I have a bare-bones Metal app setup where I attach a CAMetalLayer to a window that inherits from a NSWindow with a custom delegate. Everything else is vanilla. I'm also using metal-cpp and metal shader converter. I'm running into a issue where the application runs fine in the beginning, but once I resize the window, it starts hitching. It turns out that [CAMetalLayer nextDrawable:] frequently (but not always) takes around a full second (plus or minus a few milliseconds) to return once drawableSize has been updated. I've tried setting allowsNextDrawableTimeout to false which doesn't work; it returns a valid drawable after a second instead of nil. Setting displaySyncEnabled to false reduces the likelihood of this happening to around 50% from 90%+ but does not eliminate it. Setting maximumDrawableCount to 2 or 3 does not seem to make a difference. By dumping the resource IDs of the returned textures I've noticed something interesting: Before resizing, the layer seems to shuffle between 2 textures or at least 2 resource IDs, but after resizing it starts to create new textures for each returned drawable. Occasionally it seems to reuse a previous resource ID, but it does not seem to have anything to do with whether the method returns quickly or not. Why does this happen, and how can I fix it? Should I create a new CAMetalLayer when resizing the window instead of updating drawableSize?

I am interested in learning the Metal framework for rendering development. However, most of Apple’s official documentation uses Objective-C code. Therefore, I am seeking guidance on whether it is more advantageous for me to focus solely on learning Swift to gain proficiency in Metal.

I am making a framework in C++ using metal-cpp, basically a small game engine. I am also consequently using metal-cpp-extensions provided in LearnMetalCPP to make applications work. For one of my classes, I needed to add AppKit.hpp inside a public header file, so I moved it and its associate headers(NSApplication.hpp, NSMenu.hpp, etc.) from Project headers to Public in Build Phases' Headers, however, it started giving me the error "cast of C pointer type 'void *' to Objective-C pointer type 'Class' requires a bridged cast" at several points in the AppKit headers. They don't appear when AppKit and its associates are in the Project headers, or when they are in the Private headers and no headers import it. I imagined that disabling Objective-C ARC and Using __bridge casts outside of ARC in Build Settings would solve it, but it didn't budge. I imagined it wouldn't involve actively changing the headers would be the answer, but even if I try to put __bridge before the problematic casts, it didn't recognize __bridge. How do I solve this? And why is it only happening in Public and not Project headers?

Hello, Apple! This post is a bug report for Metal driver in MacOS Sequoia. I'm working on opensource game engine and one of my users reported a bug, on "MacOS Sequoia" + "AMD Radeon RX 6900 XT".  Engine crashes, when liking a compute pipeline, with following NSError: "Compiler encountered an internal error: I" Offended shader (depth aware blur): https://shader-playground.timjones.io/27565de40391f62f078c891077ba758c On my end, compiling same shader on M1 (Sonoma) or with offline compiler doesn't reproduce the issue. Post with bug-report on github: https://github.com/Try/OpenGothic/issues/712 Looking forward for your help and driver fix ;)

I'm trying to pass a buffer of float2 items from CPU to GPU. In the kernel, I can provide a parameter for the buffer: device const float2* values for example. How do I specify float2 as the type for the MTL::Buffer? I managed to get the code to work by "cheating" by defining a simple class that has the same data members as a float2, but there is probably a better way. class Coord_f { public: float x{0.0f}; float y{0.0f}; }; then using code to allocate like this: NS::TransferPtr(device->newBuffer(n_elements * sizeof(Coord_f), MTL::ResourceStorageModeManaged)) The headers for metal-cpp do not appear to define vector objects like float2, but I'm doubtless missing something. Thanks.

I am trying to use the SVGF denoiser to denoise my ray traced shadows (and also other textures later). I do get a smoothed image, but with wonky denoising. I need the depth-normal textures and motion textures for the SVGF and assume that these are badly filled in my case. However, neither in the above linked documentation nor in the WWDC19 video I find how they should be defined. I am looking to answers to: Is depth in red or alpha channel for the depth-normal texture? Are the normals in screen space? Is depth linear? Is it distance or z coordinate in view space? Or even logarithmically scaled or something else? Are the motion vectors supposed to be in pixels per frame? What is the orientation of the axis? Is y up or down? Are there are other restrictions on the formats? Also the linked code did not help me (I have not found any SVGF so far; also all the code is in Objective-C++, not Swift, but that's a different topic). So how should I fill these textures. Can someone point me to the documentation where these kinds of questions are answered?

What evidence exists that it's safe to call nextDrawable() on CAMetalLayer off the main thread? I have seen developers claiming that it's OK, but the official docs are silent on the topic. Attempting to do so with Strict Concurrency Checking set to Complete complains that CAMetalLayer is not @Sendable. I want to call it off the main thread since there doesn't seem to be any way to prevent it from blocking the UI for up to a second. I have read hints and allegations that this won't happen if you avoid asking for too many drawables, but that doesn't seem to be true 100% of the time in my experience. Supposing it is allowed, I wonder how races are handled such as when the layer's size is changed on the main thread, or if the layer is removed from the layer hierarchy.

Is there a working example of imageblock_slice with implicit layout somewhere? I get a compilation error when i write this: imageblock_slilce color_slice = img_blk.slice(frag->color); Error: No matching member function for call to 'slice' candidate template ignored: couldn't infer template argument 'E' candidate function template not viable: requires 2 arguments, but 1 was provided Too few template arguments for class template 'imageblock_slice' It seems the syntax has changed since the Imageblocks presentation https://developer.apple.com/videos/play/tech-talks/603/ I tried supplying the struct type of the image block between <> but it still does not work.

I have an M1 Pro with a 16-core GPU. When I run a shader with 8193 threads, atomic_thread_fence is violated across the boundary between thread 8191 (the last thread in the 7th threadgroup) and 8192 (the first thread in the 9th threadgroup). I've attached the Metal and Swift files, but I'll repost the relevant kernel here. It's a function that launches N threads to iterate through a binary tree from the leaves, where the first thread to reach the parent terminates and the second one populates it with the sum of the nodes two children. // clang-format off void sum(device const int& size, device const int* __restrict__ in, device int* __restrict__ out, device atomic_int* visited, uint i [[thread_position_in_grid]]) { // clang-format on int val = in[i]; uint cur = (size + i - 1); out[cur] = val; atomic_thread_fence(mem_flags::mem_device, memory_order_seq_cst); cur = (cur - 1) / 2; int proceed = atomic_fetch_add_explicit(&visited[cur], 1, memory_order_relaxed); while (proceed == 1) { uint left = 2 * cur + 1; uint right = 2 * cur + 2; uint val_left = out[left]; uint val_right = out[right]; uint val_cur = val_left + val_right; out[cur] = val_cur; if (cur == 0) { break; } cur = (cur - 1) / 2; atomic_thread_fence(mem_flags::mem_device, memory_order_seq_cst); proceed = atomic_fetch_add_explicit(&visited[cur], 1, memory_order_relaxed); } } What I'm observing is that thread 8192 hits the atomic_fetch_add first and terminates, while thread 8191 hits it second (observes that thread 8192 had incremented it by 1) and proceeds into the loop. Thread 8191 reads out[16383] (which it populated with 8191) and out[16384] (which thread 8192 populated with 8192 prior to the atomic_thread_fence). Instead of reading 8192 from out[16384] though, it reads 0. Maybe I'm missing something but this seems like a pretty clear violation of the atomic_thread_fence which (I thought) was supposed to guarantee that the write from thread 8192 to out[16384] would be visible to any thread observing the effects of the following atomic_fetch_add. Is atomic_fetch_add not a store operation? Modifying it to an atomic_store or atomic_exchange still results in the bug. Adding another atomic_thread_fence between the atomic_fetch_add and reading of out also doesn't change anything. I only begin to observe this on grid sizes of 8193 and upwards. That's 9 threadgroups per grid, which I assume could be related to my M1 Pro GPU having 16 cores. Running the same example on an A17 Pro GPU doesn't show any of this behavior up through a tested grid size of 4194303 (2^22-1), at which point testing larger grid sizes starts to run into other issues so I can't test anything larger. Removing the atomic_thread_fences on both the M1 and A17 cause the test to fail at much smaller grid sizes, as expected. sum.metal main.swift

I’ve been trying to run Jurassic World Evolution 2 using the Game Porting Toolkit on macOS, but the game doesn’t launch and crashes immediately. Based on the error and research, it seems the issue is related to missing support for D3D12_TILED_RESOURCES_TIER_2 in the Metal API. If this is the case, does anyone know if support for tiled resources is planned for future updates of the toolkit? Or are there any potential workarounds for bypassing this limitation?

I have a very basic usdz file from this repo I call loadTextures() after loading the usdz via MDLAsset. Inspecting the MDLTexture object I can tell it is assigning a colorspace of linear rgb instead of srgb although the image file in the usdz is srgb. This causes the textures to ultimately render as over saturated. In the code I later convert the MDLTexture to MTLTexture via MTKTextureLoader but if I set the srgb option it seems to ignore it. This significantly impacts the usefulness of Model I/O if it can't load a simple usdz texture correctly. Am I missing something? Thanks!

Hi! I just installed GPTK2 on my new Mac , but the Terminal gave “Error:OpenSSL1.1 has been disabled.” How should I fix it?Or waiting for the GPTK2 beta4? Thanks.

Currently looking for Metal developers to port Quake 2 RTX to Metal RT in order to give Apple Silicon Macs an amazing Pathtracing demo, This project falls under NightSightProductions who is also working on a Portal 2 with RTX Remaster. if you are interested and want to help further Mac gaming, message me here or on discord at king_vulpes

When generating large arrays of random numbers, NaNs show up. They also show up at the same indices when using the same seed, leading me to believe that this is a bug with MPSMatrixRandom's normally distributed Float32 random number distribution. Happens with both Philox and MTGP32. Is this intentional and how do I work around this? See the original post for a MWE in Swift and Julia: https://github.com/JuliaGPU/Metal.jl/issues/474

I am building a MacOS desktop app (https://anukari.com) that is using Metal compute to do real-time audio/DSP processing, as I have a problem that is highly parallelizable and too computationally expensive for the CPU. However it seems that the way in which I am using the GPU, even when my app is fully compute-limited, the OS never increases the power/performance state. Because this is a real-time audio synthesis application, it's a huge problem to not be able to take advantage of the full clock speeds that the GPU is capable of, because the app can't keep up with real-time. I discovered this issue while profiling the app using Instrument's Metal tracing (and Game tracing) modes. In the profiling configuration under "Metal Application" there is a drop-down to select the "Performance State." If I run the application under Instruments with Performance State set to Maximum, it runs amazingly well, and all my problems go away. For comparison, when I run the app on its own, outside of Instruments, the expensive GPU computation it's doing takes around 2x as long to complete, meaning that the app performs half as well. I've done a ton of work to micro-optimize my Metal compute code, based on every scrap of information from the WWDC videos, etc. A problem I'm running into is that I think that the more efficient I make my code, the less it signals to the OS that I want high GPU clock speeds! I think part of why the OS is confused is that in most use cases, my computation can be done using only a small number of Metal threadgroups. I'm guessing that the OS heuristics see that only a small fraction of the GPU is saturated and fail to scale up the power/clock state. I'm not sure what to do here; I'm in a bit of a bind. One possibility is that I intentionally schedule busy work -- spin threadgroups just to waste energy and signal to the OS that I need higher clock speeds. This is obviously a really bad idea, but it might work. Is there any other (better) way for my app to signal to the OS that it is doing real-time latency-sensitive computation on the GPU and needs the clock speeds to be scaled up? Note that game mode is not really an option, as my app also runs as an AU plugin inside hosts like Garageband, so it can't be made fullscreen, etc.

We have a production Metal app with a complex multithreaded Metal pipeline. When everything is operating smoothly, it works great. Even when extremely overloaded, it does not crash for days at a time. This isn't good enough for our users. Unfortunately, when I have zero visibility into id, I have no way of knowing when metal is "done" with an id. When overloaded, stale metal render passes need to be 'aborted', which results in metal callbacks not being called. for example, these callbacks may not be called after an aborted pass: id<MTLCommandBuffer> m_cmdbuf; [m_cmdbuf addScheduledHandler:^(id <MTLCommandBuffer> cb) { cpr->scheduled = MachAbsoluteTime(); }]; [m_cmdbuf addCompletedHandler:^(id <MTLCommandBuffer> cb) { cpr->completed = MachAbsoluteTime(); }]; For the moment, our workaround is a system which waits a few seconds after we "think" a rendering pass should be done with all its (aborted) resources before releasing buffers. This is not ideal, to say the least. So, in summary, my question is, it would be nice to be able to 'query' an id to know when metal is done with it, so that we know that its safe to release it along with our own internal resources. Is there any such (undocumented) mechanism? I have exhaustively read all existing Metal documentation many times. An idea that I've been toying with... it would be nice to have something akin to Zombie detection running all the time for id only. In OpenGL, it was OK to use a released texture... you may display a bad frame, but not CRASH!. Is there any similar option for id?

Project: I have some data wich could be transformed by shader, result may be kept in rgb channels of image. Great. But now to mix dozens of those results? Not one by one, image after image, but all at once. Something like „complicated average” color of particular pixel from all delivered images. Is it possible?

my app use mtkview to render video, but [MTKView initwithFrame:device] takes 2-3s in some some 2019 macbook pro, system macos 15.0.1. how can I do?

Hello, I am using MTKView to display: camera preview & video playback. I am testing on iPhone 16. App crashes at a random moment whenever MTKView is rendering CIImage. MetalView: public enum MetalActionType { case image(CIImage) case buffer(CVPixelBuffer) } public struct MetalView: UIViewRepresentable { let mtkView = MTKView() public let actionPublisher: any Publisher<MetalActionType, Never> public func makeCoordinator() -> Coordinator { Coordinator(self) } public func makeUIView(context: UIViewRepresentableContext<MetalView>) -> MTKView { guard let metalDevice = MTLCreateSystemDefaultDevice() else { return mtkView } mtkView.device = metalDevice mtkView.framebufferOnly = false mtkView.clearColor = MTLClearColor(red: 0, green: 0, blue: 0, alpha: 0) mtkView.drawableSize = mtkView.frame.size mtkView.delegate = context.coordinator mtkView.isPaused = true mtkView.enableSetNeedsDisplay = true mtkView.preferredFramesPerSecond = 60 context.coordinator.ciContext = CIContext( mtlDevice: metalDevice, options: [.priorityRequestLow: true, .highQualityDownsample: false]) context.coordinator.metalCommandQueue = metalDevice.makeCommandQueue() context.coordinator.actionSubscriber = actionPublisher.sink { type in switch type { case .buffer(let pixelBuffer): context.coordinator.updateCIImage(pixelBuffer) break case .image(let image): context.coordinator.updateCIImage(image) break } } return mtkView } public func updateUIView(_ nsView: MTKView, context: UIViewRepresentableContext<MetalView>) { } public class Coordinator: NSObject, MTKViewDelegate { var parent: MetalView var metalCommandQueue: MTLCommandQueue! var ciContext: CIContext! private var image: CIImage? { didSet { Task { @MainActor in self.parent.mtkView.setNeedsDisplay() //<--- call Draw method } } } var actionSubscriber: (any Combine.Cancellable)? private let operationQueue = OperationQueue() init(_ parent: MetalView) { self.parent = parent operationQueue.qualityOfService = .background super.init() } public func mtkView(_ view: MTKView, drawableSizeWillChange size: CGSize) { } public func draw(in view: MTKView) { guard let drawable = view.currentDrawable, let ciImage = image, let commandBuffer = metalCommandQueue.makeCommandBuffer(), let ci = ciContext else { return } //making sure nothing is nil, now we can add the current frame to the operationQueue for processing operationQueue.addOperation( MetalOperation( drawable: drawable, drawableSize: view.drawableSize, ciImage: ciImage, commandBuffer: commandBuffer, pixelFormat: view.colorPixelFormat, ciContext: ci)) } //consumed by Subscriber func updateCIImage(_ img: CIImage) { image = img } //consumed by Subscriber func updateCIImage(_ buffer: CVPixelBuffer) { image = CIImage(cvPixelBuffer: buffer) } } } now the MetalOperation class: private class MetalOperation: Operation, @unchecked Sendable { let drawable: CAMetalDrawable let drawableSize: CGSize let ciImage: CIImage let commandBuffer: MTLCommandBuffer let pixelFormat: MTLPixelFormat let ciContext: CIContext init( drawable: CAMetalDrawable, drawableSize: CGSize, ciImage: CIImage, commandBuffer: MTLCommandBuffer, pixelFormat: MTLPixelFormat, ciContext: CIContext ) { self.drawable = drawable self.drawableSize = drawableSize self.ciImage = ciImage self.commandBuffer = commandBuffer self.pixelFormat = pixelFormat self.ciContext = ciContext } override func main() { let width = Int(drawableSize.width) let height = Int(drawableSize.height) let ciWidth = Int(ciImage.extent.width) //<-- Thread 22: EXC_BAD_ACCESS (code=1, address=0x5e71f5490) A bad access to memory terminated the process. let ciHeight = Int(ciImage.extent.height) let destination = CIRenderDestination( width: width, height: height, pixelFormat: pixelFormat, commandBuffer: commandBuffer, mtlTextureProvider: { [self] () -> MTLTexture in return drawable.texture }) let transform = CGAffineTransform( scaleX: CGFloat(width) / CGFloat(ciWidth), y: CGFloat(height) / CGFloat(ciHeight)) do { try ciContext.startTask(toClear: destination) try ciContext.startTask(toRender: ciImage.transformed(by: transform), to: destination) } catch { } commandBuffer.present(drawable) commandBuffer.commit() commandBuffer.waitUntilCompleted() } } Now I am no Metal expert, but I believe it's a very simple execution that shouldn't cause memory leak especially after we have already checked for whether CIImage is nil or not. I have also tried running this code without OperationQueue and also tried with @autoreleasepool but none of them has solved this problem. Am I missing something?

Hi, A user sent us a crash report that indicates an error occurring just after loading the default Metal library of our app. Application Specific Information: Crashing on exception: *** -[__NSArrayM objectAtIndex:]: index 0 beyond bounds for empty array The report pointed me to these (simplified) lines of codes in the library setup: _vertexFunctions = [[NSMutableArray alloc] init]; _fragmentFunctions = [[NSMutableArray alloc] init]; id<MTLLibrary> library = [device newDefaultLibrary]; 2 vertex shaders and 5 fragment shaders are then loaded and stored in these two arrays using this method: -(BOOL) addShaderNamed:(NSString *)name library:(id<MTLLibrary>)library isFragment:(BOOL)isFragment { id shader = [library newFunctionWithName:name]; if (!shader) { ALOG(@"Error : Unable to find the shader named : “%@”", name); return NO; } [(isFragment ? _fragmentFunctions : _vertexFunctions) addObject:shader]; return YES; } As you can see, the arrays are not filled if the method fails... however, a few lines later, they are used without checking if they are really filled, and that causes the crash... But this coding error doesn't explain why no shader of a certain type (or both types) have been added to the array, meaning: why -newFunctionWithName: returned nil for all given names (since the implied array appears completely empty)? Clue This error has only be detected once by a user running the app on macOS 10.13 with a NVIDIA Web Driver instead of the default macOS graphic driver. Moreover, it wasn't possible to reproduce the problem on the same OS using the native macOS driver. So my question is: is it some known conflicts between NVIDIA drivers and the use of Metal libraries? Or does this case would require some specific options in the Metal implementation? Any help appreciated, thanks!