idun 0.1.0__tar.gz

idun-0.1.0/CMakeLists.txt

@@ -0,0 +1,39 @@
+# Native layer of Idun. It builds the Metal runtime; a CUDA target with the same C ABI conventions would join it
+# (enable_language(CUDA) behind an option), which is why this is CMake and not a one-line clang call.
+cmake_minimum_required(VERSION 3.28)
+# Before project(), or it does not apply; FORCE so that an existing build tree takes it too (a build without it targets
+# the SDK's macOS, 26, and cannot load on 15).
+set(CMAKE_OSX_DEPLOYMENT_TARGET "15.0" CACHE STRING "MSL 3.2 needs macOS 15" FORCE)
+project(idun_native LANGUAGES C CXX OBJCXX)
+
+if(NOT APPLE)
+  message(FATAL_ERROR "The Metal runtime builds on macOS only; there is no CUDA backend yet.")
+endif()
+
+set(CMAKE_CXX_STANDARD 20)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_OBJCXX_STANDARD 20)
+set(CMAKE_OBJCXX_STANDARD_REQUIRED ON)
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
+
+option(IDUN_COVERAGE "Instrument native code for llvm-cov (make coverage-native)" OFF)
+set(IDUN_NATIVE_OUT "${CMAKE_SOURCE_DIR}/src/idun/_native" CACHE PATH "Where libidun_metal.dylib is written")
+if(SKBUILD)  # a wheel build (scikit-build-core): the build tree, then installed into the wheel, never the source tree
+  set(IDUN_NATIVE_OUT "${CMAKE_BINARY_DIR}/lib")
+endif()
+
+add_library(idun_metal SHARED native/metal/idun_metal.mm)
+target_include_directories(idun_metal PUBLIC native/include)
+target_compile_options(idun_metal PRIVATE -fobjc-arc -fvisibility=hidden -Wall -Wextra -Wpedantic -Wshadow
+                                            -Wconversion -Werror)
+target_link_libraries(idun_metal PRIVATE "-framework Metal" "-framework Foundation")
+set_target_properties(idun_metal PROPERTIES LIBRARY_OUTPUT_DIRECTORY "${IDUN_NATIVE_OUT}")
+
+if(IDUN_COVERAGE)
+  target_compile_options(idun_metal PRIVATE -O0 -fprofile-instr-generate -fcoverage-mapping)
+  target_link_options(idun_metal PRIVATE -fprofile-instr-generate)
+endif()
+
+if(SKBUILD)
+  install(TARGETS idun_metal LIBRARY DESTINATION idun/_native)
+endif()

idun-0.1.0/CMakePresets.json

@@ -0,0 +1,27 @@
+{
+  "version": 6,
+  "configurePresets": [
+    {
+      "name": "release",
+      "displayName": "Release: optimized, written into src/idun/_native",
+      "generator": "Ninja",
+      "binaryDir": "${sourceDir}/build/release",
+      "cacheVariables": { "CMAKE_BUILD_TYPE": "Release" }
+    },
+    {
+      "name": "coverage",
+      "displayName": "Coverage: llvm-cov instrumented, written into build/coverage",
+      "generator": "Ninja",
+      "binaryDir": "${sourceDir}/build/coverage",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Debug",
+        "IDUN_COVERAGE": "ON",
+        "IDUN_NATIVE_OUT": "${sourceDir}/build/coverage/lib"
+      }
+    }
+  ],
+  "buildPresets": [
+    { "name": "release", "configurePreset": "release" },
+    { "name": "coverage", "configurePreset": "coverage" }
+  ]
+}

idun-0.1.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

idun-0.1.0/NOTICE

@@ -0,0 +1,39 @@
+idun
+Copyright 2026 Anton Shaniuk
+
+Licensed under the Apache License, Version 2.0 (LICENSE).
+
+The Metal GEMM kernel (idun/nn/kernels/metal/common.metal) is adapted from MLX's Steel GEMM
+(https://github.com/ml-explore/mlx): its tile loader's thread layout and its fragment lane coordinates. MLX's copyright
+and permission notice:
+
+  MIT License
+
+  Copyright © 2023 Apple Inc.
+
+  Permission is hereby granted, free of charge, to any person obtaining a copy
+  of this software and associated documentation files (the "Software"), to deal
+  in the Software without restriction, including without limitation the rights
+  to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+  copies of the Software, and to permit persons to whom the Software is
+  furnished to do so, subject to the following conditions:
+
+  The above copyright notice and this permission notice shall be included in all
+  copies or substantial portions of the Software.
+
+  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+  OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+  SOFTWARE.
+
+idun_tabm implements TabM: Yury Gorishniy, Akim Kotelnikov and Artem Babenko, "TabM: Advancing Tabular Deep Learning
+with Parameter-Efficient Ensembling", ICLR 2025, arXiv:2410.24210.
+
+idun_realmlp implements RealMLP-TD: David Holzmüller, Léo Grinsztajn and Ingo Steinwart, "Better by Default: Strong
+Pre-Tuned MLPs and Boosted Trees on Tabular Data", NeurIPS 2024, arXiv:2407.04491.
+
+idun_gbm follows LightGBM's model: Guolin Ke et al., "LightGBM: A Highly Efficient Gradient Boosting Decision Tree",
+NeurIPS 2017.

idun-0.1.0/PKG-INFO

@@ -0,0 +1,90 @@
+Metadata-Version: 2.4
+Name: idun
+Version: 0.1.0
+Summary: Tabular learners for the Apple silicon GPU (Metal): idun_gbm, idun_tabm and idun_realmlp, with CPU reference backends.
+Keywords: gradient boosting,tabular,deep learning,TabM,RealMLP,Metal,Apple silicon,GPU
+Author: Anton Shaniuk
+License-Expression: Apache-2.0
+License-File: LICENSE
+License-File: NOTICE
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Science/Research
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: <3.15,>=3.12
+Requires-Dist: numpy<3,>=2.5
+Requires-Dist: numba<0.69,>=0.68
+Description-Content-Type: text/markdown
+
+# idun
+
+Tabular learning on the Apple silicon GPU. idun trains three learners on Metal, each with a CPU reference backend:
+
+- **idun_gbm** (`idun.gbm`): histogram gradient boosting with LightGBM's model (depth-wise or leaf-wise trees,
+  quantized gradients, categorical splits, seed bags). Its CPU and Metal backends build the same model, bit for bit.
+- **idun_tabm** (`idun.tabm`): TabM (Gorishniy et al., ICLR 2025), a parameter-efficient ensemble of MLPs.
+- **idun_realmlp** (`idun.realmlp`): RealMLP-TD (Holzmüller et al., NeurIPS 2024).
+
+Every fit is deterministic: the same data, parameters and seed give the same bits, run after run (the networks on the
+same GPU). The estimators follow scikit-learn's conventions without depending on it.
+
+idun is in alpha (0.1): the interfaces may still change.
+
+## Install
+
+```bash
+pip install idun            # Apple silicon with macOS 15 or later: the Metal wheel; elsewhere the CPU-only wheel
+pip install idun-numerai    # the Numerai port
+```
+
+Python 3.12–3.14; the only dependencies are NumPy and numba. Metal needs an Apple silicon Mac (M1 or newer) with
+macOS 15 or later. On Linux and Intel Macs the same package installs without the Metal library and runs the CPU
+backends; asking for Metal there raises `idun.errors.BackendUnavailableError` rather than falling back.
+
+## Quick start
+
+```python
+import idun
+from idun.gbm import GBMRegressor
+from idun.realmlp import RealMLPRegressor
+from idun.tabm import TabMClassifier
+
+gbm = GBMRegressor(num_trees=300, learning_rate=0.05).fit(X_train, y_train)
+pred = gbm.predict(X_test)
+cpu = GBMRegressor(num_trees=300, learning_rate=0.05, backend="cpu").fit(X_train, y_train)  # the same trees
+
+clf = TabMClassifier(categorical=["sector"]).fit(X_train, labels)  # any labels, two classes or more
+proba = clf.predict_proba(X_test)
+
+mlp = RealMLPRegressor(n_seeds=4).fit(X_train, y_train)  # a bag of four seeds; predicts their mean
+
+gbm.save("gbm.idun")
+same = idun.load("gbm.idun")
+```
+
+`X` is a 2-D NumPy array, a pandas DataFrame or an Arrow table; missing values are allowed. The backend is `"metal"`
+by default; pass `backend="cpu"` (idun_gbm) or `backend="reference"` (the networks' float64 reference) to run on the
+CPU. Nothing switches backends behind your back.
+
+Below the estimators: `idun.Dataset` bins a table once for reuse; the functional layer (`idun.gbm.train`,
+`idun.tabm.train_tabm`, `idun.realmlp.train_realmlp`, seed bags through `train_seeds` and `predict_mean`) trains on
+binned data or on `idun.data.PackedDataset`, a memory-mapped, date-sorted panel for data larger than memory, which
+Metal reads without a copy. The docstrings document every public name.
+
+## How it compares
+
+On five OpenML tasks on an Apple M1 (8 GB): idun_gbm's RMSE is within 3 % of the median of LightGBM, XGBoost and
+scikit-learn's HistGradientBoosting at the same nominal settings; idun_tabm and idun_realmlp are within 2 % of their
+reference implementations (the `tabm` package and pytabkit on Torch's MPS device) on every task and better on most, and
+idun_tabm fits in 0.45 of the reference's time.
+
+## Licence
+
+Apache-2.0. The NOTICE file lists the work idun builds on, including MLX's Steel GEMM (MIT), whose structure the
+networks' GEMM kernel adapts.

idun-0.1.0/README.md

@@ -0,0 +1,66 @@
+# idun
+
+Tabular learning on the Apple silicon GPU. idun trains three learners on Metal, each with a CPU reference backend:
+
+- **idun_gbm** (`idun.gbm`): histogram gradient boosting with LightGBM's model (depth-wise or leaf-wise trees,
+  quantized gradients, categorical splits, seed bags). Its CPU and Metal backends build the same model, bit for bit.
+- **idun_tabm** (`idun.tabm`): TabM (Gorishniy et al., ICLR 2025), a parameter-efficient ensemble of MLPs.
+- **idun_realmlp** (`idun.realmlp`): RealMLP-TD (Holzmüller et al., NeurIPS 2024).
+
+Every fit is deterministic: the same data, parameters and seed give the same bits, run after run (the networks on the
+same GPU). The estimators follow scikit-learn's conventions without depending on it.
+
+idun is in alpha (0.1): the interfaces may still change.
+
+## Install
+
+```bash
+pip install idun            # Apple silicon with macOS 15 or later: the Metal wheel; elsewhere the CPU-only wheel
+pip install idun-numerai    # the Numerai port
+```
+
+Python 3.12–3.14; the only dependencies are NumPy and numba. Metal needs an Apple silicon Mac (M1 or newer) with
+macOS 15 or later. On Linux and Intel Macs the same package installs without the Metal library and runs the CPU
+backends; asking for Metal there raises `idun.errors.BackendUnavailableError` rather than falling back.
+
+## Quick start
+
+```python
+import idun
+from idun.gbm import GBMRegressor
+from idun.realmlp import RealMLPRegressor
+from idun.tabm import TabMClassifier
+
+gbm = GBMRegressor(num_trees=300, learning_rate=0.05).fit(X_train, y_train)
+pred = gbm.predict(X_test)
+cpu = GBMRegressor(num_trees=300, learning_rate=0.05, backend="cpu").fit(X_train, y_train)  # the same trees
+
+clf = TabMClassifier(categorical=["sector"]).fit(X_train, labels)  # any labels, two classes or more
+proba = clf.predict_proba(X_test)
+
+mlp = RealMLPRegressor(n_seeds=4).fit(X_train, y_train)  # a bag of four seeds; predicts their mean
+
+gbm.save("gbm.idun")
+same = idun.load("gbm.idun")
+```
+
+`X` is a 2-D NumPy array, a pandas DataFrame or an Arrow table; missing values are allowed. The backend is `"metal"`
+by default; pass `backend="cpu"` (idun_gbm) or `backend="reference"` (the networks' float64 reference) to run on the
+CPU. Nothing switches backends behind your back.
+
+Below the estimators: `idun.Dataset` bins a table once for reuse; the functional layer (`idun.gbm.train`,
+`idun.tabm.train_tabm`, `idun.realmlp.train_realmlp`, seed bags through `train_seeds` and `predict_mean`) trains on
+binned data or on `idun.data.PackedDataset`, a memory-mapped, date-sorted panel for data larger than memory, which
+Metal reads without a copy. The docstrings document every public name.
+
+## How it compares
+
+On five OpenML tasks on an Apple M1 (8 GB): idun_gbm's RMSE is within 3 % of the median of LightGBM, XGBoost and
+scikit-learn's HistGradientBoosting at the same nominal settings; idun_tabm and idun_realmlp are within 2 % of their
+reference implementations (the `tabm` package and pytabkit on Torch's MPS device) on every task and better on most, and
+idun_tabm fits in 0.45 of the reference's time.
+
+## Licence
+
+Apache-2.0. The NOTICE file lists the work idun builds on, including MLX's Steel GEMM (MIT), whose structure the
+networks' GEMM kernel adapts.

idun-0.1.0/native/include/idun_metal.h

@@ -0,0 +1,111 @@
+/*
+ * idun_metal.h - C ABI of the Idun Metal runtime (libidun_metal.dylib).
+ *
+ * Python binds this header through ctypes (src/idun/metal/runtime.py). Conventions, which a CUDA backend would follow
+ * too (docs/architecture/backend-interface.md):
+ *   - every function is extern "C", takes and returns only C types, and never lets an exception or Objective-C
+ *     exception cross the boundary;
+ *   - fallible functions return IDUN_OK (0) on success, or a negative IDUN_ERR_* code with a NUL-terminated message
+ *     written to (err, err_len); err may be NULL when err_len is 0; functions returning a handle return NULL on
+ *     failure and write the message the same way;
+ *   - the caller owns the memory it passes in; the library owns the handles it returns until the matching
+ *     _destroy/_free; kernels belong to their context;
+ *   - IDUN_ABI_VERSION changes on every incompatible change, and the binding refuses to load a mismatched library.
+ */
+#ifndef IDUN_METAL_H
+#define IDUN_METAL_H
+
+#include <stddef.h>
+#include <stdint.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define IDUN_ABI_VERSION 6
+
+#define IDUN_OK 0
+#define IDUN_ERR_NO_DEVICE (-1)
+#define IDUN_ERR_COMPILE (-2)
+#define IDUN_ERR_ARGUMENT (-3)
+#define IDUN_ERR_STATE (-4) /* a call out of order: dispatch without a batch, a kernel before a library, ... */
+#define IDUN_ERR_GPU (-5)   /* a command buffer failed on the device */
+
+#define IDUN_NAME_LEN 128
+#define IDUN_MAX_BUFFERS 30  /* buffer indices 0..29; the params block goes to index n_buffers */
+#define IDUN_MAX_PARAMS 4096 /* setBytes limit */
+
+typedef struct idun_device_info {
+  char name[IDUN_NAME_LEN];
+  uint64_t max_threadgroup_memory_bytes;
+  uint64_t recommended_working_set_bytes;
+  uint64_t max_buffer_bytes;
+  uint32_t max_threads_per_threadgroup;
+  uint32_t page_size_bytes;
+  uint32_t has_unified_memory;
+  uint32_t supports_apple7; /* M1 family and newer: what the idun_gbm Metal backend needs */
+  uint32_t supports_apple9; /* M4 family: SIMD width 32, 32 KiB threadgroup memory */
+  uint32_t supports_metal4;
+} idun_device_info;
+
+typedef struct idun_context idun_context; /* device, queue, library, kernels, the open command batch */
+typedef struct idun_kernel idun_kernel;   /* a compute pipeline, owned by its context */
+typedef struct idun_buffer idun_buffer;   /* a shared-storage MTLBuffer */
+
+/* IDUN_ABI_VERSION of the compiled library. */
+int32_t idun_abi_version(void);
+
+/* The source text prepended to every kernel library before compilation (determinism.md: FP contraction off). */
+const char *idun_source_prelude(void);
+
+/* Describe the system default Metal device. */
+int32_t idun_probe(idun_device_info *out, char *err, size_t err_len);
+
+/* Compile MSL `source` with Idun's fixed compile options and prelude; on failure the compiler log goes to err. */
+int32_t idun_compile_check(const char *source, char *err, size_t err_len);
+
+/* A context on the system default device: one command queue, at most one library, at most one open batch. */
+idun_context *idun_context_create(char *err, size_t err_len);
+/* Destroy a context and its kernels (buffers are freed separately). Commits nothing: an open batch is dropped;
+ * committed batches are waited for (their errors are not reported). */
+void idun_context_destroy(idun_context *ctx);
+
+/* Compile `source` (with the prelude and fixed options) as the context's library; replaces kernels of a previous one.
+ */
+int32_t idun_library_load(idun_context *ctx, const char *source, char *err, size_t err_len);
+/* The compute pipeline of kernel function `name` of the loaded library. */
+idun_kernel *idun_kernel_get(idun_context *ctx, const char *name, char *err, size_t err_len);
+/* Largest threadgroup the pipeline accepts, and its SIMD width. */
+uint32_t idun_kernel_max_threads(const idun_kernel *kernel);
+uint32_t idun_kernel_simd_width(const idun_kernel *kernel);
+
+/* A new zero-filled shared buffer of `bytes` (> 0); its CPU address goes to *contents. */
+idun_buffer *idun_buffer_alloc(idun_context *ctx, size_t bytes, void **contents, char *err, size_t err_len);
+/* Wrap caller memory without copying: `page_aligned` and `bytes` must be multiples of the page size. The memory must
+ * stay valid and mapped until idun_buffer_free; the GPU may read it (and write it, if the mapping is writable). */
+idun_buffer *idun_buffer_wrap(idun_context *ctx, void *page_aligned, size_t bytes, char *err, size_t err_len);
+void idun_buffer_free(idun_buffer *buffer);
+
+/* Open a command batch: one command buffer with one serial compute encoder. */
+int32_t idun_batch_begin(idun_context *ctx, char *err, size_t err_len);
+/* Encode one dispatch into the open batch: buffers at indices 0..n_buffers-1, `params` (if any) at index n_buffers,
+ * `threadgroup_bytes` (a multiple of 16; 0 for none) of dynamic threadgroup memory at threadgroup index 0, `groups`
+ * threadgroups of `threads` threads. Dispatches of one batch run in order. */
+int32_t idun_dispatch(idun_context *ctx, const idun_kernel *kernel, idun_buffer *const *buffers, uint32_t n_buffers,
+                      const void *params, size_t params_bytes, uint32_t threadgroup_bytes, const uint32_t groups[3],
+                      const uint32_t threads[3], char *err, size_t err_len);
+/* Commit the open batch without waiting (v5). Batches run in the order they are committed, each after the one before
+ * it finished, so the host can encode the next batch while the GPU runs this one. */
+int32_t idun_batch_commit(idun_context *ctx, char *err, size_t err_len);
+/* Wait, oldest first, until at most `keep` committed batches are left unwaited (v5); report the GPU time of those
+ * waited for (sum of GPUEndTime - GPUStartTime). Returns IDUN_ERR_GPU with the first failure's message if any failed.
+ */
+int32_t idun_batch_wait(idun_context *ctx, uint32_t keep, double *gpu_seconds, char *err, size_t err_len);
+/* Commit the open batch and wait for it and every batch committed before it; report their GPU time. */
+int32_t idun_batch_commit_wait(idun_context *ctx, double *gpu_seconds, char *err, size_t err_len);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* IDUN_METAL_H */