Go Arrays vs. Slices: Why is Local Slice Access Faster Than Local Array Access?

Array vs. Slice: Accessing Speed

In Go, a slice and an array are two closely related data structures used to store elements of a specific data type. While they share similarities, one key difference lies in their access speed, particularly when comparing global and local instances.

Benchmark Results:

To assess the performance difference, a benchmark was conducted using the following functions:

• BenchmarkSliceGlobal: Accesses elements of a global slice.
• BenchmarkArrayGlobal: Accesses elements of a global array.
• BenchmarkSliceLocal: Accesses elements of a local slice.
• BenchmarkArrayLocal: Accesses elements of a local array.

The results consistently exhibited a faster accessing speed for global arrays than for global slices. However, the local slice significantly outperformed the local array.

Explanation:

To determine the reason for this discrepancy, the amd64 assembly of the local array and slice benchmark functions was examined.

• Local Array: Loads the address of the array from memory multiple times, introducing overhead on each access.
• Local Slice: Avoids repetitive memory loading by computing exclusively on registers.

This suggests that the local slice benefits from efficient register utilization, while the local array incurs additional overhead by constantly loading the array's address.

Additionally, the array version invokes the runtime.duffcopy function, a lengthy assembly routine, while the slice version does not. This further contributes to the performance disparity.

The above is the detailed content of Go Arrays vs. Slices: Why is Local Slice Access Faster Than Local Array Access?. For more information, please follow other related articles on the PHP Chinese website!