Merge pull request #124 from nschloe/prevent-div-by-0

Prevent div by 0

.github/ISSUE_TEMPLATE/bug_report.md

@@ -14,6 +14,7 @@ A minimal code example that reproduces the problem would be a big help if you ca
 
 **Diagnose**
 I may ask you to cut and paste the output of the following command.
+
 ```
 pip freeze | grep perfplot
 ```

.github/ISSUE_TEMPLATE/feature_request.md

@@ -3,7 +3,7 @@ name: Feature request
 about: Suggest an idea for this project
 title: "[REQUEST]"
 labels: Needs triage
-assignees: ''
+assignees: ""
 ---
 
 Consider posting in https://github.com/nschloe/perfplot/discussions for feedback before raising a feature request.

.github/workflows/ci.yml

@@ -3,10 +3,10 @@ name: ci
 on:
   push:
     branches:
-    - main
+      - main
   pull_request:
     branches:
-    - main
+      - main
 
 jobs:
   lint:
@@ -25,13 +25,13 @@ jobs:
       matrix:
         python-version: [3.7, 3.8, 3.9]
     steps:
-    - uses: actions/setup-python@v2
-      with:
-        python-version: ${{ matrix.python-version }}
-    - uses: actions/checkout@v2
-    - name: Test with tox
-      run: |
-        pip install tox
-        tox -- --cov perfplot --cov-report xml --cov-report term
-    - uses: codecov/codecov-action@v1
-      if: ${{ matrix.python-version == '3.9' }}
+      - uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+      - uses: actions/checkout@v2
+      - name: Test with tox
+        run: |
+          pip install tox
+          tox -- --cov perfplot --cov-report xml --cov-report term
+      - uses: codecov/codecov-action@v1
+        if: ${{ matrix.python-version == '3.9' }}

README.md

@@ -19,6 +19,7 @@ testing snippets with input parameters (e.g., the size of an array) and plotting
 results.
 
 For example, to compare different NumPy array concatenation methods, the script
+
 ```python
 import numpy as np
 import perfplot
@@ -47,29 +48,35 @@ perfplot.show(
     # flops=lambda n: 3*n,  # FLOPS plots
 )
 ```
+
 produces
 
-![](https://nschloe.github.io/perfplot/concat.svg) | ![](https://nschloe.github.io/perfplot/relative.svg)
-| --- | --- |
+| ![](https://nschloe.github.io/perfplot/concat.svg) | ![](https://nschloe.github.io/perfplot/relative.svg) |
+| -------------------------------------------------- | ---------------------------------------------------- |
 
 Clearly, `stack` and `vstack` are the best options for large arrays.
 
 (By default, perfplot asserts the equality of the output of all snippets, too.)
 
 If your plot takes a while to generate, you can also use
+
 <!--pytest-codeblocks:skip-->
+
 ```python
 perfplot.live(
     # ...
 )
 ```
+
 <img alt="live" src="https://nschloe.github.io/perfplot/live.gif" width="40%">
 
 with the same arguments as above. It will plot the updates live.
 
 Benchmarking and plotting can be separated. This allows multiple plots of the same data,
 for example:
+
 <!--pytest-codeblocks:skip-->
+
 ```python
 out = perfplot.bench(
     # same arguments as above (except the plot-related ones, like time_unit or log*)
@@ -80,28 +87,32 @@ out.save("perf.png", transparent=True, bbox_inches="tight")
 
 Other examples:
 
-  * [Making a flat list out of list of lists in Python](https://stackoverflow.com/a/45323085/353337)
-  * [Most efficient way to map function over numpy array](https://stackoverflow.com/a/46470401/353337)
-  * [numpy: most efficient frequency counts for unique values in an array](https://stackoverflow.com/a/43096495/353337)
-  * [Most efficient way to reverse a numpy array](https://stackoverflow.com/a/44921013/353337)
-  * [How to add an extra column to an numpy array](https://stackoverflow.com/a/40218298/353337)
-  * [Initializing numpy matrix to something other than zero or one](https://stackoverflow.com/a/45006691/353337)
+- [Making a flat list out of list of lists in Python](https://stackoverflow.com/a/45323085/353337)
+- [Most efficient way to map function over numpy array](https://stackoverflow.com/a/46470401/353337)
+- [numpy: most efficient frequency counts for unique values in an array](https://stackoverflow.com/a/43096495/353337)
+- [Most efficient way to reverse a numpy array](https://stackoverflow.com/a/44921013/353337)
+- [How to add an extra column to an numpy array](https://stackoverflow.com/a/40218298/353337)
+- [Initializing numpy matrix to something other than zero or one](https://stackoverflow.com/a/45006691/353337)
 
 ### Installation
 
 perfplot is [available from the Python Package
 Index](https://pypi.org/project/perfplot/), so simply do
+
 ```
 pip install perfplot
 ```
+
 to install.
 
 ### Testing
 
 To run the perfplot unit tests, check out this repository and type
+
 ```
 tox
 ```
 
 ### License
+
 This software is published under the [GPLv3 license](https://www.gnu.org/licenses/gpl-3.0.en.html).

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = perfplot
-version = 0.9.9
+version = 0.9.10
 author = Nico Schlömer
 author_email = [email protected]
 description = Performance plots for Python code snippets

src/perfplot/_main.py

@@ -229,13 +229,17 @@ def __next__(self):
             # and the time for gauging how many more repetitions are to be done. If the
             # initial time doesn't exceed the target time, append as many repetitions as
             # the first measurement suggests. If the kernel is fast, the measurement
-            # with one repetition only can be somewhat off, but most of the time it's
-            # good enough.
+            # with one repetition only can be somewhat off because the CPU needs to spin
+            # up first. The actual times are only reached after a few hundred
+            # nanoseconds of computation. Most of the time it's okay though.
             t0_ns = time.time_ns()
             val = kernel(data)
             t1_ns = time.time_ns()
             t_ns = t1_ns - t0_ns
 
+            if t_ns == 0:
+                raise RuntimeError("Measured 0 ns for a function call. Try again?")
+
             if self.equality_check:
                 if k == 0:
                     reference = val