Add DHAT to vello cpu example

Cargo.toml

@@ -156,3 +156,8 @@ syn = { version = "2.0.101", features = ["full", "extra-traits"] }
 quote = "1.0.40"
 serde = { version = "1.0.225", default-features = false }
 serde_json = "1.0.145"
+
+# Profiling profile: release optimizations with debug info for demangled backtraces.
+[profile.profiling]
+inherits = "release"
+debug = 1

sparse_strips/vello_cpu/examples/winit/Cargo.toml

@@ -9,9 +9,13 @@ publish = false
 [lints]
 workspace = true
 
+[features]
+dhat-heap = ["dep:dhat"]
+
 [dependencies]
 winit = { workspace = true }
 vello_common = { workspace = true }
 vello_cpu = { workspace = true, features = ["multithreading"] }
 vello_example_scenes = { workspace = true, features = ["cpu"] }
 softbuffer = "0.4.6"
+dhat = { version = "0.3.3", optional = true }

sparse_strips/vello_cpu/examples/winit/README.md

@@ -37,3 +37,33 @@ This example uses multi-threaded CPU rendering. The number of threads used is de
 
 For better performance, always run with the `--release` flag.
 
+## Heap Profiling with DHAT
+
+This example supports heap profiling using the [dhat](https://docs.rs/dhat) crate. This is useful for analyzing memory allocation patterns and identifying allocation hotspots.
+
+### Running with Heap Profiling
+
+```sh
+cargo run -p vello_cpu_winit --profile profiling --features dhat-heap
+```
+
+The program will run slower than normal due to allocation tracking.
+
+When you exit the application, you'll see a summary like:
+
+```
+dhat: Total:     1,256 bytes in 6 blocks
+dhat: At t-gmax: 1,256 bytes in 6 blocks
+dhat: At t-end:  1,256 bytes in 6 blocks
+dhat: The data has been saved to dhat-heap.json, and is viewable with dhat/dh_view.html
+```
+
+### Viewing Results
+
+Open the generated `dhat-heap.json` file in [DHAT's online viewer](https://nnethercote.github.io/dh_view/dh_view.html) to explore:
+
+- Total allocations and their call sites
+- Peak heap usage (t-gmax)
+- Memory still allocated at exit (t-end)
+- Allocation hotspots with full backtraces
+

sparse_strips/vello_cpu/examples/winit/src/main.rs

@@ -8,6 +8,10 @@
     reason = "truncation has no appreciable impact in this demo"
 )]
 
+#[cfg(feature = "dhat-heap")]
+#[global_allocator]
+static ALLOC: dhat::Alloc = dhat::Alloc;
+
 #[cfg(not(target_arch = "wasm32"))]
 use std::env;
 use std::num::NonZeroU32;
@@ -55,6 +59,9 @@ struct App {
 }
 
 fn main() {
+    #[cfg(feature = "dhat-heap")]
+    let _profiler = dhat::Profiler::new_heap();
+
     #[cfg(not(target_arch = "wasm32"))]
     let (scenes, start_scene_index) = {
         let mut start_scene_index = 0;