atomr-accel 0.1.0__tar.gz

atomr_accel-0.1.0/PKG-INFO

@@ -0,0 +1,140 @@
+Metadata-Version: 2.4
+Name: atomr-accel
+Version: 0.1.0
+Classifier: Development Status :: 4 - Beta
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Rust
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Topic :: Scientific/Engineering
+Classifier: Topic :: Software Development :: Libraries
+Requires-Dist: numpy>=1.24
+Requires-Dist: pytest>=7 ; extra == 'test'
+Requires-Dist: numpy>=1.24 ; extra == 'test'
+Provides-Extra: test
+Summary: Actor-shaped face for compute-acceleration backends (NVIDIA CUDA shipping; ROCm / Metal / oneAPI / Vulkan future), on the atomr actor runtime.
+Keywords: cuda,gpu,atomr,actor,ml,accelerator
+Home-Page: https://github.com/rustakka/atomr-accel
+Author: atomr-accel contributors
+License: Apache-2.0
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
+Project-URL: Source, https://github.com/rustakka/atomr-accel
+
+# atomr-accel (Python)
+
+Python bindings for [atomr-accel](../..) — drive an actor-supervised
+NVIDIA CUDA pipeline directly from Python without juggling streams,
+contexts, or hand-rolled retry loops.
+
+```python
+import numpy as np
+import atomr_accel
+
+with atomr_accel.System.open("my-app") as sys:
+    dev = sys.spawn_device(device_id=0)         # real CUDA device
+
+    # Allocate two N×N f32 buffers on-device.
+    n = 256
+    a = dev.allocate_f32(n * n)
+    b = dev.allocate_f32(n * n)
+    c = dev.allocate_f32(n * n)
+
+    # Upload from numpy.
+    dev.copy_from_numpy(a, np.ones(n * n, dtype=np.float32))
+    dev.copy_from_numpy(b, np.full(n * n, 2.0, dtype=np.float32))
+
+    # Run cuBLAS SGEMM — the call blocks until the kernel finishes.
+    dev.sgemm(a, b, c, m=n, n=n, k=n, alpha=1.0, beta=0.0)
+
+    # Pull the result back into a fresh numpy array.
+    result = dev.copy_to_numpy(c)
+    print(result.reshape(n, n))
+```
+
+For hosts without a GPU, pass `mock=True` to `spawn_device` and the
+device replies `Unrecoverable("...mock mode")` for any kernel call —
+useful for testing the surrounding plumbing in CI.
+
+## Install
+
+The wheel builds with [maturin](https://www.maturin.rs/):
+
+```bash
+# from this directory
+pip install maturin pytest numpy
+maturin develop --release       # builds + installs into the active venv
+pytest tests/                    # runs the no-GPU smoke suite
+```
+
+For a release wheel:
+
+```bash
+maturin build --release --no-default-features --features extension-module
+# wheel lands in target/wheels/
+```
+
+The pure-Python facade (`python/atomr_accel/__init__.py`) re-exports
+the native classes and exception types. Downstream libraries import
+from `atomr_accel` and treat `atomr_accel._native` as private.
+
+## Feature flags
+
+The Rust crate matches `atomr-accel`'s feature gating so the wheel can
+be built minimal or full:
+
+| Feature           | Adds                                            |
+|-------------------|-------------------------------------------------|
+| (default)         | `System`, `Device`, `GpuBuffer`, exceptions     |
+| `curand`          | `RngGenerator`                                  |
+| `nvrtc`           | `NvrtcKernel`                                   |
+| `cudnn` / `cufft` / `cusolver` / `cublaslt` / `nccl` | placeholder; future Python surfaces |
+| `core-libs` / `training-libs` / `full-cuda`         | aggregates                          |
+
+```bash
+maturin develop --features atomr-accel-py/curand,atomr-accel-py/nvrtc
+```
+
+## Public API
+
+| Class / function                | What it wraps                                       |
+|---------------------------------|-----------------------------------------------------|
+| `atomr_accel.System.open(name)`  | A `atomr_core::actor::ActorSystem` lifetime         |
+| `system.spawn_device(id, mock=)` | A `DeviceActor` (real or mock)                      |
+| `device.allocate_f32(len)`      | `DeviceMsg::AllocateF32` → `GpuBuffer`              |
+| `device.copy_from_numpy(buf, np)` | H2D `CopyFromHostF32`                              |
+| `device.copy_to_numpy(buf)`     | D2H `CopyToHostF32` → numpy `float32` array         |
+| `device.sgemm(a,b,c,m,n,k,...)` | cuBLAS SGEMM via `BlasActor`                        |
+| `device.stats()`                | `DeviceMsg::Stats` → `DeviceLoad`                   |
+| `GpuBuffer.is_stale()`          | Generation token check vs. `DeviceState`            |
+| `GpuRuntimeError` (and subclasses) | Typed `GpuError` mapping                         |
+
+Every method blocks the calling thread until the underlying actor
+replies (the GIL is released for the duration via `py.allow_threads`).
+Async wrappers can be layered later via
+`pyo3_async_runtimes::tokio::future_into_py`.
+
+## How it works
+
+Three pieces:
+
+1. **A shared tokio runtime.** The first call to
+   `System.open(...)` initializes a multi-threaded scheduler; every
+   subsequent call reuses it. Implemented in `src/runtime.rs` via
+   `pyo3-async-runtimes::tokio::init`.
+2. **The `_native` extension module.** `src/lib.rs` registers
+   `System`, `Device`, `GpuBuffer`, exceptions, and (feature-gated)
+   `RngGenerator` / `NvrtcKernel`. Each Python class wraps a typed
+   `ActorRef<...>` from atomr-accel and converts replies via
+   `errors::map_gpu`.
+3. **The pure-Python facade** at `python/atomr_accel/__init__.py`. Hides
+   `_native`; documents the API; gives downstream libraries a stable
+   import path.
+
+See [docs/python-bridge.md](../../docs/python-bridge.md) for the
+full architecture write-up.
+
+## License
+
+Apache-2.0.
+

atomr_accel-0.1.0/README.md

@@ -0,0 +1,116 @@
+# atomr-accel (Python)
+
+Python bindings for [atomr-accel](../..) — drive an actor-supervised
+NVIDIA CUDA pipeline directly from Python without juggling streams,
+contexts, or hand-rolled retry loops.
+
+```python
+import numpy as np
+import atomr_accel
+
+with atomr_accel.System.open("my-app") as sys:
+    dev = sys.spawn_device(device_id=0)         # real CUDA device
+
+    # Allocate two N×N f32 buffers on-device.
+    n = 256
+    a = dev.allocate_f32(n * n)
+    b = dev.allocate_f32(n * n)
+    c = dev.allocate_f32(n * n)
+
+    # Upload from numpy.
+    dev.copy_from_numpy(a, np.ones(n * n, dtype=np.float32))
+    dev.copy_from_numpy(b, np.full(n * n, 2.0, dtype=np.float32))
+
+    # Run cuBLAS SGEMM — the call blocks until the kernel finishes.
+    dev.sgemm(a, b, c, m=n, n=n, k=n, alpha=1.0, beta=0.0)
+
+    # Pull the result back into a fresh numpy array.
+    result = dev.copy_to_numpy(c)
+    print(result.reshape(n, n))
+```
+
+For hosts without a GPU, pass `mock=True` to `spawn_device` and the
+device replies `Unrecoverable("...mock mode")` for any kernel call —
+useful for testing the surrounding plumbing in CI.
+
+## Install
+
+The wheel builds with [maturin](https://www.maturin.rs/):
+
+```bash
+# from this directory
+pip install maturin pytest numpy
+maturin develop --release       # builds + installs into the active venv
+pytest tests/                    # runs the no-GPU smoke suite
+```
+
+For a release wheel:
+
+```bash
+maturin build --release --no-default-features --features extension-module
+# wheel lands in target/wheels/
+```
+
+The pure-Python facade (`python/atomr_accel/__init__.py`) re-exports
+the native classes and exception types. Downstream libraries import
+from `atomr_accel` and treat `atomr_accel._native` as private.
+
+## Feature flags
+
+The Rust crate matches `atomr-accel`'s feature gating so the wheel can
+be built minimal or full:
+
+| Feature           | Adds                                            |
+|-------------------|-------------------------------------------------|
+| (default)         | `System`, `Device`, `GpuBuffer`, exceptions     |
+| `curand`          | `RngGenerator`                                  |
+| `nvrtc`           | `NvrtcKernel`                                   |
+| `cudnn` / `cufft` / `cusolver` / `cublaslt` / `nccl` | placeholder; future Python surfaces |
+| `core-libs` / `training-libs` / `full-cuda`         | aggregates                          |
+
+```bash
+maturin develop --features atomr-accel-py/curand,atomr-accel-py/nvrtc
+```
+
+## Public API
+
+| Class / function                | What it wraps                                       |
+|---------------------------------|-----------------------------------------------------|
+| `atomr_accel.System.open(name)`  | A `atomr_core::actor::ActorSystem` lifetime         |
+| `system.spawn_device(id, mock=)` | A `DeviceActor` (real or mock)                      |
+| `device.allocate_f32(len)`      | `DeviceMsg::AllocateF32` → `GpuBuffer`              |
+| `device.copy_from_numpy(buf, np)` | H2D `CopyFromHostF32`                              |
+| `device.copy_to_numpy(buf)`     | D2H `CopyToHostF32` → numpy `float32` array         |
+| `device.sgemm(a,b,c,m,n,k,...)` | cuBLAS SGEMM via `BlasActor`                        |
+| `device.stats()`                | `DeviceMsg::Stats` → `DeviceLoad`                   |
+| `GpuBuffer.is_stale()`          | Generation token check vs. `DeviceState`            |
+| `GpuRuntimeError` (and subclasses) | Typed `GpuError` mapping                         |
+
+Every method blocks the calling thread until the underlying actor
+replies (the GIL is released for the duration via `py.allow_threads`).
+Async wrappers can be layered later via
+`pyo3_async_runtimes::tokio::future_into_py`.
+
+## How it works
+
+Three pieces:
+
+1. **A shared tokio runtime.** The first call to
+   `System.open(...)` initializes a multi-threaded scheduler; every
+   subsequent call reuses it. Implemented in `src/runtime.rs` via
+   `pyo3-async-runtimes::tokio::init`.
+2. **The `_native` extension module.** `src/lib.rs` registers
+   `System`, `Device`, `GpuBuffer`, exceptions, and (feature-gated)
+   `RngGenerator` / `NvrtcKernel`. Each Python class wraps a typed
+   `ActorRef<...>` from atomr-accel and converts replies via
+   `errors::map_gpu`.
+3. **The pure-Python facade** at `python/atomr_accel/__init__.py`. Hides
+   `_native`; documents the API; gives downstream libraries a stable
+   import path.
+
+See [docs/python-bridge.md](../../docs/python-bridge.md) for the
+full architecture write-up.
+
+## License
+
+Apache-2.0.

atomr_accel-0.1.0/atomr/Cargo.toml

@@ -0,0 +1,198 @@
+[workspace]
+resolver = "2"
+members = [
+    "crates/atomr",
+    "crates/atomr-core",
+    "crates/atomr-config",
+    "crates/atomr-macros",
+    "crates/atomr-testkit",
+    "crates/atomr-remote",
+    "crates/atomr-cluster",
+    "crates/atomr-cluster-tools",
+    "crates/atomr-cluster-sharding",
+    "crates/atomr-cluster-metrics",
+    "crates/atomr-distributed-data",
+    "crates/atomr-persistence",
+    "crates/atomr-persistence-query",
+    "crates/atomr-persistence-tck",
+    "crates/atomr-persistence-query-inmemory",
+    "crates/atomr-persistence-sql",
+    "crates/atomr-persistence-redis",
+    "crates/atomr-persistence-mongodb",
+    "crates/atomr-persistence-cassandra",
+    "crates/atomr-persistence-aws",
+    "crates/atomr-persistence-azure",
+    "crates/atomr-streams",
+    "crates/atomr-telemetry",
+    "crates/atomr-dashboard",
+    "crates/atomr-coordination",
+    "crates/atomr-discovery",
+    "crates/atomr-di",
+    "crates/atomr-hosting",
+    "crates/atomr-serialization-hyperion",
+    "xtask",
+    "examples/pingpong",
+    "examples/chat",
+    "examples/fault-tolerance",
+    "examples/event-sourced-counter",
+    "examples/cluster-pubsub-chat",
+    "examples/sharded-keyvalue",
+    "benches/ping_throughput",
+    "crates/atomr-profiler",
+    "crates/py-bindings/pycore",
+    "crates/py-bindings/pytestkit",
+    "crates/py-bindings/pyremote",
+    "crates/py-bindings/pycluster",
+    "crates/py-bindings/pycluster-tools",
+    "crates/py-bindings/pycluster-sharding",
+    "crates/py-bindings/pyddata",
+    "crates/py-bindings/pypersistence",
+    "crates/py-bindings/pystreams",
+    "crates/py-bindings/pycoordination",
+    "crates/py-bindings/pydiscovery",
+    "crates/py-bindings/pydi",
+    "crates/py-bindings/pyhosting",
+]
+exclude = ["akka.net"]
+
+[workspace.package]
+version = "0.1.0"
+edition = "2021"
+rust-version = "1.78"
+license = "Apache-2.0"
+repository = "https://github.com/rustakka/atomr"
+homepage = "https://github.com/rustakka/atomr"
+authors = ["atomr contributors"]
+
+[workspace.dependencies]
+# Async / runtime
+tokio = { version = "1.39", features = ["full"] }
+tokio-util = { version = "0.7", features = ["codec"] }
+async-trait = "0.1"
+futures = "0.3"
+futures-util = "0.3"
+pin-project-lite = "0.2"
+
+# Logging / tracing
+tracing = "0.1"
+tracing-subscriber = { version = "0.3", features = ["env-filter", "fmt"] }
+
+# Concurrency primitives
+dashmap = "6"
+parking_lot = "0.12"
+crossbeam-channel = "0.5"
+crossbeam-queue = "0.3"
+flume = "0.11"
+
+# Serialization
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+bincode = { version = "2.0.0-rc.3", features = ["serde"] }
+bytes = "1"
+
+# Protobuf (remote)
+prost = "0.13"
+prost-build = "0.13"
+
+# Config
+config = { version = "0.14", default-features = false, features = ["toml", "json", "yaml"] }
+toml = "0.8"
+
+# Error / utility
+thiserror = "1"
+anyhow = "1"
+once_cell = "1"
+
+# Testing
+proptest = "1"
+criterion = { version = "0.5", default-features = false, features = ["html_reports"] }
+
+# Macros
+proc-macro2 = "1"
+quote = "1"
+syn = { version = "2", features = ["full", "extra-traits"] }
+
+# Python bindings
+pyo3 = { version = "0.22", default-features = false }
+pyo3-async-runtimes = { version = "0.22", default-features = false, features = ["tokio-runtime"] }
+
+# UUID / time
+uuid = { version = "1", features = ["v4", "serde"] }
+chrono = { version = "0.4", default-features = false, features = ["clock", "serde"] }
+
+# Persistence drivers (used by provider crates; kept here so versions stay in sync)
+sqlx = { version = "0.8", default-features = false, features = ["runtime-tokio-rustls", "macros", "any", "chrono", "uuid"] }
+fred = { version = "10", default-features = false, features = ["i-std", "transactions"] }
+mongodb = { version = "3" }
+scylla = { version = "1" }
+aws-config = { version = "1" }
+aws-sdk-dynamodb = { version = "1" }
+reqwest = { version = "0.12", default-features = false, features = ["rustls-tls", "json"] }
+hmac = "0.12"
+sha2 = "0.10"
+base64 = "0.22"
+url = "2"
+
+# HTTP / web (atomr-dashboard)
+axum = { version = "0.7", default-features = false, features = ["json", "ws", "http1", "tokio", "query", "matched-path"] }
+tower = { version = "0.5", default-features = false, features = ["util"] }
+tower-http = { version = "0.5", default-features = false, features = ["cors", "trace", "fs"] }
+rust-embed = { version = "8" }
+clap = { version = "4", features = ["derive"] }
+
+# Internal
+atomr = { path = "crates/atomr", version = "0.1.0" }
+atomr-core = { path = "crates/atomr-core", version = "0.1.0" }
+atomr-config = { path = "crates/atomr-config", version = "0.1.0" }
+atomr-macros = { path = "crates/atomr-macros", version = "0.1.0" }
+atomr-testkit = { path = "crates/atomr-testkit", version = "0.1.0" }
+atomr-remote = { path = "crates/atomr-remote", version = "0.1.0" }
+atomr-cluster = { path = "crates/atomr-cluster", version = "0.1.0" }
+atomr-cluster-tools = { path = "crates/atomr-cluster-tools", version = "0.1.0" }
+atomr-cluster-sharding = { path = "crates/atomr-cluster-sharding", version = "0.1.0" }
+atomr-cluster-metrics = { path = "crates/atomr-cluster-metrics", version = "0.1.0" }
+atomr-distributed-data = { path = "crates/atomr-distributed-data", version = "0.1.0" }
+atomr-persistence = { path = "crates/atomr-persistence", version = "0.1.0" }
+atomr-persistence-query = { path = "crates/atomr-persistence-query", version = "0.1.0" }
+atomr-persistence-query-inmemory = { path = "crates/atomr-persistence-query-inmemory", version = "0.1.0" }
+atomr-persistence-tck = { path = "crates/atomr-persistence-tck", version = "0.1.0" }
+atomr-persistence-sql = { path = "crates/atomr-persistence-sql", version = "0.1.0" }
+atomr-persistence-redis = { path = "crates/atomr-persistence-redis", version = "0.1.0" }
+atomr-persistence-mongodb = { path = "crates/atomr-persistence-mongodb", version = "0.1.0" }
+atomr-persistence-cassandra = { path = "crates/atomr-persistence-cassandra", version = "0.1.0" }
+atomr-persistence-aws = { path = "crates/atomr-persistence-aws", version = "0.1.0" }
+atomr-persistence-azure = { path = "crates/atomr-persistence-azure", version = "0.1.0" }
+atomr-streams = { path = "crates/atomr-streams", version = "0.1.0" }
+atomr-telemetry = { path = "crates/atomr-telemetry", version = "0.1.0" }
+atomr-dashboard = { path = "crates/atomr-dashboard", version = "0.1.0" }
+atomr-coordination = { path = "crates/atomr-coordination", version = "0.1.0" }
+atomr-discovery = { path = "crates/atomr-discovery", version = "0.1.0" }
+atomr-di = { path = "crates/atomr-di", version = "0.1.0" }
+atomr-hosting = { path = "crates/atomr-hosting", version = "0.1.0" }
+
+[profile.release]
+opt-level = 3
+lto = "thin"
+codegen-units = 1
+
+[profile.bench]
+opt-level = 3
+lto = "thin"
+
+# Phase 0 lint baseline. See docs/idiomatic-rust.md (P-4).
+#
+# `todo`/`unimplemented` are denied workspace-wide because the audit
+# found 0 sites; any regression breaks CI immediately.
+#
+# `unwrap_used`/`expect_used`/`panic` (and `print_stdout`/
+# `print_stderr`/`dbg_macro` in library crates) are NOT promoted to
+# `warn` here — the existing CI gate runs `cargo clippy ... -D
+# warnings` and the audit found pre-existing sites that would break
+# it. Instead, the per-crate counts are tracked by `cargo xtask
+# audit` against `docs/reports/audit-2026-04.json`; CI runs
+# `cargo xtask audit --check` and fails on regression. Phases that
+# touch a given crate ratchet that crate's lints up to `warn`/`deny`
+# once the corresponding sites are cleaned up (see Phase 13).
+[workspace.lints.clippy]
+todo                 = "deny"
+unimplemented        = "deny"

atomr_accel-0.1.0/atomr/crates/atomr-cluster/Cargo.toml

@@ -0,0 +1,36 @@
+[package]
+name = "atomr-cluster"
+description = "Membership, gossip, reachability, vector clocks, and split-brain resolvers for atomr clusters."
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+repository.workspace = true
+homepage.workspace = true
+readme = "README.md"
+documentation = "https://docs.rs/atomr-cluster"
+keywords = ["actor", "cluster", "distributed", "gossip"]
+categories = ["network-programming", "concurrency"]
+
+[dependencies]
+atomr-core = { workspace = true }
+atomr-remote = { workspace = true }
+atomr-config = { workspace = true }
+tokio = { workspace = true }
+serde = { workspace = true }
+serde_json = { workspace = true }
+bincode = { workspace = true }
+parking_lot = { workspace = true }
+thiserror = { workspace = true }
+tracing = { workspace = true }
+async-trait = { workspace = true }
+
+[dev-dependencies]
+atomr-testkit = { workspace = true }
+
+[lints]
+workspace = true
+
+[package.metadata.docs.rs]
+all-features = true
+rustdoc-args = ["--cfg", "docsrs"]