gpulayer 0.3.0__tar.gz

gpulayer-0.3.0/.gitignore

@@ -0,0 +1,11 @@
+__pycache__/
+*.pyc
+.venv/
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+dist/
+*.egg-info/
+.gpulayer/
+examples/*/outputs/
+gpulayer_runs/

gpulayer-0.3.0/CHANGELOG.md

@@ -0,0 +1,139 @@
+# Changelog
+
+All notable changes to GPUlayer. Versions follow [semantic versioning](https://semver.org);
+before 1.0, minor versions may change public APIs, with migration notes here.
+
+## 0.3.0 (2026-10-30)
+
+First release on PyPI (`pip install gpulayer`), under the Apache-2.0 licence. Tested on 30
+machines in two GCP regions, with a multi-machine training comparison against PyTorch
+Lightning.
+
+### Scale and speed
+
+- Scheduling no longer slows down as the queue grows: a submit costs 0.5 ms at 2,000 queued
+  jobs instead of 44 ms (`benchmarks/submit_queue.py`).
+- Behind `gpulayer up`, idle workers wait until something changes instead of re-checking
+  every 0.2 s, and a change wakes only the workers or job waiters it concerns. On 30 workers
+  a 480-run training sweep finished in 45 s instead of 102 s.
+- SQLite runs with `synchronous=NORMAL` (WAL): commits no longer wait for the disk (2.25 ms
+  to 0.02 ms on a GCP standard disk). Consistent after a crash; a power loss can roll back
+  the last commits.
+- The Trainer does less work per step (0.60 to 0.585 ms on a small CPU model; a plain
+  PyTorch loop is 0.55).
+- Validated on 30 workers in two regions: distributed jobs on up to 16 machines (32
+  processes), fail-fast across 8 machines in 4.9 s, retries after killing workers mid-run.
+  One training job on 8 machines ran 3.4x faster than on one, and 7% sooner than PyTorch
+  Lightning under torchrun (benchmarks/results/2026-10-30-gcp-fleet).
+- Known limit: bursts of very short jobs reach about 25 jobs/s, bounded by the server using
+  one CPU core (ADR-0016, item 6).
+
+### Features and fixes
+
+- `--gpus 0,1|none` on `gpulayer up`, `join` and `worker start`: a worker offers only those
+  GPUs, so several workers on one machine no longer offer the same GPU (the known issue in
+  0.2.0). `--gpus none` makes a CPU-only worker on a GPU machine.
+- CUDA jobs get `CUDA_DEVICE_ORDER=PCI_BUS_ID`, so `CUDA_VISIBLE_DEVICES` selects the GPUs
+  that were reserved (`nvidia-smi` numbering) on machines with mixed GPU models.
+- Trainer: `max_steps`, `check_val_every_n_epoch`, `trainer.test()` with `Module.test_step`
+  and `trainer.predict()` with `Module.predict_step` (ADR-0018).
+- `ModelCheckpoint(save_top_k=N)` keeps the N best epochs (also after remote runs), and a
+  `LearningRateMonitor` callback records each epoch's learning rate.
+- `val_check_interval` (batches or a fraction of the epoch) and `metrics.csv` next to
+  `metrics.json`, with step, mid-epoch validation and epoch rows.
+- A launched training run returns its metrics even with checkpointing off (`metrics.json` is
+  written every epoch).
+- Apache-2.0 licence (ADR-0012), PyPI package metadata, and a trusted-publishing release
+  workflow (`.github/workflows/publish.yml`).
+- Benchmarks: `benchmarks/cloud/gcp.sh` builds an image and starts a server plus a fleet of
+  workers; `benchmarks/scale_test.py`; `benchmarks/lightning_comparison/scaling_run.py` and
+  `multinode.py` compare multi-machine training with PyTorch Lightning.
+
+## 0.2.0 (2026-10-30)
+
+Validated on an NVIDIA GPU and across machines, faster distributed start-up and HTTP API, and
+a measured comparison with PyTorch Lightning.
+
+### Migration from 0.1.0
+
+- Distributed jobs default to `launcher="direct"`. Upgrade the server and the workers
+  together: 0.1.0 workers are not chosen for `direct` jobs, and a 0.1.0 server rejects them.
+  Pass `launcher="torchrun"` (`--launcher torchrun`) to keep the old behaviour.
+
+### Changes
+
+- `gpulayer mcp`: a Model Context Protocol server on stdio, so AI and coding agents can list
+  resources, submit commands (with a project directory and inputs), wait, read logs, cancel
+  and download outputs; `--read-only` for monitoring. Tested with the official MCP Python SDK.
+- Agent Skill for the CLI in `integrations/skills/gpulayer`.
+- Guide: AI and coding agents.
+- Trainer callbacks: `Callback` hooks, `EarlyStopping` and `ModelCheckpoint` (best checkpoint,
+  also returned from remote runs); when one DDP rank asks to stop, every rank stops after the
+  same epoch.
+- Faster distributed jobs: `DistributedJob.launcher="direct"` is the new default. Workers
+  start the ranks themselves, with the same environment torchrun sets, instead of through
+  torchrun's elastic agent: a 2 x 2 job on one Mac takes 0.93 s instead of about 2.1 s.
+  `launcher="torchrun"` (`--launcher torchrun`) keeps the old behaviour.
+  Migration: workers and the server must both be upgraded to run `direct` jobs. v0.1.0
+  workers report only `torchrun` and are not chosen for them; a v0.1.0 server rejects the
+  new value, so submit with `launcher="torchrun"` there.
+- Faster HTTP API: clients and workers keep their connection open, and the server sets
+  `TCP_NODELAY` and runs TLS handshakes per connection. On Linux a request dropped from 43 ms
+  to 0.6 ms, and 500 small jobs over two machines ran 7.1x faster (7.8 to 55.6 jobs/s).
+- Validated on NVIDIA: a Tesla T4 on GCP (CUDA detection, CUDA jobs, Trainer in fp32, bf16,
+  fp16 and with `torch.compile`, GPU reservation), and on two machines over a real network
+  (distributed jobs across both, fail-fast, hundreds of jobs). `scripts/validate_gpu.sh`
+  repeats the single-machine part on any NVIDIA machine.
+- Performance and comparison: a docs page and README section with charts
+  (`scripts/make_perf_charts.py`), the GPUlayer vs PyTorch Lightning harness
+  (`benchmarks/lightning_comparison/`), `benchmarks/multi_machine.py`, raw results in
+  `benchmarks/results/2026-10-30-gcp/`, and `benchmarks/cloud/gcp.sh` to rerun them on Google
+  Cloud.
+- NVIDIA CUDA independent jobs are now `hardware-validated` in the support matrix.
+
+## 0.1.0 (2026-10-29)
+
+The first release. GPUlayer runs commands, Python functions and distributed jobs on the
+machines you can reach, from a laptop to servers that joined a control plane.
+
+### Run work
+
+- Job types: commands in any language, Python functions (`@gl.remote`, `.map()`), and
+  distributed jobs launched with torchrun (static rendezvous, gang scheduling, fail-fast).
+- PyTorch integration: `Module`, `Trainer` (in-process, multi-process DDP, multi-node,
+  remote), `Runtime` and `launch()`; gradient accumulation with DDP `no_sync`.
+- Runtime profiles (`pytorch-cuda`, `pytorch-mps`, `jax-cuda`, `cupy-cuda`, `mlx`, ...).
+
+### Place and run it well
+
+- Capability-aware scheduler with per-worker reasons when a job cannot run.
+- Several tasks per worker, each with its own reserved GPUs (default one slot per NVIDIA or
+  AMD GPU); `spread` placement keeps machines evenly busy.
+- Event-driven dispatch: long-poll worker acquire and job waits, background log shipping.
+  One job on an idle worker: 626 ms to 31 ms; burst throughput 27 to 87 jobs/s (one Mac).
+- Per-rank log streams for distributed jobs.
+
+### Servers and workers
+
+- `gpulayer up` (control plane with TLS and a pinned self-signed certificate) and
+  `gpulayer join`; named, expiring, revocable tokens with roles (admin, submitter, join,
+  metrics); rate-limited logins; worker artifact scoping; retention and `gpulayer gc`.
+
+### Observe
+
+- Web dashboard: jobs, live logs, placement reasons, workers, and activity charts (jobs,
+  CPU load, memory, task processes, NVIDIA GPU utilisation).
+- Prometheus `/metrics`, a Grafana dashboard and a Prometheus/Grafana Compose setup.
+
+### Install
+
+- `install.sh` (curl one-liner, private virtualenv, no sudo), a Docker image for server and
+  workers (CPU or CUDA base) with Compose, and CI that tests all of it.
+
+### Validation status
+
+- Tested on CPU and Apple Silicon (PyTorch MPS and MLX), with several workers on one host,
+  and in Linux CI including a Docker server/worker test.
+- NVIDIA CUDA execution and multi-machine runs are implemented but not validated on hardware.
+- Cloud provisioning is simulated (`FakeProvider`); cloud VMs can join with `gpulayer join`.
+- Not a sandbox for untrusted code. See docs/project/status.md.

gpulayer-0.3.0/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.

gpulayer-0.3.0/PKG-INFO

@@ -0,0 +1,216 @@
+Metadata-Version: 2.5
+Name: gpulayer
+Version: 0.3.0
+Summary: Run any workload on every GPU and CPU you can reach: job orchestration across local, remote and cloud machines, distributed PyTorch, a web dashboard and an MCP server for AI agents.
+Project-URL: Homepage, https://www.gpulayer.dev
+Project-URL: Documentation, https://www.gpulayer.dev/docs/
+Project-URL: Source, https://github.com/antonellof/gpulayer
+Project-URL: Changelog, https://github.com/antonellof/gpulayer/blob/main/CHANGELOG.md
+Project-URL: Issues, https://github.com/antonellof/gpulayer/issues
+Author: Antonello
+License-Expression: Apache-2.0
+License-File: LICENSE
+Keywords: apple-silicon,cuda,distributed,gpu,mcp,mlops,orchestration,pytorch,scheduler
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: System :: Distributed Computing
+Requires-Python: >=3.10
+Requires-Dist: packaging>=23
+Requires-Dist: pydantic<3,>=2.5
+Provides-Extra: dev
+Requires-Dist: mypy>=1.8; extra == 'dev'
+Requires-Dist: prometheus-client>=0.20; extra == 'dev'
+Requires-Dist: pytest>=7.4; extra == 'dev'
+Requires-Dist: ruff>=0.4; extra == 'dev'
+Provides-Extra: examples
+Requires-Dist: mlx>=0.10; (sys_platform == 'darwin' and platform_machine == 'arm64') and extra == 'examples'
+Requires-Dist: numpy>=1.20; extra == 'examples'
+Requires-Dist: torch>=2.1; extra == 'examples'
+Provides-Extra: torch
+Requires-Dist: torch>=2.1; extra == 'torch'
+Description-Content-Type: text/markdown
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/logo.svg" alt="GPUlayer" width="300">
+</p>
+
+<p align="center">
+  <b>Run any workload on every GPU and CPU you can reach.</b><br>
+  <a href="#install">Install</a> ·
+  <a href="#quickstart">Quickstart</a> ·
+  <a href="#for-ai-and-coding-agents">AI agents</a> ·
+  <a href="#why-gpulayer">Why GPUlayer</a> ·
+  <a href="#performance-and-comparison">Performance</a> ·
+  <a href="#architecture">Architecture</a> ·
+  <a href="https://www.gpulayer.dev/docs/">Docs</a> ·
+  <a href="https://github.com/antonellof/gpulayer/tree/main/examples">Examples</a> ·
+  <a href="https://github.com/antonellof/gpulayer/blob/main/CHANGELOG.md">Changelog</a>
+</p>
+
+GPUlayer sends a command, a Python function or a distributed job to machines that can run
+it (this one, remote servers, clusters, data centers, cloud VMs, or several at once), starts
+it there and brings back logs, outputs and results. PyTorch gets a deeper integration on top.
+Documentation: [gpulayer.dev/docs](https://www.gpulayer.dev/docs/).
+
+> **v0.3.0.** Tested on CPU, Apple Silicon (PyTorch MPS and MLX) and an
+> NVIDIA Tesla T4, and on up to 30 machines on Google Cloud. Several GPUs per machine and NCCL
+> across machines are implemented but not validated yet; cloud provisioning is simulated. See
+> [Status and roadmap](https://github.com/antonellof/gpulayer/blob/main/docs/project/status.md).
+
+## Install
+
+```bash
+pip install gpulayer
+```
+
+That's it. Check what this machine can run:
+
+```bash
+gpulayer resources detect
+```
+
+To train with PyTorch, install it too: `pip install "gpulayer[torch]"`.
+
+<details>
+<summary>Other ways to install</summary>
+
+| Method | Command |
+|---|---|
+| As a standalone CLI with [uv](https://docs.astral.sh/uv/) | `uv tool install gpulayer` |
+| Install script (own virtual environment, no sudo) | `curl -fsSL https://raw.githubusercontent.com/antonellof/gpulayer/main/install.sh \| sh` |
+| Docker | `docker build -t gpulayer .` (see [deploy/docker](https://github.com/antonellof/gpulayer/tree/main/deploy/docker)) |
+| From source, to run the examples | `git clone https://github.com/antonellof/gpulayer && cd gpulayer && pip install -e ".[dev,examples]"` |
+
+More in [Installation](https://github.com/antonellof/gpulayer/blob/main/docs/get-started/installation.md).
+</details>
+
+## Quickstart
+
+```bash
+# Any command as a job: code and inputs shipped, outputs returned
+gpulayer submit --accelerator cuda --source ./app -- ./render.sh
+
+# PyTorch: the same script in this process, as 2-process DDP, or on 2 nodes
+python examples/trainer_regression/train.py
+python examples/trainer_regression/train.py --accelerator cpu --devices 2
+python examples/trainer_regression/train.py --accelerator cpu --nodes 2 --devices 2
+
+# More machines: run this on the server, then paste the printed `gpulayer join` line elsewhere
+gpulayer up
+
+# See what ran
+gpulayer dashboard
+```
+
+```python
+import gpulayer as gl
+
+@gl.remote(accelerator="mps", frameworks={"mlx": ">=0.10"})   # any library, any GPU
+def benchmark(n: int) -> dict: ...
+
+benchmark.map([1024, 2048, 4096])                           # in parallel
+```
+
+![GPUlayer dashboard](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/dashboard/overview-dark.png)
+
+The [full quickstart](https://github.com/antonellof/gpulayer/blob/main/docs/get-started/quickstart.md) walks through each step with its output.
+
+## For AI and coding agents
+
+`gpulayer mcp` is an MCP server, so Claude Code, Claude Desktop, Cursor, Codex and VS Code
+agents can list GPUs, submit jobs, wait, read logs and download outputs. An
+[Agent Skill](https://github.com/antonellof/gpulayer/blob/main/integrations/skills/gpulayer/SKILL.md) covers agents that use the CLI.
+
+```bash
+claude mcp add gpulayer -- gpulayer mcp
+```
+
+See [AI and coding agents](https://github.com/antonellof/gpulayer/blob/main/docs/guides/ai-agents.md).
+
+## Why GPUlayer
+
+Most GPU code assumes the GPU is on the machine where you run it. With GPUlayer you describe
+what a workload needs (an accelerator, GPU memory, frameworks) and it runs on a machine that
+has it, wherever that machine is.
+
+| Situation | Without GPUlayer | With GPUlayer |
+|---|---|---|
+| Code on a laptop, GPU elsewhere | ssh, copy code, rebuild the environment, tmux, copy results back | `gpulayer submit --target remote -- <cmd>` |
+| Several GPU machines | track free machines by hand; jobs crash on the wrong GPU or library | one queue; each job lands where it can run, or is rejected with the reason |
+| NVIDIA servers and Apple Silicon Macs | separate scripts and setup per machine | ask for `cuda` or `mps`; one interface |
+| Many independent tasks | loops and ad hoc scripts | `fn.map([...])` across free GPUs |
+| Multi-node training | `torchrun` on every node with matching ranks and addresses | `Trainer(num_nodes=2, devices=4)` |
+
+On one machine with one GPU, running your code directly is simpler. GPUlayer runs jobs that
+finish (long-running services are planned) and is not a sandbox for untrusted code. More in
+[Why GPUlayer](https://www.gpulayer.dev/docs/#why-gpulayer).
+
+## Performance and comparison
+
+Measured on one NVIDIA Tesla T4 against PyTorch Lightning, with the same model, data and
+optimizer:
+
+![Training step time: GPUlayer and PyTorch Lightning](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/perf/training-step-light.svg#gh-light-mode-only)
+![Training step time: GPUlayer and PyTorch Lightning](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/perf/training-step-dark.svg#gh-dark-mode-only)
+
+- GPUlayer's Trainer is 1.3 to 1.4x faster per step on small models and the same on a
+  compute-bound CNN, and starts about 0.6 s sooner.
+- One training job on 8 machines runs 3.4x faster than on one, and GPUlayer finishes it 7%
+  sooner than Lightning under torchrun (39.9 s vs 42.8 s; 9% sooner on one machine).
+- Distributed jobs start 1.1 to 2.3 s faster than through torchrun; tested on up to 16
+  machines (32 processes) and 30 workers across two GCP regions.
+- Across two machines: a distributed job spanning both, fail-fast in 2.7 s, and 55.6 small
+  jobs per second.
+
+![One training job on more machines](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/perf/strong-scaling-light.svg#gh-light-mode-only)
+![One training job on more machines](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/perf/strong-scaling-dark.svg#gh-dark-mode-only)
+
+Method, all charts and raw logs:
+[Performance and comparison](https://github.com/antonellof/gpulayer/blob/main/docs/get-started/performance-and-comparison.md). Reproduce them
+with [`benchmarks/`](https://github.com/antonellof/gpulayer/blob/main/benchmarks/README.md).
+
+## Architecture
+
+![Your workloads go through GPUlayer to the machines you can reach](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/architecture-light.svg#gh-light-mode-only)
+![Your workloads go through GPUlayer to the machines you can reach](https://raw.githubusercontent.com/antonellof/gpulayer/main/docs/assets/architecture-dark.svg#gh-dark-mode-only)
+
+GPUlayer has three parts:
+
+- **Clients** submit work: the `gpulayer` CLI, the Python SDK (`@gl.remote`, `gl.Client`), the
+  PyTorch `Trainer` and `launch()`, and the MCP server for AI agents. A job states what it
+  needs (accelerator, GPU count and memory, frameworks), not which machine to use.
+- **The control plane** keeps the job queue, picks a worker for each job and records the
+  reason when no worker fits. It stores code bundles, inputs, outputs and logs, and checks
+  tokens. For work on one machine it runs inside your process; for several machines it runs
+  as a server (`gpulayer up`).
+- **Workers**, one per machine, report their GPUs and installed frameworks, ask the control
+  plane for tasks, run them as normal processes (one per rank for distributed jobs),
+  stream logs back and upload outputs. Workers open every connection themselves, so they can
+  sit behind NAT. A cloud VM is a worker that joined with `gpulayer join`.
+
+A job goes from `queued` to `scheduled` when a worker fits, then `running`, and ends as
+`succeeded`, `failed`, `cancelled` or `timed_out`. Frameworks such as PyTorch, JAX, MLX and
+CuPy still decide how the computation uses the GPU; GPUlayer decides which GPU runs it. More in
+[Architecture overview](https://github.com/antonellof/gpulayer/blob/main/docs/architecture/overview.md).
+
+## Development
+
+```bash
+uv pip install -e '.[dev,examples]'
+.venv/bin/pytest && .venv/bin/ruff check src tests && .venv/bin/mypy src
+```
+
+See [Contributing](https://github.com/antonellof/gpulayer/blob/main/docs/project/contributing.md).
+
+## License
+
+Apache License 2.0, see [LICENSE](https://github.com/antonellof/gpulayer/blob/main/LICENSE).