diff --git a/impeller/renderer/backend/metal/surface_mtl.mm b/impeller/renderer/backend/metal/surface_mtl.mm
index 117ef36ab13e9..b84f3a144058d 100644
--- a/impeller/renderer/backend/metal/surface_mtl.mm
+++ b/impeller/renderer/backend/metal/surface_mtl.mm
@@ -266,14 +266,29 @@ - (void)flutterPrepareForPresent:(nonnull id<MTLCommandBuffer>)commandBuffer;
           flutterPrepareForPresent:command_buffer];
     }
 
+    // Intel iOS simulators do not seem to give backpressure on Metal drawable
+    // aquisition, which can result in Impeller running head of the GPU
+    // workload by dozens of frames. Slow this process down by blocking
+    // on submit until the last command buffer is at least scheduled.
+#if defined(FML_OS_IOS_SIMULATOR) && defined(FML_ARCH_CPU_X86_64)
+    constexpr bool alwaysWaitForScheduling = true;
+#else
+    constexpr bool alwaysWaitForScheduling = false;
+#endif  // defined(FML_OS_IOS_SIMULATOR) && defined(FML_ARCH_CPU_X86_64)
+
     // If the threads have been merged, or there is a pending frame capture,
     // then block on cmd buffer scheduling to ensure that the
     // transaction/capture work correctly.
     if ([[NSThread currentThread] isMainThread] ||
-        [[MTLCaptureManager sharedCaptureManager] isCapturing]) {
+        [[MTLCaptureManager sharedCaptureManager] isCapturing] ||
+        alwaysWaitForScheduling) {
       TRACE_EVENT0("flutter", "waitUntilScheduled");
       [command_buffer commit];
+#if defined(FML_OS_IOS_SIMULATOR) && defined(FML_ARCH_CPU_X86_64)
+      [command_buffer waitUntilCompleted];
+#else
       [command_buffer waitUntilScheduled];
+#endif  // defined(FML_OS_IOS_SIMULATOR) && defined(FML_ARCH_CPU_X86_64)
       [drawable_ present];
     } else {
       // The drawable may come from a FlutterMetalLayer, so it can't be