Providing a response and feedback to this: https://origin-devforums.apple.com/forums/thread/827311 and https://developer.apple.com/forums/thread/827043?page=1#889020022 I have a created a feedback/bug report with ID FB22823733 Feedback Report: I have created a feedback report as well like recommended with this ID: FB22823733, with more elaborate images of my implementation, also see here More clarity: Yes, this is for an iOS app(iOS 18+) but testing and debugging on Xcode(MacOS 15.7.7, M4 24GB), which traps or hangs on getting to [PhotogrammetrySample] even while using the lazy sequence and the contentsOf as specified in your docs. I also tried using the PhotogrammetrySession folder run but its still failed with : CoreOC.PhotogrammetrySession.Error 6 and I understand to mean alignment failed. What can be done or do you guys expose any functions that can be used to aid, or handle these internally, can't find any internally. The ObjectMasks are actually segmentation masks from an ML algorithm . To replicate try calling PhotogrammetrySession(input: [PhotogrammetrySample]) with contentsOf as captured on your documentation, even with like 30 image set or is there something I'm missing.

Xcode hangs when I call PhotogrammetrySession(input: [PhotogrammetrySample]) with objectMask set and traps on some devices, see the attached screenshot, it gets to the function and hangs. Even the folder reconstruction also doesn't complete as it can't find alignment and displays the CoreOC.PhotogrammetrySession.Error 6 and I understand to mean alignment failed. In this case it failed while object masking was ON, so RealityKit could not find enough consistent feature tracks inside the masked pixels across the image set. What can be done or do you guys expose any functions that can be used to aid, or handle these internally, can't find any internally. The ObjectMasks are actually segmentation masks from an ML algorithm . To replicate try calling PhotogrammetrySession(input: [PhotogrammetrySample]) with contentsOf as captured on your documentation, even with like 30 image set or is there something I'm missing. I will appreciate a timely response and willing to provide more clarity and informations, thank you so much for your understanding

Our application is trying to read all resolutions of an external monitor. We have observed that, for the external monitor there is a mismatch in resolution list in our application and the resolution list in system settings. We are using the apple API "CGDisplayCopyAllDisplayModes" to read the resolutions.

Metal GPU Driver Crash on M5 Pro + macOS 26.5 — kIOGPUCommandBufferCallbackErrorOutOfMemory with <2GB working sets Summary The Metal driver AGXMetalG17X 351.2 on macOS 26.5 (25F71) for the M5 Pro chip crashes with kIOGPUCommandBufferCallbackErrorOutOfMemory (00000008) when running LLM inference workloads with working sets as small as ~1.5GB, despite 24GB of unified memory being available and Apple Diagnostics confirming the hardware is fully functional. This affects multiple tools: MLX, llama.cpp (Metal backend), and native apps using Metal for inference. System Component Value Model MacBook Pro (Mac17,9) Chip Apple M5 Pro (applegpu_g17s) GPU Cores 16 RAM 24 GB LPDDR5 macOS 26.5 (25F71) Metal Metal 4 GPU Driver AGXMetalG17X 351.2 Xcode 26.5 (17F42) Reproduction MLX (Python) pip install mlx mlx-lm python -m mlx_lm.generate \ --model mlx-community/Qwen2.5-3B-Instruct-4bit \ --max-tokens 10 \ --prompt "Hello" Expected: Normal text generation Actual: Crash with: libc++abi: terminating due to uncaught exception of type std::runtime_error: [METAL] Command buffer execution failed: Insufficient Memory (00000008:kIOGPUCommandBufferCallbackErrorOutOfMemory) llama.cpp brew install llama.cpp llama-cli --model model.gguf --prompt "Hello" --n-predict 20 --n-gpu-layers 99 Expected: Fast GPU generation Actual: Process hangs indefinitely Test Results Tool Model Peak Memory Result MLX Qwen2.5-0.5B-4bit 0.36 GB ✅ Works MLX Qwen2.5-1.5B-4bit 0.98 GB ✅ Works MLX Qwen3-1.7B-4bit 1.01 GB ✅ Works MLX Qwen2.5-3B-4bit ~1.5 GB ❌ Metal OOM crash MLX Qwen3-4B-4bit ~2.1 GB ❌ Metal OOM crash MLX Qwen3-8B-4bit ~4.5 GB ❌ Metal OOM crash llama.cpp Qwen2.5-0.5B GGUF ~0.5 GB ❌ Hangs with GPU llama.cpp Qwen2.5-0.5B GGUF ~0.5 GB ✅ Works with CPU only Key Evidence Hardware is healthy — Apple Diagnostics passed all tests Basic Metal works — matmul, array ops work fine CPU inference works — llama.cpp with -ngl 0 runs correctly The error is NOT about actual memory exhaustion — kIOGPUCommandBufferCallbackErrorOutOfMemory means the kernel rejects the Metal memory commit, not that physical memory is full. The system reports 17.76GB available for Metal working set. Crash Log Extract Thread 31 Crashed: 0 libsystem_kernel.dylib __pthread_kill + 8 1 libsystem_pthread.dylib pthread_kill + 296 2 libsystem_c.dylib abort + 148 3 Metal MTLReportFailure.cold.1 + 48 4 Metal MTLReportFailure + 576 5 Metal -[_MTLCommandBuffer addCompletedHandler:] + 104 ... Exception Type: EXC_CRASH (SIGABRT) Termination Reason: Namespace SIGNAL, Code 6, Abort trap: 6 Related Issues ml-explore/mlx#3586 — Metal compiler regression on macOS 26.5 ml-explore/mlx#3534 — M5 float32 precision issue ml-explore/mlx#3568 — M5 random divergence ml-explore/mlx#3539 — Metal residency OOM (M4 Max) Request Please investigate the AGXMetalG17X driver for M5 Pro on macOS 26.5. The driver appears to incorrectly reject Metal memory commits for LLM inference workloads, even when the working set is well within the system's reported limits (1.5GB requested vs 17.76GB available). Happy to provide full crash logs, sysdiagnose archives, or run additional tests.

One of my users updated his iPhone 16 Pro to iOS 26.5, and Apple TV to tvOS 26.5, both of them unable to pair a Microsoft Xbox One (Model 1708) gamepad, before updating (updated from iOS 18.x) it works normally. He also got an old iPhone on iOS 16.7 which was working fine with that gamepad. Any one met the same issue? feedback no. FB22809309

Hi all, I'm working on updating my audio visualizer app. I'm adding new visualizers based on Metal 4 compute shaders. They worked in iOS 26.4 and iOS 26.5 up until beta 3. However, after that, the visualizers started crashing the phone and forcing a restart. On the latest version of iOS 26.5, the crash is still there. I submitted feedback, but haven't heard anything back just yet. I was wondering if others have faced this same issue, and if there are any workarounds. Here is my repo if you want to look at the code (forgive me if it's sloppy, I'm quite new to graphics programming and Metal): https://github.com/aabagdi/VisualMan/tree/main Thank you!

In my turn-based game, I receive GKListener event receivedTurnEventForMatch and decode the match.matchData. On occasion, the matchData is clearly stale and is from the previous turn. If I call the MatchMaker ViewController up and select that same match, the data is not stale, so it's not a matter of not calling endTurn. I have tried both loadMatchWithID and loadMatchesWithCompletionHandler after receiving the receivedTurnEventForMatch, but the data is still stale. Advice?

Hi, Can Apple provide some reassurance on the continued support for SpriteKit in the coming years? At WWDC 2025, Apple deprecated SceneKit, which was alarming for a SpriteKit fan and user. I'm aware Apple doesn't comment on future plans. However, committing to a specific framework for larger projects is a significant investment in time (and money). Developers need to know whether to continue investing in proprietary technology that could be deprecated at any time. Some reassurance regarding SpriteKit's future would go a long way. I have moved my project from SpriteKit to RealityKit. RealityKit is certainly interesting, but I'd gladly switch back to SpriteKit. SpriteKit has things going for it that no other Apple framework provides: Live drawing of shapes, paths, text, video, and images at 120 FPS across all Apple devices. RealityKit offers no procedural drawing API equivalent to shape nodes. Particles and physics fields are very easy and fun to play with. The rigid body engine has a good feel. By comparison, RealityKit sleeps too aggressively, and doesn't have spring joints. SpriteKit is lightweight and runs well on older devices. In an ideal world, I'd also love to see SpriteKit improved: A deterministic physics engine Metal shader support Soft shadows for lights One can dream. Thank you!

Hi, I have a game which is available on iOS and tvOS and the gameplay is basically the same on both platforms. I'd like to enable my players to seamlessly continue their play sessions on all available platforms. Unfortunately, the new GameSave framework is not available on tvOS and neither is the old GKSavedGame API. I'm asking you to enable the new GameSave framework in a future version of tvOS. Thank you. I've also reported via Feedback Assistant with Feedback Number: FB22791051

Hi, I'm implementing the new iOS 26 GameSave API in my game and I'm seeing an unexpected behavior when I try to handle custom conflict resolution. Here's what happens: directoryState.conflictedVersions is nil when directoryState.state is GSSyncStateConflicted. According to the API docs, this should never happen because there is no way to resolve the conflict. Here's the setup: I have 2 devices. Device 1 - iPad 10G iOS 26.5, and Device 2 - iPhone SE 3G iOS 26.5 and I run the same code on both devices. I disable the WiFi on device 2 to create a conflict I open the directory using [GSSyncedDirectory openDirectoryForContainerIdentifier: nil] Run [dir finishSyncingWithCompletionHandler:] After the callback returns I check dir.directoryState.state If the state is GSSyncStateReady, GSSyncStateOffline, or GSSyncStateLocal I write some data in a file and close the dir. I re-enable the WiFi on Device 2 and run the same code Now, on Device 2 dir.directoryState.state is GSSyncStateConflicted, as expected, but dir.directoryState.conflictedVersions is NIL and there is no way to continue and resolve the conflict. I'm expecting dir.directoryState.conflictedVersions to contain an array of GSSyncedDirectoryVersion objects as described in the API docs. One interesting bit is that if I replace [dir finishSyncingWithCompletionHandler:] with [dir finishSyncing: window completionHandler:] the UI conflict resolution works as expected. I'm using Objective C to call the API, if that matters. I've also reported via Feedback Assistant with Feedback Number: FB22790831

The CGSetDisplayTransferByTable() is not working on the latest round of Mac hardware, namely the MacBook Neo (external display), MacBook M5 Pro (both built-in and external display) and possibly the M5 Max. All tested apps (BetterDisplay, MonitorControl, f.lux, Lunar) exhibit the very issue both in macOS Tahoe 26.3 and macOS Tahoe 26.4 RC. Tested on multiple Macs and installations on the MacBook Neo and MacBook M5 Pro. This issue breaks several display related macOS apps. Way to reproduce the issue using an affected app: Install the app BetterDisplay (https://betterdisplay.pro) Launch the app, open the app menu, choose Image Adjustments and try to adjust colors. Adjustments take no effect Way to reproduce the issue programmatically: Attempt to use the affected macOS API feature: https://developer.apple.com/documentation/coregraphics/cgsetdisplaytransferbytable(::::_:) Here are the FB numbers: FB22273730 (Filed this one as a developer on an unaffected MBP M3 Max) FB22273782 (Filed from an affected MBP M5 Pro running 26.4 RC, with debug info attached)

I recompiled my turn-based app to the (now mandatory) SDK 26 (Xcode 26.4 to be exact), and Game Center is no longer sending notifications of turn changes or game invites to either my iPad Air running 15.8.8 or my iPhone SE3 running 26.4.2.

I have noticed that the performance drop on SpriteKit-based projects running on iOS 26 is still ongoing With iOS 26 back in Sep 2025 a framerate problem was introduced. My app was always running smoothly with 60fps even on very old devices suddenly started to stutter with 40fps - and lower on a rather normal iPhone 13. This problem continued with BETA 26.1 The problem was fixed in 26.2. But 26.3 brought the problem back and its still ongoing with 26.4 of yesterday This is easily reproducible with a very simple example // // BareboneSpriteKitApp.swift // BareboneSpriteKit // // Created by Bernd Beyreuther on 24.02.26. // import SwiftUI import SpriteKit @main struct BareboneSpriteKitApp: App { var body: some Scene { WindowGroup { BareboneSceneView() } } } final class BareboneScene: SKScene { override func didMove(to view: SKView) { size = view.bounds.size scaleMode = .resizeFill anchorPoint = CGPoint(x: 0.5, y: 0.5) backgroundColor = .darkGray let s = SKSpriteNode(color: .cyan, size: CGSize(width: 64, height: 64)) addChild(s) let action = SKAction.rotate(byAngle: .pi, duration: 2) s.run(.repeatForever(action)) let t = SKLabelNode(text: deviceInfoString()) t.fontSize = 15 t.position.y = -100 addChild(t) } } struct BareboneSceneView: View { var body: some View { SpriteView( scene: BareboneScene(), debugOptions: [.showsFPS] ) .ignoresSafeArea() } } func deviceInfoString() -> String { let os = ProcessInfo.processInfo.operatingSystemVersion let osString = "iOS \(os.majorVersion).\(os.minorVersion).\(os.patchVersion)" let model = UIDevice.current.model // "iPhone", "iPad" let machine = { var sysinfo = utsname() uname(&sysinfo) return withUnsafePointer(to: &sysinfo.machine) { ptr -> String in ptr.withMemoryRebound(to: CChar.self, capacity: 1) { cptr in String(cString: cptr) } } }() // z.B. "iPhone15,2" return "Model Identifier: \(model) (\(machine)), \(osString)" } I file a bugreport via Feedback Assistant FB22038921 The problem is no around for such a long time ! This is deeply concerning, because it questions if it is really feasable to continue to develop using Spritekit ?

In my turn-based app, existing matches created with a newer version (5.1.2) are not showing up in the MatchMaker ViewController (or even when I call GKTurnBasedMatch loadMatchesWithCompletionHandler) on a device running version 5.1. (Newer version is run from Xcode download to device). Note that in AppStoreConnect, I have the earliest compatible version set at 4.5. This was not a problem before using the (now mandatory) SDK 26 (Xcode 26.4 to be exact).

This started out as a plea for help, but in preparing this post I discovered the root cause. I'm posting it as a lesson learned in hopes it will help someone. I've spent a good chunk of March trying to get AR-mode working again in my unreleased game. I had it working with SceneKit and ARView 5 years ago, but since 2024 I've been converting the game to use RealityKit and RealityView on iOS, macOS, visionOS, and tvOS. I've been having no joy getting AR mode to work on iOS. I get the pass-through device video but the game content isn't anchored to the floor but rather anchored to the screen. I made a simple project with just a simple shape in the middle of a RealityView and an overlay with a SwiftUI toggle to go in and out of AR-mode. At first, my simple project worked, and I couldn't figure out what was different in the logic. Both projects used the same logic: func transitionToXR(_ content: inout RealityViewCameraContent) { content.remove(gameBoard.rootEntity) content.add(xrAnchor) content.camera = .spatialTracking Self.anchorStateChangedSubscription = content.subscribe(to: SceneEvents.AnchoredStateChanged.self) { event in if event.anchor == xrAnchor, event.isAnchored { xrAnchor.addChild(gameBoard.rootEntity) } } } Then I made an alternate version of my view, and reproduced the same "anchored to the screen not the floor" issue. I compared the code side-by-side and finally saw the difference! The one that didn't work, like my game, had a property 'cameraEntity' which is initialized with PerspectiveCamera(), position and look-at configured, then added as a child of the root entity. So, the simple fix was to remove 'cameraEntity' from the root entity before adding it to the detected AnchorEntity when going into AR-mode. Then when leaving AR-mode, I add back 'cameraEntity' as a child of the root entity and configure it again. So the lesson learned is: make sure there isn't a PerspectiveCamera in the tree of Entities added to an AnchorEntity with a .spatialTracking content camera. Apple: let me know if you think this is a bug or if I was being dumb. If a bug, I can use Feedback Assistant to report this. If I was being dumb, it wouldn't be the first time. :-)

Hi, I’m trying to use the new TouchController framework together with a custom Metal rendering pipeline using SpriteKit’s SKRenderer. Right now there seems to be almost no real documentation or end-to-end example showing how the framework is intended to work. Or is it just my own inexperience that I cant get it to work? My setup is: MTKView custom Renderer SKRenderer SKScene rendering SpriteKit manually through: skRenderer.render(withViewport:commandBuffer:renderPassDescriptor:) I managed to get: Metal rendering working SpriteKit rendering through SKRenderer TCTouchController connecting successfully But I still struggle to get the actual controls to render reliably on screen. Is there any minimal working example showing: MTKView SKRenderer TCTouchController rendering controls with render(using:) all together in one pipeline? Even a very small sample project would help a lot. Thanks!

Setting massProperties on a USDZ-loaded entity corrupts its transform (position/scale → NaN) Category: RealityKit / visionOS Environment: visionOS 26.x Summary: Modifying PhysicsBodyComponent.massProperties at runtime on an entity loaded from a .usdz file silently corrupts the entity's transform. The translation and scale components become NaN, while rotation remains valid. The corruption occurs during the next RealityKit scene update cycle (e.g., during an await suspension on the main actor). Steps to Reproduce: Author a .usda file with a RigidBody component including valid m_mass and m_inertia values Load the entity at runtime via Entity(named: "MyEntity.usdz", in: bundle) Read the existing PhysicsBodyComponent, modify massProperties, and set it back: if var physics = entity.components[PhysicsBodyComponent.self] { physics.massProperties = .init( mass: 1.944, inertia: SIMD3<Float>(0.013, 0.026, 0.016), centerOfMass: (position: .zero, orientation: .init()) ) entity.components.set(physics) } Add the entity to the scene Perform any await call (e.g., TextureResource(contentsOf:)) that yields to the main actor, allowing a RealityKit scene update to run Expected: The entity retains its position and scale with the updated mass/inertia values. Actual: The entity's transform becomes corrupted: Transform( scale: SIMD3<Float>(nan, nan, nan), rotation: simd_quatf(real: 1.0, imag: SIMD3<Float>(0.0, 0.0, 0.0)), // ← fine translation: SIMD3<Float>(-nan, -nan, -nan) ) Additional findings from investigation: The entity's physics mode does not matter — corruption occurs even when the entity is .kinematic Creating a new PhysicsBodyComponent(...) and replacing the existing one also triggers the bug Modifying other properties on the same component (linearDamping, angularDamping, material, mode) does not cause corruption — only massProperties triggers it The parent entity's transform remains valid The computed mass/inertia values themselves are valid (finite, positive) The corruption is silent — no error, no warning, no crash from RealityKit itself Workaround: Author mass and inertia values directly in the .usda file and do not modify massProperties at runtime: def RealityKitStruct "massFrame" { float3 m_inertia = (0.02, 0.02, 0.038) float m_mass = 2.5 def RealityKitStruct "m_pose" { } } Other PhysicsBodyComponent properties (damping, material, mode) can safely be modified at runtime.

I'm updating our app to support metal 4, but the metal 4 types don't seem to get recognized when targeting simulator. Is it known if metal 4 will be supported in the near future, or am I setting up the app wrong?

Hey all, I'm building a tile-based puzzle game for iOS and have been experimenting with using SwiftUI for the HUD (score, timer, pause menu, settings sheet) layered on top of a SpriteView that hosts the actual gameplay scene. So far the integration has been smoother than I expected — SpriteView drops cleanly into a ZStack, and I can drive SwiftUI state from the SpriteKit scene via an ObservableObject shared between them. That said, I've run into a few rough edges that I'd love some input on. The biggest one is touch handling: when a SwiftUI overlay (like a semi-transparent pause button) sits over the SpriteView, taps near the edges of the button occasionally get swallowed by the underlying scene, even when the button's hit area looks correct in the view debugger. I've tried .contentShape(Rectangle()) and bumping the frame, which helps but doesn't fully eliminate it. Curious if anyone has landed on a reliable pattern here, especially for transient overlays like toast notifications that need to ignore touches everywhere except on the toast itself. The other thing I'm weighing is animation ownership. Right now, gameplay animations (tile slides, match effects) live in SpriteKit, and HUD animations (score pop, combo counter) live in SwiftUI with withAnimation. It works, but the two animation systems don't share a clock, so when I want a "tile matched → score increment" effect to feel synchronized, I end up dispatching from the SKScene back to the ObservableObject and hoping the frame timing lines up. Has anyone found a cleaner way to coordinate timing across the two, or is this just the cost of mixing the frameworks? Would also love to hear from anyone who has shipped a game with this hybrid setup — any gotchas around Scene Phase transitions, backgrounding, or memory pressure that bit you in production? Thanks!

swift test (SwiftPM CLI) fails to decode RealityKit custom components from USD files, even though entity.components.has(MyComponent.self) returns true. Typed access via entity.components[MyComponent.self] returns nil. This forces projects that use RealityKit custom components to use xcodebuild test exclusively. Minimal repro: github.com/mesqueeb/swiftpm-realitykit-custom-component-repro Repro steps git clone https://github.com/mesqueeb/swiftpm-realitykit-custom-component-repro cd swiftpm-realitykit-custom-component-repro swift test --filter componentsPresentButNotDecodableInSwiftTest Observed ✅ entity.components.has(ReproComponent.self) returns true ❌ entity.components[ReproComponent.self] returns nil Expected If has(...) returns true for a registered custom component, typed lookup should decode and return non-nil. Notes Running the same test via xcodebuild test works correctly The component is properly registered and the USDA file correctly references it This affects any project that relies on custom RealityKit components in tests — there is no swift test workaround Feedback ID: FB22099519 Environment: macOS 15.5, Xcode 16.4, Swift 6.1