Prepare benchmarks + ~1.7x prepare perf improvement (#121)

* Add prepare benchmarks

* Skip unnecessary peaks

* Cite sample sources

benches/state.rs

@@ -0,0 +1,35 @@
+use pollster::block_on;
+
+pub struct State {
+    pub device: wgpu::Device,
+    pub queue: wgpu::Queue,
+}
+
+impl State {
+    pub fn new() -> Self {
+        let instance = wgpu::Instance::new(wgpu::InstanceDescriptor {
+            backends: wgpu::Backends::all(),
+            flags: wgpu::InstanceFlags::empty(),
+            dx12_shader_compiler: wgpu::Dx12Compiler::Fxc,
+            gles_minor_version: wgpu::Gles3MinorVersion::Automatic,
+        });
+
+        let adapter = block_on(wgpu::util::initialize_adapter_from_env_or_default(
+            &instance, None,
+        ))
+        .unwrap();
+
+        let (device, queue) = block_on(adapter.request_device(
+            &wgpu::DeviceDescriptor {
+                label: Some("Benchmark Device"),
+                required_features: adapter.features(),
+                required_limits: adapter.limits(),
+                memory_hints: wgpu::MemoryHints::Performance,
+            },
+            None,
+        ))
+        .unwrap();
+
+        Self { device, queue }
+    }
+}