tensorstudio 1.0.0__tar.gz → 1.0.1__tar.gz

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/.github/workflows/ci.yml

@@ -30,6 +30,10 @@ jobs:
         run: mypy --python-version ${{ matrix.python-version }} python/tensorstudio
       - name: Tests
         run: pytest -q
+      - name: Runner scripts
+        run: |
+          python test_all.py --help
+          python benchmark_all.py --help
       - name: Example
         run: python examples/basic_tensor_ops.py
 
@@ -59,6 +63,10 @@ jobs:
         run: mypy --python-version ${{ matrix.python-version }} python/tensorstudio
       - name: Tests
         run: pytest -q
+      - name: Runner scripts
+        run: |
+          python test_all.py --help
+          python benchmark_all.py --help
       - name: Example
         run: python examples/basic_tensor_ops.py
 
@@ -88,5 +96,9 @@ jobs:
         run: mypy --python-version ${{ matrix.python-version }} python/tensorstudio
       - name: Tests
         run: pytest -q
+      - name: Runner scripts
+        run: |
+          python test_all.py --help
+          python benchmark_all.py --help
       - name: Example
         run: python examples/basic_tensor_ops.py

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/.github/workflows/publish-testpypi.yml

@@ -106,3 +106,4 @@ jobs:
         uses: pypa/gh-action-pypi-publish@release/v1
         with:
           repository-url: https://test.pypi.org/legacy/
+          skip-existing: true

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/.github/workflows/publish.yml

@@ -6,7 +6,7 @@ on:
     types: [published]
   push:
     tags:
-      - "v1.0.0"
+      - "v*"
 
 permissions:
   contents: read
@@ -103,5 +103,11 @@ jobs:
           pattern: dist-*
           path: dist
           merge-multiple: true
+      - name: Verify artifacts
+        run: |
+          python -m pip install -U twine
+          python -m twine check dist/*
       - name: Publish to PyPI
         uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          skip-existing: true

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## 1.0.1
+
+- Added `test_all.py` for one-command local release checks covering lint,
+  typing, tests, examples, docs, build, and package metadata validation.
+- Added `benchmark_all.py` for one-command full benchmark report generation.
+- Added explicit benchmark win columns for NumPy, TensorFlow, PyTorch, and JAX,
+  plus fastest-library reporting for each benchmark case.
+- Hardened release publishing triggers so future `v*` tags are eligible for
+  the publish workflow instead of only `v1.0.0`.
+- Updated docs to describe benchmark interpretation, local release checks, and
+  site navigation for benchmark, hardware, roadmap, and usage pages.
+
 ## 1.0.0
 
 - Promoted the CPU-only stable API foundation from release candidate status.

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.18)
 
-project(tensorstudio VERSION 1.0.0 LANGUAGES CXX)
+project(tensorstudio VERSION 1.0.1 LANGUAGES CXX)
 
 find_package(Python COMPONENTS Interpreter Development.Module REQUIRED)
 set(PYBIND11_FINDPYTHON ON)

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: tensorstudio
-Version: 1.0.0
+Version: 1.0.1
 Summary: TensorStudio is a compact C++ tensor and autograd engine with a Python API for learning, experimentation, and lightweight ML workloads.
 Keywords: tensor,autograd,machine-learning,cpp,pybind11
 Author: TensorStudio contributors
@@ -65,7 +65,7 @@ Description-Content-Type: text/markdown
 TensorStudio is a compact C++ tensor and autograd engine with a Python API for
 learning, experimentation, and lightweight ML workloads.
 
-TensorStudio `1.0.0` is a CPU-only stable API foundation. It is eager-only,
+TensorStudio `1.0.1` is a CPU-only stable API foundation. It is eager-only,
 intentionally small, and not a replacement for mature ML frameworks.
 
 ## Install
@@ -227,13 +227,18 @@ performance is still experimental. Benchmarks live in `benchmarks/` and can be
 run locally:
 
 ```bash
+python benchmark_all.py
 python benchmarks/benchmark_report.py
 ```
 
-On one Windows CPython 3.10 run for `1.0.0`, TensorStudio beat NumPy on 13
+`benchmark_all.py` writes `benchmarks/results.md` and includes explicit win
+columns for NumPy, TensorFlow, PyTorch, and JAX when those libraries are
+available locally.
+
+On one Windows CPython 3.10 run for `1.0.1`, TensorStudio beat NumPy on 13
 small activation/reduction benchmark cases and lost on 76 NumPy-comparable
-cases. Against PyTorch CPU `2.12.1+cpu`, TensorStudio won 68 local cases and
-lost 26. The strongest local wins were small eager operations where framework
+cases. Against PyTorch CPU `2.12.1+cpu`, TensorStudio won 71 local cases and
+lost 23. The strongest local wins were small eager operations where framework
 dispatch overhead dominates; larger matrix multiplication, larger transcendental
 activations, and larger autograd workloads remain faster in PyTorch and NumPy.
 See `benchmarks/results.md` for the full table, platform details, and exact
@@ -243,11 +248,11 @@ Snapshot from that local run:
 
 | operation | shape | TensorStudio | NumPy | PyTorch CPU | TS vs NumPy | TS vs PyTorch |
 |---|---:|---:|---:|---:|---:|---:|
-| `sigmoid` | `(32,)` | 0.0018 ms | 0.0049 ms | 0.0638 ms | 2.7548x | 35.6408x |
-| `mean` | `(32,)` | 0.0016 ms | 0.0071 ms | 0.0110 ms | 4.3608x | 6.7855x |
-| `chain_relu` | `(128,)` | 0.0089 ms | 0.0039 ms | 0.0545 ms | 0.4410x | 6.1384x |
-| `matmul` | `(256, 256)` | 3.0180 ms | 0.4115 ms | 0.1392 ms | 0.1363x | 0.0461x |
-| `elementwise_backward` | `(1024,)` | 2.3844 ms | n/a | 0.1830 ms | n/a | 0.0767x |
+| `sigmoid` | `(32,)` | 0.0018 ms | 0.0036 ms | 0.0575 ms | 2.0414x | 32.2110x |
+| `mean` | `(32,)` | 0.0016 ms | 0.0069 ms | 0.0112 ms | 4.4051x | 7.1262x |
+| `chain_relu` | `(128,)` | 0.0086 ms | 0.0039 ms | 0.0567 ms | 0.4488x | 6.5977x |
+| `matmul` | `(256, 256)` | 2.4215 ms | 0.4027 ms | 0.0832 ms | 0.1663x | 0.0343x |
+| `elementwise_backward` | `(1024,)` | 2.3527 ms | n/a | 0.1887 ms | n/a | 0.0802x |
 
 Speedup is `competitor median / TensorStudio median`, so values above `1.0x`
 favor TensorStudio.
@@ -273,6 +278,7 @@ unsafe because pickle can execute arbitrary code.
 
 ```bash
 python -m pip install -e ".[dev,docs]"
+python test_all.py --skip-build
 ruff check .
 mypy python/tensorstudio
 pytest -q
@@ -285,6 +291,7 @@ scikit-build-core, and C++20.
 
 ## Release Checklist
 
+- `python test_all.py` passes locally.
 - `ruff check .` passes.
 - `mypy python/tensorstudio` passes.
 - `pytest -q` passes on Windows, Linux, and macOS.
@@ -309,6 +316,7 @@ tokens or print secrets.
 - CPU backend only.
 - Eager execution only.
 - No CUDA or Metal backend yet.
+- No BLAS-backed matrix multiplication yet.
 - No graph compiler or distributed runtime.
 - No convolution layers yet.
 - No sparse tensors or advanced indexing.

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/README.md

@@ -7,7 +7,7 @@
 TensorStudio is a compact C++ tensor and autograd engine with a Python API for
 learning, experimentation, and lightweight ML workloads.
 
-TensorStudio `1.0.0` is a CPU-only stable API foundation. It is eager-only,
+TensorStudio `1.0.1` is a CPU-only stable API foundation. It is eager-only,
 intentionally small, and not a replacement for mature ML frameworks.
 
 ## Install
@@ -169,13 +169,18 @@ performance is still experimental. Benchmarks live in `benchmarks/` and can be
 run locally:
 
 ```bash
+python benchmark_all.py
 python benchmarks/benchmark_report.py
 ```
 
-On one Windows CPython 3.10 run for `1.0.0`, TensorStudio beat NumPy on 13
+`benchmark_all.py` writes `benchmarks/results.md` and includes explicit win
+columns for NumPy, TensorFlow, PyTorch, and JAX when those libraries are
+available locally.
+
+On one Windows CPython 3.10 run for `1.0.1`, TensorStudio beat NumPy on 13
 small activation/reduction benchmark cases and lost on 76 NumPy-comparable
-cases. Against PyTorch CPU `2.12.1+cpu`, TensorStudio won 68 local cases and
-lost 26. The strongest local wins were small eager operations where framework
+cases. Against PyTorch CPU `2.12.1+cpu`, TensorStudio won 71 local cases and
+lost 23. The strongest local wins were small eager operations where framework
 dispatch overhead dominates; larger matrix multiplication, larger transcendental
 activations, and larger autograd workloads remain faster in PyTorch and NumPy.
 See `benchmarks/results.md` for the full table, platform details, and exact
@@ -185,11 +190,11 @@ Snapshot from that local run:
 
 | operation | shape | TensorStudio | NumPy | PyTorch CPU | TS vs NumPy | TS vs PyTorch |
 |---|---:|---:|---:|---:|---:|---:|
-| `sigmoid` | `(32,)` | 0.0018 ms | 0.0049 ms | 0.0638 ms | 2.7548x | 35.6408x |
-| `mean` | `(32,)` | 0.0016 ms | 0.0071 ms | 0.0110 ms | 4.3608x | 6.7855x |
-| `chain_relu` | `(128,)` | 0.0089 ms | 0.0039 ms | 0.0545 ms | 0.4410x | 6.1384x |
-| `matmul` | `(256, 256)` | 3.0180 ms | 0.4115 ms | 0.1392 ms | 0.1363x | 0.0461x |
-| `elementwise_backward` | `(1024,)` | 2.3844 ms | n/a | 0.1830 ms | n/a | 0.0767x |
+| `sigmoid` | `(32,)` | 0.0018 ms | 0.0036 ms | 0.0575 ms | 2.0414x | 32.2110x |
+| `mean` | `(32,)` | 0.0016 ms | 0.0069 ms | 0.0112 ms | 4.4051x | 7.1262x |
+| `chain_relu` | `(128,)` | 0.0086 ms | 0.0039 ms | 0.0567 ms | 0.4488x | 6.5977x |
+| `matmul` | `(256, 256)` | 2.4215 ms | 0.4027 ms | 0.0832 ms | 0.1663x | 0.0343x |
+| `elementwise_backward` | `(1024,)` | 2.3527 ms | n/a | 0.1887 ms | n/a | 0.0802x |
 
 Speedup is `competitor median / TensorStudio median`, so values above `1.0x`
 favor TensorStudio.
@@ -215,6 +220,7 @@ unsafe because pickle can execute arbitrary code.
 
 ```bash
 python -m pip install -e ".[dev,docs]"
+python test_all.py --skip-build
 ruff check .
 mypy python/tensorstudio
 pytest -q
@@ -227,6 +233,7 @@ scikit-build-core, and C++20.
 
 ## Release Checklist
 
+- `python test_all.py` passes locally.
 - `ruff check .` passes.
 - `mypy python/tensorstudio` passes.
 - `pytest -q` passes on Windows, Linux, and macOS.
@@ -251,6 +258,7 @@ tokens or print secrets.
 - CPU backend only.
 - Eager execution only.
 - No CUDA or Metal backend yet.
+- No BLAS-backed matrix multiplication yet.
 - No graph compiler or distributed runtime.
 - No convolution layers yet.
 - No sparse tensors or advanced indexing.

tensorstudio-1.0.1/benchmark_all.py

@@ -0,0 +1,48 @@
+from __future__ import annotations
+
+import argparse
+from pathlib import Path
+
+from benchmarks.benchmark_report import run_benchmarks
+
+ALL_SECTIONS = {
+    "elementwise",
+    "matmul",
+    "reductions",
+    "activations",
+    "autograd",
+    "training_loop",
+}
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Run every TensorStudio benchmark section.")
+    parser.add_argument(
+        "--section",
+        action="append",
+        choices=sorted(ALL_SECTIONS),
+        help="Run only the selected section. Repeat for multiple sections.",
+    )
+    parser.add_argument(
+        "--output",
+        type=Path,
+        default=Path("benchmarks") / "results.md",
+        help="Markdown report path. Defaults to benchmarks/results.md.",
+    )
+    parser.add_argument(
+        "--print",
+        action="store_true",
+        help="Print the full Markdown report after writing it.",
+    )
+    args = parser.parse_args()
+
+    selected_sections = set(args.section or ALL_SECTIONS)
+    report = run_benchmarks(selected_sections, args.output)
+    if args.print:
+        print(report)
+    else:
+        print(f"Wrote benchmark report to {args.output}")
+
+
+if __name__ == "__main__":
+    main()

tensorstudio-1.0.1/benchmarks/__init__.py

@@ -0,0 +1,2 @@
+"""Local TensorStudio benchmark helpers."""
+

{tensorstudio-1.0.0 → tensorstudio-1.0.1}/benchmarks/benchmark_report.py

@@ -539,12 +539,24 @@ def _format_float(value: float | None) -> str:
     return f"{value:.4f}"
 
 
+def _format_win(value: float | None) -> str:
+    if value is None:
+        return "n/a"
+    return "yes" if value > 1.0 else "no"
+
+
 def _speedup(ts_ms: float | None, competitor_ms: float | None) -> float | None:
     if ts_ms is None or competitor_ms is None or ts_ms <= 0:
         return None
     return competitor_ms / ts_ms
 
 
+def _fastest_library(medians: dict[str, float]) -> str:
+    if not medians:
+        return "n/a"
+    return min(medians.items(), key=lambda item: item[1])[0]
+
+
 def _run_cases(cases: Iterable[BenchmarkCase], libraries: list[Library]) -> dict[str, Stats]:
     results: dict[str, Stats] = {}
     for case in cases:
@@ -640,10 +652,13 @@ def _render_report(
     lines.append("")
     lines.append(
         "| category | operation | shape | library | median ms | mean ms | min ms | "
-        "max ms | std ms | TS vs NumPy | TS vs TensorFlow | TS vs PyTorch | TS vs JAX | result |"
+        "max ms | std ms | TS vs NumPy | TS vs TensorFlow | TS vs PyTorch | TS vs JAX | "
+        "win vs NumPy | win vs TensorFlow | win vs PyTorch | win vs JAX | "
+        "fastest library | result |"
     )
     lines.append(
-        "|---|---|---:|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---|"
+        "|---|---|---:|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|"
+        "---|---|---|---|---|---|"
     )
 
     for case in cases:
@@ -668,6 +683,7 @@ def _render_report(
                 medians.get("JAX CPU dispatch"),
             ),
         }
+        fastest = _fastest_library(medians)
         for library in libraries:
             stats = results.get(_result_key(case, library.name))
             if stats is None:
@@ -692,13 +708,19 @@ def _render_report(
                 f"{_format_float(speedups['NumPy'])} | "
                 f"{_format_float(speedups['TensorFlow CPU eager'])} | "
                 f"{_format_float(speedups['PyTorch CPU'])} | "
-                f"{_format_float(speedups['JAX CPU dispatch'])} | {result} |"
+                f"{_format_float(speedups['JAX CPU dispatch'])} | "
+                f"{_format_win(speedups['NumPy'])} | "
+                f"{_format_win(speedups['TensorFlow CPU eager'])} | "
+                f"{_format_win(speedups['PyTorch CPU'])} | "
+                f"{_format_win(speedups['JAX CPU dispatch'])} | "
+                f"{fastest} | {result} |"
             )
     lines.append("")
     lines.append(
         "Speedup columns are `competitor median / TensorStudio median`; values above 1.0 mean"
     )
     lines.append("TensorStudio was faster for that specific case.")
+    lines.append("Win columns say whether that same speedup is above 1.0 for the competitor.")
     lines.append("")
     return "\n".join(lines)