobQueue *>(p_job_queue);
uint32_t num_tasks = job_queue->num_tasks;
uint32_t total_threads = WorkerThreadPool::get_singleton()->get_thread_count();
uint32_t start = p_index * num_tasks / total_threads;
uint32_t end = (p_index + 1 == total_threads) ? num_tasks : ((p_index + 1) * num_tasks / total_threads);
for (uint32_t i = start; i < end; i++) {
_digest_row_task(job_queue->job_params, job_queue->job_tasks[i]);
}
}
void image_compress_cvtt(Image *p_image, Image::UsedChannels p_channels) {
uint64_t start_time = OS::get_singleton()->get_ticks_msec();
if (p_image->is_compressed()) {
return; //do not compress, already compressed
}
int w = p_image->get_width();
int h = p_image->get_height();
bool is_ldr = (p_image->get_format() <= Image::FORMAT_RGBA8);
bool is_hdr = (p_image->get_format() >= Image::FORMAT_RF) && (p_image->get_format() <= Image::FORMAT_RGBE9995);
if (!is_ldr && !is_hdr) {
return; // Not a usable source format
}
cvtt::Options options;
uint32_t flags = cvtt::Flags::Default;
flags |= cvtt::Flags::BC7_RespectPunchThrough;
if (p_channels == Image::USED_CHANNELS_RG) { //guessing this is a normal map
flags |= cvtt::Flags::Uniform;
}
options.flags = flags;
Image::Format target_format = Image::FORMAT_BPTC_RGBA;
bool is_signed = false;
if (is_hdr) {
if (p_image->get_format() != Image::FORMAT_RGBH) {
p_image->convert(Image::FORMAT_RGBH);
}
const uint8_t *rb = p_image->get_data().ptr();
const uint16_t *source_data = reinterpret_cast<const uint16_t *>(&rb[0]);
int pixel_element_count = w * h * 3;
for (int i = 0; i < pixel_element_count; i++) {
if ((source_data[i] & 0x8000) != 0 && (source_data[i] & 0x7fff) != 0) {
is_signed = true;
break;
}
}
target_format = is_signed ? Image::FORMAT_BPTC_RGBF : Image::FORMAT_BPTC_RGBFU;
} else {
p_image->convert(Image::FORMAT_RGBA8); //still uses RGBA to convert
}
const uint8_t *rb = p_image->get_data().ptr();
Vector<uint8_t> data;
int64_t target_size = Image::get_image_data_size(w, h, target_format, p_image->has_mipmaps());
int mm_count = p_image->has_mipmaps() ? Image::get_image_required_mipmaps(w, h, target_format) : 0;
data.resize(target_size);
int shift = Image::get_format_pixel_rshift(target_format);
uint8_t *wb = data.ptrw();
int64_t dst_ofs = 0;
CVTTCompressionJobQueue job_queue;
job_queue.job_params.is_hdr = is_hdr;
job_queue.job_params.is_signed = is_signed;
job_queue.job_params.options = options;
job_queue.job_params.bytes_per_pixel = is_hdr ? 6 : 4;
cvtt::Kernels::ConfigureBC7EncodingPlanFromQuality(job_queue.job_params.bc7_plan, 5);
// Amdahl's law (Wikipedia)
// If a program needs 20 hours to complete using a single thread, but a one-hour portion of the program cannot be parallelized,
// therefore only the remaining 19 hours (p = 0.95) of execution time can be parallelized, then regardless of how many threads are devoted
// to a parallelized execution of this program, the minimum execution time cannot be less than one hour.
//
// The number of executions with different inputs can be increased while the latency is the same.
Vector<CVTTCompressionRowTask> tasks;
for (int i = 0; i <= mm_count; i++) {
int bw = w % 4 != 0 ? w + (4 - w % 4) : w;
int bh = h % 4 != 0 ? h + (4 - h % 4) : h;
int64_t src_ofs = p_image->get_mipmap_offset(i);
const uint8_t *in_bytes = &rb[src_ofs];
uint8_t *out_bytes = &wb[dst_ofs];
for (int y_start = 0; y_start < h; y_start += 4) {
CVTTCompressionRowTask row_task;
row_task.width = w;
row_task.height = h;
row_task.y_start = y_start;
row_task.in_mm_bytes = in_bytes;
row_task.out_mm_bytes = out_bytes;
tasks.push_back(row_task);
out_bytes += 16 * (bw / 4);
}
dst_ofs += (MAX(4, bw) * MAX(4, bh)) >> shift;
w = MAX(w / 2, 1);
h = MAX(h / 2, 1);
}
const CVTTCompressionRowTask *tasks_rb = tasks.ptr();
job_queue.job_tasks = &tasks_rb[0];
job_queue.num_tasks = static_cast<uint32_t>(tasks.size());
WorkerThreadPool::GroupID group_task = WorkerThreadPool::get_singleton()->add_native_group_task(&_digest_job_queue, &job_queue, WorkerThreadPool::get_singleton()->get_thread_count(), -1, true, SNAME("CVTT Compress"));
WorkerThreadPool::get_singleton()->wait_for_group_task_completion(group_task);
p_image->set_data(p_image->get_width(), p_image->get_height(), p_image->has_mipmaps(), target_format, data);
print_verbose(vformat("CVTT: Encoding took %d ms.", OS::get_singleton()->get_ticks_msec() - start_time));
}
void image_decompress_cvtt(Image *p_image) {
Image::Format target_format;
bool is_signed = false;
bool is_hdr = false;
Image::Format input_format = p_image->get_format();
switch (input_format) {
case Image::FORMAT_BPTC_RGBA:
target_format = Image::FORMAT_RGBA8;
break;
case Image::FORMAT_BPTC_RGBF:
case Image::FORMAT_BPTC_RGBFU:
target_format = Image::FORMAT_RGBH;
is_signed = (input_format == Image::FORMAT_BPTC_RGBF);
is_hdr = true;
break;
default:
return; // Invalid input format
};
int w = p_image->get_width();
int h = p_image->get_height();
const uint8_t *rb = p_image->get_data().ptr();
Vector<uint8_t> data;
int64_t target_size = Image::get_image_data_size(w, h, target_format, p_image->has_mipmaps());
int mm_count = p_image->get_mipmap_count();
data.resize(target_size);
uint8_t *wb = data.ptrw();
int bytes_per_pixel = is_hdr ? 6 : 4;
int64_t dst_ofs = 0;
for (int i = 0; i <= mm_count; i++) {
int64_t src_ofs = p_image->get_mipmap_offset(i);
const uint8_t *in_bytes = &rb[src_ofs];
uint8_t *out_bytes = &wb[dst_ofs];
cvtt::PixelBlockU8 output_blocks_ldr[cvtt::NumParallelBlocks];
cvtt::PixelBlockF16 output_blocks_hdr[cvtt::NumParallelBlocks];
for (int y_start = 0; y_start < h; y_start += 4) {
int y_end = y_start + 4;
for (int x_start = 0; x_start < w; x_start += 4 * cvtt::NumParallelBlocks) {
uint8_t input_blocks[16 * cvtt::NumParallelBlocks];
memset(input_blocks, 0, sizeof(input_blocks));
unsigned int num_real_blocks = ((w - x_start) + 3) / 4;
if (num_real_blocks > cvtt::NumParallelBlocks) {
num_real_blocks = cvtt::NumParallelBlocks;
}
memcpy(input_blocks, in_bytes, 16 * num_real_blocks);
in_bytes += 16 * num_real_blocks;
int x_end = x_start + 4 * num_real_blocks;
if (is_hdr) {
if (is_signed) {
cvtt::Kernels::DecodeBC6HS(output_blocks_hdr, input_blocks);
} else {
cvtt::Kernels::DecodeBC6HU(output_blocks_hdr, input_blocks);
}
} else {
cvtt::Kernels::DecodeBC7(output_blocks_ldr, input_blocks);
}
for (int y = y_start; y < y_end; y++) {
int first_input_element = (y - y_start) * 4;
uint8_t *row_start;
if (y >= h) {
row_start = out_bytes + (h - 1) * (w * bytes_per_pixel);
} else {
row_start = out_bytes + y * (w * bytes_per_pixel);
}
for (int x = x_start; x < x_end; x++) {
uint8_t *pixel_start;
if (x >= w) {
pixel_start = row_start + (w - 1) * bytes_per_pixel;
} else {
pixel_start = row_start + x * bytes_per_pixel;
}
int block_index = (x - x_start) / 4;
int block_element = (x - x_start) % 4 + first_input_element;
if (is_hdr) {
memcpy(pixel_start, output_blocks_hdr[block_index].m_pixels[block_element], bytes_per_pixel);
} else {
memcpy(pixel_start, output_blocks_ldr[block_index].m_pixels[block_element], bytes_per_pixel);
}
}
}
}
}
dst_ofs += w * h * bytes_per_pixel;
w >>= 1;
h >>= 1;
}
p_image->set_data(p_image->get_width(), p_image->get_height(), p_image->has_mipmaps(), target_format, data);
}