diff --git a/examples/common/boilerplate.rs b/examples/common/boilerplate.rs
index b6bf25e68e..9f70707fa1 100644
--- a/examples/common/boilerplate.rs
+++ b/examples/common/boilerplate.rs
@@ -164,7 +164,6 @@ pub fn main_wrapper<E: Example>(
         precache_flags: E::PRECACHE_SHADER_FLAGS,
         device_pixel_ratio,
         clear_color: Some(ColorF::new(0.3, 0.0, 0.0, 1.0)),
-        //scatter_gpu_cache_updates: false,
         debug_flags,
         //allow_texture_swizzling: false,
         ..options.unwrap_or(webrender::RendererOptions::default())
diff --git a/webrender/src/device/gl.rs b/webrender/src/device/gl.rs
index 1bfad787c1..f77bbe388a 100644
--- a/webrender/src/device/gl.rs
+++ b/webrender/src/device/gl.rs
@@ -790,7 +790,7 @@ impl ProgramSourceInfo {
         let source_and_digest = SHADERS.get(&name).expect("Shader not found");
 
         // Hash the renderer name.
-        hasher.write(device.renderer_name.as_bytes());
+        hasher.write(device.capabilities.renderer_name.as_bytes());
 
         // Hash the prefix string.
         build_shader_prefix_string(
@@ -999,6 +999,8 @@ pub struct Capabilities {
     pub supports_nonzero_pbo_offsets: bool,
     /// Whether the driver supports specifying the texture usage up front.
     pub supports_texture_usage: bool,
+    /// The name of the renderer, as reported by GL
+    pub renderer_name: String,
 }
 
 #[derive(Clone, Debug)]
@@ -1083,7 +1085,6 @@ pub struct Device {
 
     max_texture_size: i32,
     max_texture_layers: u32,
-    renderer_name: String,
     cached_programs: Option<Rc<ProgramCache>>,
 
     // Frame counter. This is used to map between CPU
@@ -1554,6 +1555,7 @@ impl Device {
                 supports_texture_swizzle,
                 supports_nonzero_pbo_offsets,
                 supports_texture_usage,
+                renderer_name,
             },
 
             color_formats,
@@ -1578,7 +1580,6 @@ impl Device {
 
             max_texture_size,
             max_texture_layers,
-            renderer_name,
             cached_programs,
             frame_id: GpuFrameId(0),
             extensions,
@@ -1998,7 +1999,7 @@ impl Device {
                     error!(
                       "Failed to load a program object with a program binary: {} renderer {}\n{}",
                       &info.base_filename,
-                      self.renderer_name,
+                      self.capabilities.renderer_name,
                       error_log
                     );
                     if let Some(ref program_cache_handler) = cached_programs.program_cache_handler {
diff --git a/webrender/src/renderer.rs b/webrender/src/renderer.rs
index 49becbd30d..281b71a2d1 100644
--- a/webrender/src/renderer.rs
+++ b/webrender/src/renderer.rs
@@ -2231,9 +2231,19 @@ impl Renderer {
         let transforms_texture = VertexDataTexture::new(&mut device, ImageFormat::RGBAF32);
         let render_task_texture = VertexDataTexture::new(&mut device, ImageFormat::RGBAF32);
 
+        // On some (mostly older, integrated) GPUs, the normal GPU texture cache update path
+        // doesn't work well when running on ANGLE, causing CPU stalls inside D3D and/or the
+        // GPU driver. See https://bugzilla.mozilla.org/show_bug.cgi?id=1576637 for much
+        // more detail. To reduce the number of code paths we have active that require testing,
+        // we will enable the GPU cache scatter update path on all devices running with ANGLE.
+        // We want a better solution long-term, but for now this is a significant performance
+        // improvement on HD4600 era GPUs, and shouldn't hurt performance in a noticeable
+        // way on other systems running under ANGLE.
+        let is_angle = device.get_capabilities().renderer_name.contains("ANGLE");
+
         let gpu_cache_texture = GpuCacheTexture::new(
             &mut device,
-            options.scatter_gpu_cache_updates,
+            is_angle,
         )?;
 
         device.end_frame();
@@ -6608,7 +6618,6 @@ pub struct RendererOptions {
     pub enable_clear_scissor: bool,
     pub max_texture_size: Option<i32>,
     pub max_glyph_cache_size: Option<usize>,
-    pub scatter_gpu_cache_updates: bool,
     pub upload_method: UploadMethod,
     pub workers: Option<Arc<ThreadPool>>,
     pub enable_multithreading: bool,
@@ -6679,8 +6688,6 @@ impl Default for RendererOptions {
             enable_clear_scissor: true,
             max_texture_size: None,
             max_glyph_cache_size: None,
-            // Scattered GPU cache updates haven't met a test that would show their superiority yet.
-            scatter_gpu_cache_updates: false,
             // This is best as `Immediate` on Angle, or `Pixelbuffer(Dynamic)` on GL,
             // but we are unable to make this decision here, so picking the reasonable medium.
             upload_method: UploadMethod::PixelBuffer(VertexUsageHint::Stream),