perf:remove isinstance check

[d-v-b]

removes an expensive isinstance check inside BytesCodec._decode_single. Isinstance on runtime_checkable protocols is expensive and this particular check is in a hotspot. Without the check, we are slightly less type-safe, but users who somehow get a non-ndarray into this part of the code will get an immediate a runtime error.

[codspeed-hq]

Merging this PR will improve performance by 34.44%

⚡ 31 improved benchmarks
✅ 17 untouched benchmarks
⏩ 6 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	test_slice_indexing[None-(slice(0, None, 4), slice(0, None, 4), slice(0, None, 4))-memory_get_latency]	555.2 ms	425.9 ms	+30.37%
⚡	WallTime	test_slice_indexing[None-(slice(10, -10, 4), slice(10, -10, 4), slice(10, -10, 4))-memory]	274.6 ms	205.1 ms	+33.88%
⚡	WallTime	test_slice_indexing[None-(slice(None, None, None), slice(0, 3, 2), slice(0, 10, None))-memory_get_latency]	5.2 ms	4.1 ms	+26.75%
⚡	WallTime	test_slice_indexing[None-(slice(10, -10, 4), slice(10, -10, 4), slice(10, -10, 4))-memory_get_latency]	304.4 ms	233.2 ms	+30.54%
⚡	WallTime	test_slice_indexing[None-(slice(None, 10, None), slice(None, 10, None), slice(None, 10, None))-memory]	1,007.5 µs	891.1 µs	+13.07%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(None, None, None), slice(None, None, None), slice(None, None, None))-memory_get_latency]	567.1 ms	439.6 ms	+28.99%
⚡	WallTime	test_read_array[memory-Layout(shape=(1000000,), chunks=(1000,), shards=None)-None]	372.3 ms	276.9 ms	+34.44%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(0, 0, 0)-memory]	2 ms	1.8 ms	+10.97%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(0, None, 4), slice(0, None, 4), slice(0, None, 4))-memory]	546 ms	416.2 ms	+31.2%
⚡	WallTime	test_read_array[memory-Layout(shape=(1000000,), chunks=(1000,), shards=(1000,))-gzip]	1.5 s	1.3 s	+14.58%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(0, None, 4), slice(0, None, 4), slice(0, None, 4))-memory_get_latency]	563.9 ms	433.4 ms	+30.12%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(None, None, None), slice(0, 3, 2), slice(0, 10, None))-memory]	7.6 ms	6.3 ms	+21.79%
⚡	WallTime	test_read_array[memory-Layout(shape=(1000000,), chunks=(1000,), shards=None)-gzip]	681.8 ms	584 ms	+16.75%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(None, None, None), slice(None, None, None), slice(None, None, None))-memory]	551.2 ms	421.9 ms	+30.64%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(None, None, None), slice(0, 3, 2), slice(0, 10, None))-memory_get_latency]	8.5 ms	7.2 ms	+18.34%
⚡	WallTime	test_read_array[memory-Layout(shape=(1000000,), chunks=(100,), shards=(1000000,))-None]	3.8 s	2.9 s	+32.39%
⚡	WallTime	test_slice_indexing[None-(slice(None, None, None), slice(0, 3, 2), slice(0, 10, None))-memory]	4.8 ms	3.7 ms	+28.66%
⚡	WallTime	test_slice_indexing[(50, 50, 50)-(slice(10, -10, 4), slice(10, -10, 4), slice(10, -10, 4))-memory_get_latency]	316.7 ms	244.6 ms	+29.5%
⚡	WallTime	test_read_array[memory-Layout(shape=(1000000,), chunks=(1000,), shards=(1000,))-None]	1,162.3 ms	975.1 ms	+19.2%
⚡	WallTime	test_slice_indexing[None-(slice(0, None, 4), slice(0, None, 4), slice(0, None, 4))-memory]	502.8 ms	374.1 ms	+34.39%
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

_{Comparing d-v-b:perf/remove-isinstance-check (f0d4b2e) with main (3e7d24d)}

Footnotes

1. 6 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[d-v-b]

35% perf improvement seems good

[dcherian]

can we just have as_array_like become as_ndarray_like?