Zarr chunk access
will-moore opened this issue · 0 comments
will-moore commented
Looking to see whether loading a different chunk would affect the zarr performance, I first tried commenting-out the accessing and slicing of a zarr chunk,
diff --git a/benchmark/benchmark.py b/benchmark/benchmark.py
index 7336442..e38c69e 100644
--- a/benchmark/benchmark.py
+++ b/benchmark/benchmark.py
@@ -77,8 +77,8 @@ def test_zarr_chunk(benchmark, method):
def run(self):
data = self.group["0"]
- chunks = data.chunks
- len(data[0:chunks[0], 0:chunks[1], 0:chunks[2], 0:chunks[3], 0:chunks[4]])
+ # chunks = data.chunks
+ # len(data[0:chunks[0], 0:chunks[1], 0:chunks[2], 0:chunks[3], 0:chunks[4]])
And found that this had no significant effect on the benchmark output:
$ docker-compose run --rm benchmark -sv
...
before:
after:
But commenting out the previous line data = self.group["0"]
had the expected effect of reducing all zarr tests to 10e-5
secs.