Hi, just got an Apple M3 Pro to try it out on some Jax operations. I see the development is actively ongoing so maybe this error can help. This is the environment: Metal device set to: Apple M3 Pro systemMemory: 18.00 GB maxCacheSize: 6.00 GB jax: 0.4.26 jaxlib: 0.4.23 numpy: 1.26.4 python: 3.11.8 | packaged by conda-forge | (main, Feb 16 2024, 20:49:36) [Clang 16.0.6 ] jax.devices (1 total, 1 local): [METAL(id=0)] process_count: 1 platform: uname_result(system='Darwin', node='MKFL96VR9YT', release='23.4.0', version='Darwin Kernel Version 23.4.0: Wed Feb 21 21:44:54 PST 2024; root:xnu-10063.101.15~2/RELEASE_ARM64_T6030', machine='arm64') This is a minimal example which produces an error, I think due to the fft part: from jax import numpy as np array = np.ones((16, 16)) np.fft.fft2(array) This is the full traceback: Traceback (most recent call last): File "/Users/user/Downloads/wow.py", line 5, in <module> np.fft.fft2(array) File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/numpy/fft.py", line 216, in fft2 return _fft_core_2d('fft2', xla_client.FftType.FFT, a, s=s, axes=axes, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/numpy/fft.py", line 210, in _fft_core_2d return _fft_core(func_name, fft_type, a, s, axes, norm) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/numpy/fft.py", line 102, in _fft_core transformed = lax.fft(arr, fft_type, tuple(s)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/traceback_util.py", line 179, in reraise_with_filtered_traceback return fun(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/pjit.py", line 298, in cache_miss outs, out_flat, out_tree, args_flat, jaxpr, attrs_tracked = _python_pjit_helper( ^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/pjit.py", line 176, in _python_pjit_helper out_flat = pjit_p.bind(*args_flat, **params) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/core.py", line 2788, in bind return self.bind_with_trace(top_trace, args, params) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/core.py", line 425, in bind_with_trace out = trace.process_primitive(self, map(trace.full_raise, args), params) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/core.py", line 913, in process_primitive return primitive.impl(*tracers, **params) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/pjit.py", line 1494, in _pjit_call_impl return xc._xla.pjit(name, f, call_impl_cache_miss, [], [], donated_argnums, # type: ignore ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/pjit.py", line 1471, in call_impl_cache_miss out_flat, compiled = _pjit_call_impl_python( ^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/pjit.py", line 1406, in _pjit_call_impl_python lowering_parameters=mlir.LoweringParameters()).compile() ^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/interpreters/pxla.py", line 2369, in compile executable = UnloadedMeshExecutable.from_hlo( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/interpreters/pxla.py", line 2908, in from_hlo xla_executable, compile_options = _cached_compilation( ^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/interpreters/pxla.py", line 2718, in _cached_compilation xla_executable = compiler.compile_or_get_cached( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/compiler.py", line 266, in compile_or_get_cached return backend_compile(backend, computation, compile_options, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/profiler.py", line 335, in wrapper return func(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "/opt/anaconda3/envs/jaxmetal/lib/python3.11/site-packages/jax/_src/compiler.py", line 238, in backend_compile return backend.compile(built_c, compile_options=options) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ jaxlib.xla_extension.XlaRuntimeError: UNKNOWN: <unknown>:0: error: 'func.func' op One or more function input/output data types are not supported. <unknown>:0: note: see current operation: "func.func"() <{arg_attrs = [{mhlo.layout_mode = "default", mhlo.sharding = "{replicated}"}], function_type = (tensor<16x16xf32>) -> tensor<16x16xcomplex<f32>>, res_attrs = [{jax.result_info = "", mhlo.layout_mode = "default"}], sym_name = "main", sym_visibility = "public"}> ({ ^bb0(%arg0: tensor<16x16xf32>): %0 = "mhlo.convert"(%arg0) : (tensor<16x16xf32>) -> tensor<16x16xcomplex<f32>> %1 = "mhlo.fft"(%0) {fft_length = dense<16> : tensor<2xi64>, fft_type = #mhlo<fft_type FFT>} : (tensor<16x16xcomplex<f32>>) -> tensor<16x16xcomplex<f32>> "func.return"(%1) : (tensor<16x16xcomplex<f32>>) -> () }) : () -> () <unknown>:0: error: failed to legalize operation 'func.func' <unknown>:0: note: see current operation: "func.func"() <{arg_attrs = [{mhlo.layout_mode = "default", mhlo.sharding = "{replicated}"}], function_type = (tensor<16x16xf32>) -> tensor<16x16xcomplex<f32>>, res_attrs = [{jax.result_info = "", mhlo.layout_mode = "default"}], sym_name = "main", sym_visibility = "public"}> ({ ^bb0(%arg0: tensor<16x16xf32>): %0 = "mhlo.convert"(%arg0) : (tensor<16x16xf32>) -> tensor<16x16xcomplex<f32>> %1 = "mhlo.fft"(%0) {fft_length = dense<16> : tensor<2xi64>, fft_type = #mhlo<fft_type FFT>} : (tensor<16x16xcomplex<f32>>) -> tensor<16x16xcomplex<f32>> "func.return"(%1) : (tensor<16x16xcomplex<f32>>) -> () }) : () -> () I'd be happy running more tests should you need them, I'm new to this, so not sure which just yet. Many thanks!!

Hello, When I build an XCode project with the Apple Silicon chip, I have some issues. The project contains Pods and Swift Packages. I could not run the application at all and always got the following error: Could not find module '***' for target 'x86_64-apple-ios-simulator'; found: arm64, arm64-apple-ios-simulator, at: *** I try to resolve this issue. Always embed swift standard libraries = YES Build Active Architure Only = YES UIRequiredDeviceCapabilities = armv7 Excluded Architectures > Debug > Any iOS Simulator SDK arm64 add Open Using Rosetta Excluded Architectures > Debug > Any iOS Simulator SDK arm64 remove However, issue always come to me. :( Do you have any solution for this problem ? Thank you by advance !

Ventura 13.2.1 M1 Sonoma 14.2.1 M2 In my app I have a signal handler. When testing it with null-dereference I see that in previous MacOs versions like Monterey 12.0 x86 the signal handler is called. However, on my Silicon Ventura/Sonoma machines its not called. Tried with SIP enabled and disabled So I created a binary with code: #include <iostream> int main() { int *ptr = nullptr; std::cout << *ptr; // Dereference null pointer return 0; } Compiled it with: g++ null.cpp -o null.bin And executed it with and without sudo. The app indeed crashes because of the null dereference (and core dump is created when SIP disabled). However, no signal is recived. I am able to prove it with DTrace . DTrace script: #pragma D option quiet proc:::signal-send { @[execname, stringof(args[1]->pr_fname), args[2]] = count(); } END { printf("%20s %20s %12s %s\n", "SENDER", "RECIPIENT", "SIG", "COUNT"); printa("%20s %20s %12d %@d\n", @); } Here is the output. In the left terminal I executed the binary. In the right terminal the script output. On top of DTrace I created and MacOS endpoint-security app and subscribed to ES_EVENT_TYPE_NOTIFY_SIGNAL. Same there, no signal. Did anything change with signals on M1/M2 MacOS 13.0 ?

After enabling the signal alternative stack using sigaltstack, the thread stack region size comes to be 128 MB. The stack region size is only 8 MB without enabling the signal alternative stack. (get by calling the mach_vm_region_recurse function) It only happens on macOS 14 or later, and M series silicon. The growth of the stack region size results in a too-large minidump file when we generate a crash report using Google's crashpad library. Could anybody tell me why the size has increased so much? And is there a way to work around it?

Tried various how-tos on youtube and github. Have conda. Third step fails. conda install -c apple tensorflow-deps pip install tensorflow-macos pip install tensorflow-metal ERROR: Could not find a version that satisfies the requirement tensorflow-metal (from versions: none) ERROR: No matching distribution found for tensorflow-metal I see a lot of fixes for Intel-based Mac. None for M3. HELP!?

I have the XCode Version 15.3 (15E204a). When I try to compile my application, the following errors occur: Undefined symbol: _GDTCCTConstructiOSClientInfo Undefined symbol: _GDTCCTNetworkConnectionInfoNetworkMobileSubtype Linker command failed with exit code 1 (use -v to see invocation) Any solution to fix this issue?

I am using UDP communication in a app. Here is what i do, Initialises a UDP broadcast connection object. Bind it with a port to listen Receives the IP &amp; Port from the UDP connection to connect further with TCP connection. After updating Xcode 15.3, It works until the iPad is connected with mac in debug mode. When i create build to test remotely, it stops receiving IP &amp; Port from UDP connection. Here is how i concluded this is Xcode issue, I tried to debug this issue with Xcode 15.2 and it works as expected with debug and after creating build also. Any help / suggestion would be appreciated.

When running an iOS app as designed for iPad on an m1 Mac mini the UIImagePickerController.isSourceTypeAvailable(.camera) api returns true leading to a crash (attached) if the camera is selected to upload an image to the app as my much loved Mac mini does not have a camera. For the moment have disabled camera if platform is Mac by adding the qualification: ProcessInfo().isiOSAppOnMac == false but this seems like a bug or does the crash also happen on Macs with cameras? Other image picker options work fine. Crash log

Hello, I am developing a tool in python or nodejs to intercept flows between two applications (MITM). I want to use the FRIDA library, but when I use it I get the following error: Error: module not found at "/usr/lib/libSystem.B.dylib". Indeed, the library is not present in the folder. I tried to get help directly from FRIDA but I couldn't find any help and on the current forum I did see some posts talking about this problem but I couldn't solve it. Do you have an idea of how to solve this problem. Thank you. Ps: I'm new to the APPLE ecosystem (Mac mini - Apple M1)

We develop virtual instruments for Mac/AU and are trying to get our AU-Plugins and our Standalone player to work with Audio Workgroups. When the Standalone App or Logic Pro is in the foreground and active all is well and as expected. However when the App or Logic Pro is not in focus all my auxiliary threads are running on E-Cores. Even though they are properly joined to the processing thread's workgroup. This leads to a lot of audible drop outs because deadlines are not met anymore. The processing thread itself stays on a p-core. But has to wait for the other threads to finish. How can I opt out of this behaviour? Our users certainly have use cases where they expect the Player to run smoothly even though they currently have a different App in focus.

I've been working on an iOS project for the iPhone and would like to support running it on macOS computers with Apple Silicon. In the Targets / Supported Destinations we added "Mac (Designed for iPhone)" but experienced Thread 1: EXC_BAD_ACCESS crashes immediately when we tried to run it. We've isolated it down to Stepper UI elements in our view. Starting a new project and just trying to present a single Stepper in the ContentView, we get the same crash. Here is code that presents the issue: // ContentView.swift import SwiftUI struct ContentView: View { @State var someValue = 5 var body: some View { VStack { Stepper("Stepper", value: $someValue, in: 0...10) } } } When run from Xcode on an iOS device or the simulator, it runs fine. Trying to run it on the Mac, it crashes here: // Stepper_01App.swift import SwiftUI @main // <-- Thread 1: EXC_BAD_ACCESS (code=2, address=0x16a643f70) struct Stepper_01App: App { var body: some Scene { WindowGroup { ContentView() } } } Xcode 14.3 (14E222b), MacOS Ventura 13.3.1 (a), Mac mini M2. Target: Mac (Designed for iPhone) We have verified that the same code crashes on all the Apple Silicon Macs we have access to. Searching the Internet and Apple Developer forums I dont find other reports, so I kind of feel there must be some level of either user error or system/project misconfiguration going on? If any iOS app that used Steppers was just crashing when trying to run on a Mac, it seems like this would be a big deal. If anyone has input or can point out what we need to do differently, it would be appreciated!

Im honestly a bit lost and looking for general pointers. Here is the general flow of my project. I have an Xcode project where I want to return and convert the temperature values accessed from the apple smc and I found a GitHub repo with all the smc key sensors for the M3Pros/Max chips: https://github.com/exelban/stats/issues/1703 basically, I have all these keys stored in an array in obj-c like so: NSArray *smcKeys = @[ @"Tp01", @"Tp05", @"Tp09", @"Tp0D", @"Tp0b", @"Tp0f", @"Tp0j", @"Tp0n",@"Tp0h", @"Tp0L", @"Tp0S", @"Tp0V", @"Tp0z", @"Tp0v", @"Tp17", @"Tp1F", @"Tp1J", @"Tp1p", @"Tp1h", @"Tp1R", ]; I am passing all these keys by passing 'smcKeys' in a regular C code file I have here that is meant to open, close and read the data shown here: #include "smc.h" #include <mach/mach.h> #include <IOKit/IOKitLib.h> #include "smckeys.h" io_connect_t conn; kern_return_t openSMC(void) { kern_return_t result; kern_return_t service; io_iterator_t iterator; service = IOServiceGetMatchingServices(kIOMainPortDefault, IOServiceMatching("AppleSMC"), &iterator); if(service == 0) { printf("error: could not match dictionary"); return 0; } result = IOServiceOpen(service, mach_task_self(), 0, &conn); IOObjectRelease(service); return 0; } kern_return_t closeSMC(void) { return IOServiceClose(conn); } kern_return_t readSMC(char *smcKeys, SMCVal_t *val) { kern_return_t result; uint32_t keyCode = *(uint32_t *)smcKeys; SMCVal_t inputStruct; SMCVal_t outputStruct; inputStruct.datasize = sizeof(SMCVal_t); inputStruct.datatype = 'I' << 24; //a left shift operation. turning the I into an int by shifting the ASCII value 24 bits to the left inputStruct.data[0] = keyCode; result = IOConnectCallStructMethod(conn, 5, &inputStruct, sizeof(SMCVal_t), &outputStruct, (size_t*)&inputStruct.datasize); if (result == kIOReturnSuccess) { if (val -> datasize > 0) { if (val -> datatype == ('f' << 24 | 'l' << 16 | 't' << 8 )) { //bit shifting to from 32bit operation associated with the ASCII charecters'f', 'l', and 't', sets datatype field. double temp = *(double *)val -> data; return temp; } } } return 0.0; } Which I am then then calling the functions from this file in a swift file and converting the values to Fahrenheit but no data is being printed in my console: import IOKit public class getTemperature { public struct SMCVal_t { var datasize: UInt32 var datatype: UInt32 var data: (UInt8, UInt8, UInt8, UInt8, UInt8, UInt8, UInt8, UInt8) } @_silgen_name("openSMC") func openSMC() -> kern_return_t @_silgen_name("closeSMC") func closeSMC() -> kern_return_t @_silgen_name("readSMC") func readSMC(key: UnsafePointer<CChar>?,val: UnsafeMutablePointer<SMCVal_t>) -> kern_return_t func convertAndPrintTempValue(key:UnsafePointer<CChar>?,scale: Character, showTemp: Bool ) -> kern_return_t { let openSM = openSMC() guard openSM == 0 else { print("Failed to open SMC: \(openSM)") return kern_return_t() } let closeSM = closeSMC() guard closeSM == 0 else { print("could not close SMC: \(closeSM)") return IOServiceClose(conn) } func convertAndPrint(val: SMCVal_t) -> Double { if val.datatype == (UInt32("f".utf8.first!) << 24 | UInt32("l".utf8.first!) << 16 | UInt32("t".utf8.first!) << 8) { let extractedTemp = Double(val.data.0) return( extractedTemp * 9.0 / 5.0 + 32.0 ) } return 0.0 } let smcValue = SMCVal_t(datasize: 0, datatype: 0, data: (0,0,0,0,0,0,0,0)) let convertedVal = convertAndPrint(val: smcValue) print("Temperarure:\(convertedVal)F°") return kern_return_t() } } I know this is a lot but I am honestly looking for any tips to fill in any gaps in my knowledge for anyone who's built a similar application meant to extract any sort of data from Mac hardware.

I'm working on an app for macOS where it would be very useful to display the GPU (graphics card) workload usage as a percentage. CPU usage monitoring is easy, but GPU monitoring on Apple Silicon is next-to-impossible. Apple only seems to give us our app’s GPU usage which is not what we want, since we want to total GPU workload for the whole system. I'm using the latest version of Xcode and Swift, any ideas how to achieve this?

I'm struggling with compiling lib opus so that it works in the simulator on Apple silicon. I found a thread on the forums that seems to address part of the issue, but I am unable to build the static lib so that it shows the platform it is targeting. The thread mentions that I should be able to run otool and see a "load commands" that indicate the platform. When I run otool against the static library that we have created, it doesn't list any load commands. I don't see LC_BUILD_VERSION or LC_VERSION_MIN_***. Why would there not be any "Load command" entries? % otool -l -arch arm64 dependencies/lib/libopus.a Archive : dependencies/lib/libopus.a dependencies/lib/libopus.a(bands.o): is an LLVM bit-code file dependencies/lib/libopus.a(celt.o): is an LLVM bit-code file dependencies/lib/libopus.a(celt_encoder.o): is an LLVM bit-code file dependencies/lib/libopus.a(celt_decoder.o): is an LLVM bit-code file ... dependencies/lib/libopus.a(mlp.o): is an LLVM bit-code file dependencies/lib/libopus.a(mlp_data.o): is an LLVM bit-code file The static library has the two architectures embedded in it, but when compiling the framework for the simulator platform the linking phase complains that we are building for the simulator, but linking object code built for ios. % lipo -info dependencies/lib/libopus.a Architectures in the fat file: dependencies/lib/libopus.a are: x86_64 arm64 In case you are curious, I'm just piggybacking on this project that has a build-libopus.sh script in the root directory that builds the official open source Opus library files. My hope is to build this static library for ios, ios-simulator, and mac-catalyst platforms and then include them in a xcframework.

Is there a way in pure C (not Objective-C, and certainly not Swift) way to detect if an iOS app is running on a Mac? I'm aware of iOSAppOnMac, but AFAICT I'd need to write Objective-C code to use it. My application is quite old and large, with portions going back to the 1980s, so the burden of moving on the anything that can't be accessible directly from C is quite high.

My device has an M2 Max chip, and I am using Xcode version 15.1 Beta 3. My app runs normally in iOS and iPad simulators, but when I attempt to run it in the Vision Pro simulator, even though the compilation is successful, a dialog box appears stating, 'AppName's architectures (Intel 64-bit) include none that Apple Vision Pro can execute (arm64).' Consequently, the app is not successfully installed in the Vision Pro simulator. Additionally, my project uses Cocoapods for dependency management. I would appreciate any help, thank you!

On xcode 13, I have macos project that runs fine on intel machine. On apple silicon (M1 Plus) I get the error "A build only device cannot be used to run this target.", when I try to run from Xcode. This seems to be an ios error. All Google suggested fixes involve picking a new device which is an ios fix. Right? Bulids fine and Archive app runs fine. I get the error for both intel and Arm64 architectures. I tried building for Target deployment target device families and Deployment as: macos 12.0 sdk any suggestions?

#include <stdio.h> int main() { unsigned long a, d; __asm__ volatile ( "\n\t" "movl $0x77777777, %%eax\n\t" "movl $0xffffffff, %%ecx\n\t" "xorl %%edx, %%edx\n\t" "divl %%ecx\n\t" "cwtd\n\t" "movq %%rax, %0\n\t" "movq %%rdx, %1\n\t" : "=r"(a), "=r"(d) :: "rax", "rdx" ); printf("rax: %lx, rdx, %lx", a, d); } The minimal program above was expected to give rax: 0, rdx: 77770000 but on macOS with Rosetta 2 it gave rax: 0, rdx, 0. It causes specfic programs (e.g. Genshin Impact) to crash.