Update benchmark results

Author: gerlero

README.md

@@ -21,7 +21,7 @@
 <div align="center">
 <img alt="benchmark" src="https://github.com/gerlero/foamlib/raw/main/benchmark/benchmark.png" height="250">
 
-Parsing a `volVectorField` with 200k cells.<sup>[1](#benchmark)</sup>
+_Loading a_ volVectorField _with one million cells_<sup>[1](#benchmark)</sup>
 </div>
 
 
@@ -282,4 +282,4 @@ If you use **foamlib** in your research, please cite our paper:
 
 ## 👟 Footnotes
 
-<a id="benchmark">[1]</a> foamlib 1.3.7 vs PyFoam 2023.7 on an M3 MacBook Air. [Benchmark script](https://github.com/gerlero/foamlib/blob/main/benchmark/benchmark.py).
+<a id="benchmark">[1]</a> foamlib 1.3.9 vs. PyFoam 2023.7 (Python 3.11.13) on an M3 MacBook Air. [Benchmark script](https://github.com/gerlero/foamlib/blob/main/benchmark/benchmark.py).

benchmark/benchmark.png

@@ -7,11 +7,17 @@
 from foamlib import FoamFieldFile
 from PyFoam.RunDictionary.ParsedParameterFile import ParsedParameterFile
 
-FoamFieldFile("U").internal_field = [[0.0, 0.0, 0.0]] * 200_000
+FoamFieldFile("U").internal_field = [[0.0, 0.0, 0.0]] * 1_000_000
+with FoamFieldFile("U_binary") as f:
+    f.format = "binary"
+    f.internal_field = [[0.0, 0.0, 0.0]] * 1_000_000
 
 print(
     f"foamlib: {min(timeit.repeat(lambda: FoamFieldFile('U').internal_field, number=1))} s"
 )
+print(
+    f"foamlib (binary): {min(timeit.repeat(lambda: FoamFieldFile('U_binary').internal_field, number=1))} s"
+)
 print(
     f"PyFoam: {min(timeit.repeat(lambda: ParsedParameterFile('U')['internalField'], number=1))} s"
 )

benchmark/benchmark.py

@@ -1,2 +1,2 @@
-foamlib==1.3.7
+foamlib==1.3.9
 PyFoam==2023.7