Cache Vello render output as stitchable textures#3722
Cache Vello render output as stitchable textures#3722TrueDoctor wants to merge 15 commits intomasterfrom
Conversation
Performance Benchmark Results
|
Performance Benchmark Results
|
TrueDoctor
force-pushed
the
cache-render-output
branch
from
c8cf08b to
01b4c1f
Compare
Performance Benchmark Results
|
TrueDoctor
force-pushed
the
cache-render-output
branch
from
01b4c1f to
f79c30c
Compare
TrueDoctor
force-pushed
the
cache-render-output
branch
from
b8a716a to
6950a74
Compare
Keavon
changed the title
|
We don't store tiles, so I think the title is technically wrong / misleading |
Keavon
changed the title
TrueDoctor
force-pushed
the
cache-render-output
branch
2 times, most recently
from
1b42423 to
0af8665
Compare
TrueDoctor
force-pushed
the
cache-render-output
branch
from
0af8665 to
722d379
Compare
Keavon
force-pushed
the
cache-render-output
branch
from
865572b to
74332a5
Compare
timon-schelling
left a comment
timon-schelling
left a comment
There was a problem hiding this comment.
Generally LGTM, didn't check the cache as thoroughly. Can do that if you want but the approach is good.
|
|
||
| rows.extend_from_slice(&[header, node_graph_wires_label, graph_wire_style, use_vello, brush_tool]); | ||
| let max_region_size_description = | ||
| "Maximum render region size in total pixels (width × height). Larger values require fewer render passes. If you see rendering artifacts or nothing at all, configure a smaller value."; |
There was a problem hiding this comment.
Can we split this explanation in two parts, first explaining for non technical users and a second part going deeper and explaining the exact effect?
|
|
||
| executor.context.queue.submit([encoder.finish()]); | ||
| surface_texture.present(); | ||
| // image_texture.texture.destroy(); |
There was a problem hiding this comment.
| // image_texture.texture.destroy(); |
| context_features: graphene_std::ContextDependencies { | ||
| extract: ContextFeatures::empty(), | ||
| // We add the extract index annotation here to force the compiler to add a context nullification node before this node so the render context is properly nullified so the render cache node can do its's work | ||
| extract: ContextFeatures::INDEX, |
There was a problem hiding this comment.
could we add a separate ContextFeature for this purpose?
| } | ||
|
|
||
| #[inline] | ||
| pub fn tile_world_start(tile: &TileCoord, scale: f64) -> DVec2 { |
There was a problem hiding this comment.
| pub fn tile_world_start(tile: &TileCoord, scale: f64) -> DVec2 { | |
| pub fn tile_space_origin(tile: &TileCoord, scale: f64) -> DVec2 { |
There was a problem hiding this comment.
not sure, just something to think about
| use crate::render_node::RenderOutputType; | ||
|
|
||
| pub const TILE_SIZE: u32 = 256; | ||
| pub const MAX_CACHE_MEMORY_BYTES: usize = 512 * 1024 * 1024; |
There was a problem hiding this comment.
Maybe this could be a user setting?
| let queue = &exec.context.queue; | ||
|
|
||
| // TODO: Use texture pool to reuse existing unused textures instead of allocating fresh ones every time | ||
| let output_texture = device.create_texture(&wgpu::TextureDescriptor { |
There was a problem hiding this comment.
Could we use target texture here? Only needs to be recreated if viewport size changes?
2 participants
Cache render output as texture and only rerender necessary parts while panning
Closes: #3509, part of #1607.