Stackgo takes advantage of a simple page-based implementation that pre-allocates memory in advance using a default page size of 4096 bytes.
Although there is a rare pre-allocation overhead, on average the push time complexity is O(1).
Memory-shrinking is supported when completely freeing a page upon pop. This means that there may be an additional overhead when popping, but on average pop time complexity is still O(1).
Here the results of benchmarking stackgo
| Implementation | Processor | Operation | ns/op | B/op | allocs/op |
|---|---|---|---|---|---|
| stackgo | 2.3 GHz Intel Core i5 | PUSH | 133 | 24 | 1 |
| classic | 2.3 GHz Intel Core i5 | PUSH | 255 | 40 | 2 |
| stackgo | 2.3 GHz Intel Core i5 | POP | 15.6 | 0 | 0 |
| classic | 2.3 GHz Intel Core i5 | POP | 12.6 | 0 | 0 |
As you can see, stackgo outperforms on push operation, giving roughly the same performances on pop.