voxel-rs/src/voxel/brickmap.rs

633 lines
23 KiB
Rust

use std::collections::HashSet;
use wgpu::util::DeviceExt;
use crate::{
gfx::{self},
math,
};
#[repr(C)]
#[derive(Debug, Default, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
struct Brickmap {
pub bitmask: [u32; 16],
pub shading_table_offset: u32,
pub lod_color: u32,
}
#[repr(C)]
#[derive(Debug, Default, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
struct WorldState {
brickgrid_dims: [u32; 3],
_pad: u32,
}
#[derive(Debug, Default, Copy, Clone)]
struct BrickmapCacheEntry {
grid_idx: usize,
shading_table_offset: u32,
}
#[repr(C)]
#[derive(Debug, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
struct BrickmapUnpackElement {
cache_idx: u32,
brickmap: Brickmap,
shading_element_count: u32,
shading_elements: [u32; 512],
}
enum BrickgridFlag {
_Empty = 0,
_Unloaded = 1,
_Loading = 2,
Loaded = 4,
}
#[derive(Debug)]
pub struct BrickmapManager {
state_uniform: WorldState,
state_buffer: wgpu::Buffer,
brickgrid: Vec<u32>,
brickgrid_buffer: wgpu::Buffer,
brickmap_cache_map: Vec<Option<BrickmapCacheEntry>>,
brickmap_cache_idx: usize,
brickmap_buffer: wgpu::Buffer,
shading_table_buffer: wgpu::Buffer,
shading_table_allocator: ShadingTableAllocator,
feedback_buffer: wgpu::Buffer,
feedback_result_buffer: wgpu::Buffer,
unpack_max_count: usize,
brickgrid_staged: HashSet<usize>,
brickgrid_unpack_buffer: wgpu::Buffer,
brickmap_staged: Vec<BrickmapUnpackElement>,
brickmap_unpack_buffer: wgpu::Buffer,
}
// TODO:
// - Brickworld system
impl BrickmapManager {
pub fn new(
context: &gfx::Context,
brickgrid_dims: glam::UVec3,
brickmap_cache_size: usize,
shading_table_bucket_size: u32,
max_requested_brickmaps: u32,
max_uploaded_brickmaps: u32,
) -> Self {
let device = &context.device;
let state_uniform = WorldState {
brickgrid_dims: [brickgrid_dims.x, brickgrid_dims.y, brickgrid_dims.z],
..Default::default()
};
let brickgrid =
vec![1u32; (brickgrid_dims.x * brickgrid_dims.y * brickgrid_dims.z) as usize];
let brickmap_cache = vec![Brickmap::default(); brickmap_cache_size];
let brickmap_cache_map = vec![None; brickmap_cache.capacity()];
let shading_table_allocator = ShadingTableAllocator::new(4, shading_table_bucket_size);
let shading_table = vec![0u32; shading_table_allocator.total_elements as usize];
let mut feedback_data = vec![0u32; 4 + 4 * max_requested_brickmaps as usize];
feedback_data[0] = max_requested_brickmaps;
let feedback_data_u8 = bytemuck::cast_slice(&feedback_data);
let feedback_result_buffer = device.create_buffer(&wgpu::BufferDescriptor {
label: Some("Feedback Read"),
size: feedback_data_u8.len() as u64,
usage: wgpu::BufferUsages::COPY_DST | wgpu::BufferUsages::MAP_READ,
mapped_at_creation: false,
});
let mut brickgrid_upload_data = vec![0u32; 4 + 4 * max_uploaded_brickmaps as usize];
brickgrid_upload_data[0] = max_uploaded_brickmaps;
let brickgrid_staged = HashSet::new();
let mut brickmap_upload_data = vec![0u32; 4 + 532 * max_uploaded_brickmaps as usize];
brickmap_upload_data[0] = max_uploaded_brickmaps;
let brickmap_staged = Vec::new();
let mut buffers = gfx::BulkBufferBuilder::new()
.with_bytemuck_buffer("Brick World State", &[state_uniform])
.set_usage(wgpu::BufferUsages::STORAGE | wgpu::BufferUsages::COPY_DST)
.with_bytemuck_buffer("Brickgrid", &brickgrid)
.with_bytemuck_buffer("Brickmap Cache", &brickmap_cache)
.with_bytemuck_buffer("Shading Table", &shading_table)
.with_bytemuck_buffer("Brickgrid Unpack", &brickgrid_upload_data)
.with_bytemuck_buffer("Brickmap Unpack", &brickmap_upload_data)
.set_usage(
wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::COPY_DST
| wgpu::BufferUsages::COPY_SRC,
)
.with_buffer("Feedback", feedback_data_u8)
.build(context);
Self {
state_uniform,
brickgrid,
brickmap_cache_map,
brickmap_cache_idx: 0,
shading_table_allocator,
feedback_result_buffer,
unpack_max_count: max_uploaded_brickmaps as usize,
brickgrid_staged,
brickmap_staged,
state_buffer: buffers.remove(0),
brickgrid_buffer: buffers.remove(0),
brickmap_buffer: buffers.remove(0),
shading_table_buffer: buffers.remove(0),
brickgrid_unpack_buffer: buffers.remove(0),
brickmap_unpack_buffer: buffers.remove(0),
feedback_buffer: buffers.remove(0),
}
}
pub fn get_brickgrid_buffer(&self) -> &wgpu::Buffer {
&self.brickgrid_buffer
}
pub fn get_worldstate_buffer(&self) -> &wgpu::Buffer {
&self.state_buffer
}
pub fn get_brickmap_buffer(&self) -> &wgpu::Buffer {
&self.brickmap_buffer
}
pub fn get_shading_buffer(&self) -> &wgpu::Buffer {
&self.shading_table_buffer
}
pub fn get_feedback_buffer(&self) -> &wgpu::Buffer {
&self.feedback_buffer
}
pub fn get_feedback_result_buffer(&self) -> &wgpu::Buffer {
&self.feedback_result_buffer
}
pub fn get_brickmap_unpack_buffer(&self) -> &wgpu::Buffer {
&self.brickmap_unpack_buffer
}
pub fn get_brickgrid_unpack_buffer(&self) -> &wgpu::Buffer {
&self.brickgrid_unpack_buffer
}
pub fn get_unpack_max_count(&self) -> usize {
self.unpack_max_count
}
pub fn process_feedback_buffer(
&mut self,
context: &gfx::Context,
world: &mut super::world::WorldManager,
) {
// Get request count
let mut slice = self.feedback_result_buffer.slice(0..16);
slice.map_async(wgpu::MapMode::Read, |_| {});
context.device.poll(wgpu::Maintain::Wait);
let mut data: Vec<u32> = bytemuck::cast_slice(slice.get_mapped_range().as_ref()).to_vec();
self.feedback_result_buffer.unmap();
let request_count = data[1] as usize;
if request_count == 0 {
self.upload_unpack_buffers(context);
return;
}
// Get the position data
let range_end = 16 + 16 * request_count as u64;
slice = self.feedback_result_buffer.slice(16..range_end);
slice.map_async(wgpu::MapMode::Read, |_| {});
context.device.poll(wgpu::Maintain::Wait);
data = bytemuck::cast_slice(slice.get_mapped_range().as_ref()).to_vec();
self.feedback_result_buffer.unmap();
// Generate a sphere of voxels
let grid_dims = self.state_uniform.brickgrid_dims;
for i in 0..request_count {
// Extract brickgrid position of the requested brickmap
let grid_pos = glam::uvec3(data[i * 4], data[i * 4 + 1], data[i * 4 + 2]);
let grid_idx = math::to_1d_index(
grid_pos,
glam::uvec3(grid_dims[0], grid_dims[1], grid_dims[2]),
);
// We only want to upload voxels that are on the surface, so we cull anything
// that is surrounded by solid voxels
let grid_pos = grid_pos.as_ivec3();
let (bitmask_data, albedo_data) = Self::cull_interior_voxels(world, grid_pos);
// If there's no voxel colour data post-culling it means the brickmap is
// empty. We don't need to upload it, just mark the relevant brickgrid entry.
if albedo_data.is_empty() {
self.update_brickgrid_element(grid_idx, 0);
continue;
}
// Update the brickgrid index
self.update_brickgrid_element(
grid_idx,
Self::to_brickgrid_element(self.brickmap_cache_idx as u32, BrickgridFlag::Loaded),
);
// If there's already something in the cache spot we want to write to, we
// need to unload it.
if self.brickmap_cache_map[self.brickmap_cache_idx].is_some() {
let entry = self.brickmap_cache_map[self.brickmap_cache_idx].unwrap();
self.update_brickgrid_element(entry.grid_idx, 1);
}
// Update the shading table
let shading_idx = self
.shading_table_allocator
.try_alloc(albedo_data.len() as u32)
.unwrap() as usize;
// We're all good to overwrite the cache map entry now :)
self.brickmap_cache_map[self.brickmap_cache_idx] = Some(BrickmapCacheEntry {
grid_idx,
shading_table_offset: shading_idx as u32,
});
// Update the brickmap
let brickmap = Brickmap {
bitmask: bitmask_data,
shading_table_offset: shading_idx as u32,
lod_color: 0,
};
let shading_element_count = albedo_data.len();
let mut shading_elements = [0u32; 512];
shading_elements[..shading_element_count].copy_from_slice(&albedo_data);
let staged_brickmap = BrickmapUnpackElement {
cache_idx: self.brickmap_cache_idx as u32,
brickmap,
shading_element_count: shading_element_count as u32,
shading_elements,
};
self.brickmap_staged.push(staged_brickmap);
self.brickmap_cache_idx = (self.brickmap_cache_idx + 1) % self.brickmap_cache_map.len();
}
// Reset the request count on the gpu buffer
let data = &[0, 0, 0, 0];
context.queue.write_buffer(&self.feedback_buffer, 4, data);
self.upload_unpack_buffers(context);
// TODO: This is inaccurate if we've looped
log::info!("Num loaded brickmaps: {}", self.brickmap_cache_idx);
}
fn update_brickgrid_element(&mut self, index: usize, data: u32) {
// If we're updating a brickgrid element, we need to make sure to deallocate anything
// that's already there. The shading table gets deallocated, and the brickmap cache entry
// is marked as None.
if (self.brickgrid[index] & 0xF) == 4 {
let brickmap_idx = (self.brickgrid[index] >> 8) as usize;
let cache_map_entry = self.brickmap_cache_map[brickmap_idx];
match cache_map_entry {
Some(entry) => {
match self
.shading_table_allocator
.try_dealloc(entry.shading_table_offset)
{
Ok(_) => (),
Err(e) => log::warn!("{}", e),
}
self.brickmap_cache_map[brickmap_idx] = None;
}
None => log::warn!("Expected brickmap cache entry, found None!"),
}
}
// We're safe to overwrite the CPU brickgrid and mark for GPU upload now
self.brickgrid[index] = data;
self.brickgrid_staged.insert(index);
}
fn upload_unpack_buffers(&mut self, context: &gfx::Context) {
// Brickgrid
let mut data = Vec::new();
let mut iter = self.brickgrid_staged.iter();
let mut to_remove = Vec::new();
for _ in 0..self.unpack_max_count {
let el = iter.next();
if el.is_none() {
break;
}
let val = el.unwrap();
to_remove.push(*val as u32);
data.push(*val as u32);
data.push(self.brickgrid[*val]);
}
for val in &to_remove {
self.brickgrid_staged.remove(&(*val as usize));
}
if !data.is_empty() {
log::info!(
"Uploading {} brickgrid entries. ({} remaining)",
to_remove.len(),
self.brickgrid_staged.len()
);
}
context.queue.write_buffer(
&self.brickgrid_unpack_buffer,
4,
bytemuck::cast_slice(&[data.len()]),
);
context.queue.write_buffer(
&self.brickgrid_unpack_buffer,
16,
bytemuck::cast_slice(&data),
);
// Brickmap
let end = self.unpack_max_count.min(self.brickmap_staged.len());
let iter = self.brickmap_staged.drain(0..end);
let data = iter.as_slice();
context.queue.write_buffer(
&self.brickmap_unpack_buffer,
4,
bytemuck::cast_slice(&[end]),
);
context
.queue
.write_buffer(&self.brickmap_unpack_buffer, 16, bytemuck::cast_slice(data));
drop(iter);
if end > 0 {
log::info!(
"Uploading {} brickmap entries. ({} remaining)",
end,
self.brickmap_staged.len()
);
}
}
fn cull_interior_voxels(
world: &mut super::world::WorldManager,
grid_pos: glam::IVec3,
) -> ([u32; 16], Vec<u32>) {
// This is the data we want to return
let mut bitmask_data = [0xFFFFFFFF_u32; 16];
let mut albedo_data = Vec::<u32>::new();
// Calculate world chunk and block positions for each that may be accessed
let center_pos = Self::grid_pos_to_world_pos(world, grid_pos);
let forward_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(1, 0, 0));
let backward_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(-1, 0, 0));
let left_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(0, 0, -1));
let right_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(0, 0, 1));
let up_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(0, 1, 0));
let down_pos = Self::grid_pos_to_world_pos(world, grid_pos + glam::ivec3(0, -1, 0));
// Fetch those blocks
let center_block = world.get_block(center_pos.0, center_pos.1);
let forward_block = world.get_block(forward_pos.0, forward_pos.1);
let backward_block = world.get_block(backward_pos.0, backward_pos.1);
let left_block = world.get_block(left_pos.0, left_pos.1);
let right_block = world.get_block(right_pos.0, right_pos.1);
let up_block = world.get_block(up_pos.0, up_pos.1);
let down_block = world.get_block(down_pos.0, down_pos.1);
// Reusable array of whether cardinal neighbours are empty
let mut neighbours = [false; 6];
for z in 0..8 {
// Each z level contains two bitmask segments of voxels
let mut entry = 0u64;
for y in 0..8 {
for x in 0..8 {
// Ignore non-solids
let idx = x + y * 8 + z * 8 * 8;
let empty_voxel = super::world::Voxel::Empty;
match center_block[idx] {
super::world::Voxel::Empty => continue,
super::world::Voxel::Color(r, g, b) => {
// A voxel is on the surface if at least one of it's
// cardinal neighbours is non-solid.
neighbours[0] = if x == 7 {
forward_block[idx - 7] == empty_voxel
} else {
center_block[idx + 1] == empty_voxel
};
neighbours[1] = if x == 0 {
backward_block[idx + 7] == empty_voxel
} else {
center_block[idx - 1] == empty_voxel
};
neighbours[2] = if z == 7 {
right_block[idx - 448] == empty_voxel
} else {
center_block[idx + 64] == empty_voxel
};
neighbours[3] = if z == 0 {
left_block[idx + 448] == empty_voxel
} else {
center_block[idx - 64] == empty_voxel
};
neighbours[4] = if y == 7 {
up_block[idx - 56] == empty_voxel
} else {
center_block[idx + 8] == empty_voxel
};
neighbours[5] = if y == 0 {
down_block[idx + 56] == empty_voxel
} else {
center_block[idx - 8] == empty_voxel
};
// Set the appropriate bit in the z entry and add the
// shading data
let surface_voxel = neighbours.iter().any(|v| *v);
if surface_voxel {
entry += 1 << (x + y * 8);
let albedo = ((r as u32) << 24)
+ ((g as u32) << 16)
+ ((b as u32) << 8)
+ 255u32;
albedo_data.push(albedo);
}
}
}
}
}
let offset = 2 * z;
bitmask_data[offset] = (entry & 0xFFFFFFFF).try_into().unwrap();
bitmask_data[offset + 1] = ((entry >> 32) & 0xFFFFFFFF).try_into().unwrap();
}
(bitmask_data, albedo_data)
}
fn to_brickgrid_element(brickmap_cache_idx: u32, flags: BrickgridFlag) -> u32 {
(brickmap_cache_idx << 8) + flags as u32
}
fn grid_pos_to_world_pos(
world: &mut super::world::WorldManager,
grid_pos: glam::IVec3,
) -> (glam::IVec3, glam::UVec3) {
// We deal with dvecs here because we want a negative grid_pos to have floored
// chunk_pos
let chunk_dims = world.get_chunk_dims().as_dvec3();
let chunk_pos = (grid_pos.as_dvec3() / chunk_dims).floor();
let block_pos = grid_pos - (chunk_pos * chunk_dims).as_ivec3();
(chunk_pos.as_ivec3(), block_pos.as_uvec3())
}
}
#[derive(Debug)]
struct ShadingBucket {
global_offset: u32,
slot_count: u32,
slot_size: u32,
free: Vec<u32>,
used: Vec<u32>,
}
impl ShadingBucket {
fn new(global_offset: u32, slot_count: u32, slot_size: u32) -> Self {
let mut free = Vec::with_capacity(slot_count as usize);
for i in (0..slot_count).rev() {
free.push(i);
}
let used = Vec::with_capacity(slot_count as usize);
Self {
global_offset,
slot_count,
slot_size,
free,
used,
}
}
fn contains_address(&self, address: u32) -> bool {
let min = self.global_offset;
let max = min + self.slot_count * self.slot_size;
min <= address && address < max
}
fn try_alloc(&mut self) -> Option<u32> {
// Mark the first free index as used
let bucket_index = self.free.pop()?;
self.used.push(bucket_index);
// Convert the bucket index into a global address
Some(self.global_offset + bucket_index * self.slot_size)
}
fn try_dealloc(&mut self, address: u32) -> Result<(), String> {
log::trace!("Dealloc address: {}", address);
if !self.contains_address(address) {
let msg = format!("Address ({}) is not within bucket range.", address);
return Err(msg);
}
let local_address = address - self.global_offset;
if local_address % self.slot_size != 0 {
return Err("Address is not aligned to bucket element size.".to_string());
}
let bucket_index = local_address / self.slot_size;
if !self.used.contains(&bucket_index) {
return Err("Address is not currently allocated.".to_string());
}
// All the potential errors are out of the way, time to actually deallocate
let position = self.used.iter().position(|x| *x == bucket_index).unwrap();
self.used.swap_remove(position);
self.free.push(bucket_index);
Ok(())
}
}
#[derive(Debug)]
struct ShadingTableAllocator {
buckets: Vec<ShadingBucket>,
bucket_count: u32,
elements_per_bucket: u32,
total_elements: u32,
used_elements: u32,
}
impl ShadingTableAllocator {
fn new(bucket_count: u32, elements_per_bucket: u32) -> Self {
let total_elements = bucket_count * elements_per_bucket;
let used_elements = 0;
// Build the buckets. Ordered in ascending size
let mut buckets = Vec::with_capacity(bucket_count as usize);
for i in (0..bucket_count).rev() {
let global_offset = i * elements_per_bucket;
let slot_size = u32::pow(2, 9 - i);
let slot_count = elements_per_bucket / slot_size;
log::info!(
"Creating bucket: offset({}), slot_size({}), slot_count({})",
global_offset,
slot_size,
slot_count
);
buckets.push(ShadingBucket::new(global_offset, slot_count, slot_size));
}
Self {
buckets,
bucket_count,
elements_per_bucket,
total_elements,
used_elements,
}
}
fn try_alloc(&mut self, size: u32) -> Option<u32> {
for i in 0..self.bucket_count as usize {
let bucket = &mut self.buckets[i];
if bucket.slot_size < size {
continue;
}
let idx = bucket.try_alloc();
if idx.is_some() {
self.used_elements += bucket.slot_size;
log::trace!(
"Allocated to shader table at {}. {}/{} ({}%)",
idx.unwrap(),
self.used_elements,
self.total_elements,
((self.used_elements as f32 / self.total_elements as f32) * 100.0).floor()
);
return idx;
}
}
None
}
fn try_dealloc(&mut self, address: u32) -> Result<(), String> {
// Buckets are reverse order of their global offset so we need to reverse our idx
let mut bucket_idx = address / self.elements_per_bucket;
bucket_idx = self.bucket_count - bucket_idx - 1;
let bucket = &mut self.buckets[bucket_idx as usize];
self.used_elements -= bucket.slot_size;
bucket.try_dealloc(address)
}
}