| Commit message (Collapse) | Author | Age |
| |\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
live-coding example together
A merge of sk-text (korlovpv), gpu-s4b (nqwsztut) and demo1-b (oroqkmvu), no conflicts, built beside cloud9's c9-cli. zig build test: 192/192 steps, 151/151 tests; test-text: 52/52 steps, 4/4 tests.
Co-Authored-By: Claude Fable 5.2 <[email protected]>
Claude-Session: https://claude.ai/code/session_016Y9MNBECMWcJD5pdS4SR6T
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Moved from pardes's src/gui/gui.zig (step S4b, the last of moving its
renderer here): begin and submit acquire a frame's command buffer and
give a frame up; mapCells grows and maps the cell instance buffer for
the embedder to write in place; uploadCells sends it after the atlas's
dirty rows; record makes the frame's render pass, cleared to the
embedder's colour, and draws its groups (Group: cells, then pictures,
then decor, each a run of instances), a scroll's shifted rows clipped to
the rectangle given, then the overlay. uploadAtlas, drawImages,
pushLayerOpacity and drawOverlay are Cells' own now.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Moved from pardes's src/gui/gui.zig (step S4a of moving its renderer
here): Cells is now a struct over the embedder's device and swapchain
format holding the four pipelines (and the addon one, made on first
use), the glyph atlas's texture, nearest sampler and upload buffer, the
pictures' linear sampler, the cells', pictures' and overlay's buffers,
and the pictures' textures by key. init and deinit make and release
them; image uploads a picture's RGBA, or only its dirty rows into the
texture of the picture it replaces; mapImages, uploadImages and
drawImages carry the frame's picture instances and draws, an
ImageDraw's texture either one of Cells' own or the caller's, borrowed
(an addon's premultiplied output); uploadAtlas sends the Atlas's dirty
rows; uploadOverlay, pushLayerOpacity and drawOverlay the overlay's
triangles. The SPIR-V is the embedder's `<name>.spv` imports; Cells.zig
imports the Atlas as `font_atlas`.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Moved from pardes's src/gui/gui.zig (step S3 of moving its renderer
here): src/gpu/Present.zig, the device, the window's claim or a software
presenter through SDL_Renderer (a transparent window, which the GPU API
will not claim, or a claim refused), the present mode and frames in
flight, detach and attach for a window closed and opened again, the
frame's target (the swapchain's image, or a capture texture in test mode
and for the soft presenter), the soft present (the frame read back, put
in a streaming texture and flipped) and a captured frame's read back.
The window stays the embedder's and is passed in where it is used. What
pardes did inside these with its own state comes back instead: a soft
present's phase times (Timings), each presented frame's pixels through a
frame tap, a captured frame's mapped bytes and times (readFrame, then
releaseFrame) and their byte order (bgr). Present.zig imports the
embedder's SDL translation as `c` and Cells.zig as `gpu_cells`.
|
| |\| |
| |/
|/|
| |
| |
| |
| | |
sketchbook-next with gpu-s2 (vendor/font.{c,h}, src/font/Atlas.zig, src/gpu/Cells.zig, src/gpu/shaders/): no conflicts. Nothing of sketchbook's own uses them yet.
Co-Authored-By: Claude Fable 5.2 <[email protected]>
Claude-Session: https://claude.ai/code/session_016Y9MNBECMWcJD5pdS4SR6T
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Moved verbatim from pardes's src/gui/gui.zig (step S1 of moving its
renderer here): the seven GLSL shaders and their committed SPIR-V under
src/gpu/shaders/, and src/gpu/Cells.zig with the instance structs the
shaders read, the vertex inputs, makePipeline, the layer blend, texture2d
and growInstances (which takes the device instead of pardes's Gui).
Cells.zig imports the embedder's own SDL translation as `c`; pardes
compiles it from this checkout. `zig build shaders` refreshes the
prebuilt SPIR-V.
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A compute shader's output was heard only after the sketch read it back
(f.download, a frame later) and pushed it behind a lead of its own
choosing; spheres kept 100 ms, and a change it made left for the device
96 ms later on average (measured, dummy driver, 60 Hz).
f.audioBuffer(key, buffer, .{ .count, .offset }) streams `count` f32 a
shader wrote: the host copies them back while it runs the frame's
commands, waiting on the copy's fence (Executor.read: 0.2 ms on average
here, 2 ms at the 99th percentile on a loaded machine), and queues them
on the stream, so nothing is in flight between frames and the sketch
keeps no books. Input.lead says how far ahead to make sound: the slowest
of the last 64 frames' worth of samples (timed between the ends of the
frames' commands, where the pushes are), plus the device's pull (what
SDL takes from a stream at once, read from the device: 1024 here), plus
half a pull for each underrun, a sample less each frame after; capped at
8192. A sketch pushes `in.lead -| in.stream(key).queued` a frame, and
its change leaves after about what is queued ahead of it, a pull: 19 to
27 ms on average with SDL's 1024-sample pulls, 6.3 to 6.5 ms with 256
(SDL_AUDIO_DEVICE_SAMPLE_FRAMES; 17 under heavy load). Stream.underruns
counts the pushes that found a stream empty though the frame before had
fed it (a gap).
The voices keep their 2048 samples: mixed to the lead, a one-shot longer
than it is mixed over several frames, and voices.script's expect-audio
and expect-no-audio, which look one frame at a time, failed.
On the web, audio.buffer reads the buffer back with mapAsync and queues
it on arrival, in order, counting it as queued meanwhile; in.lead stays
at 100 ms there.
The measurements (a probe patched into the host for the purpose, kept
out of the tree) are in the 02-pardes-code scratch, .scratch/sketch/gpusynth.
Tests: the lead's arithmetic (audio.zig, in test-audio's filter); spheres'
play script checks its stream is fed. zig build test: 164/164 steps,
49/49 tests run (the rest cached); zig build test-audio: 48/48 steps, with
the voices sketch's wasm, so web.zig compiles; voices.script passes four
runs in four. web/host.js: node --check only, no browser run.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_016Y9MNBECMWcJD5pdS4SR6T
|
| |
|
|
|
|
|
|
|
|
| |
executor pardes shares
`f.postProcess(source, target, .{...})` records a fullscreen pass that samples a completed render into a distinct target image: colour at fragment binding 0 and, with `sample_depth`, depth at 1, which needs the source's latest pass to have depth and one sample (missing depth, multisampled depth and feedback loops are rejected). The scene format's version 4 carries it, the executor records it on the caller's command buffer and owns the sampler, the web host does the same with its texture and sampler pairs, and native profiling reports the work as `postprocess`. The README says how to use it.
docs/pardes-rendering.md describes the executor as pardes uses it for addon scenes: a borrowed SDL GPU device, `record` into the caller's command buffer, `run`, `output` and `postProcess`.
The working copy as it stood on 2026-11-16; message written by Claude from the diff, at the author's request.
|
| |
|