Jimmy

Results 527 comments of Jimmy

this can be disabled by default tough, only guys who need it and konws how to use it can able it.

> ```cs > [MethodImpl(MethodImplOptions.AggressiveInlining)] > private static void ArrowInline(int a, int b) => CallMethodThatReturnsInt(a, b); > ``` > > The compiler should drop the returned value in this case. Fixed.

Let's leave these non-code related tasks to NGD, @superboyiii

The reason that i am not benchmark on single Opcode is becuause 1. DOS attack needs to be in a loop; 2. single OpCode execution time is related to its...

> Why not NOP? I've told about this many times (the last iteration: [#3552 (comment)](https://github.com/neo-project/neo/issues/3552#issuecomment-2436165391)). Nop is basically the same, its 500 ms. > > not benchmark on single Opcode...

> That's exactly the point. Take `UNPACK`, it's 2048 now, make it `4 * ((num_of_elements + 4) >> 2)`, get a reasonable 4 for 0-3 elements and 8 for 3-7...

@roman-khimov this is the benchmark with `4 * ((num_of_elements + 4) >> 2)`, problem without a threshold here is that when the subitem count low, the price is high compared...

This is PACK without threshold and factor: | Method | ItemCount | Mean | Error | StdDev | Median | |------------- |---------- |----------:|----------:|----------:|----------:| | Bench_OneGAS | 4 | 98.87 ms...

> > when the subitem count low, the price is high compared to its execution time > > I see your formula is a bit more accurate. But likely you're...