makepad/platform/script/audio_aot/shaders
2026-09-29 17:21:25 +02:00
..
allpass_reverb.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
fm2_svf.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
fm4.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
pluck.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
saw_svf.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
waveshaper4x.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00
wavetable_pad.splash script: audio_aot, realtime audio shaders compiled from Splash -- a new crate (makepad-script-audio-aot) that compiles a subset of Splash with logic and dataflow (control flow, loops with bounded iteration, integers and booleans, fixed arrays and structs, per-voice state that persists across samples, delay lines and ring buffers, interpolated table reads, a seeded noise source, audio-rate, control-rate and event ports, explicit feedback delays) into an SSA IR with a reference interpreter and a native ARM64 backend, fused per graph and run per block with no allocation, locks or unbounded time on the audio thread; edits compile on a worker and swap at a block boundary. 64 voices of 2-op FM plus an SVF at 48 kHz take 7.4% of one core natively (the interpreter alone would need 178%). math_aot's plan of record (MATH_AOT.md) joins the tree with a pointer to it. Checked: cargo test --release -p makepad-script-audio-aot (16). 2026-09-29 17:21:25 +02:00