[Mono-devel-list] First CIL Regex performance result

[Ben Maurer]

Ok, lemme add some of my own:

> 	IL_000d:  ldc.i4 0
> 	IL_0012:  add 

x + 0 = x

Also, you have alot of code like:

> 	IL_0080:  ldarg.0 
> 	IL_0081:  ldfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::scan_ptr
> 	IL_0086:  ldarg.0 
> 	IL_0087:  ldfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::text_end
> 	IL_008c:  blt IL_0096

I wonder if you should store these into locals, and then store them back
when you are done. Local storage should be much faster.


> 	IL_0069:  ret
> 
> //****  Branch Code 
> 	IL_006a:  ldarg.0 
> 	IL_006b:  callvirt instance void class [System]'System.Text.RegularExpressions.CILMachineBase'::'Checkpoint'()
> 	IL_0070:  br IL_0080
> 
> 	IL_0075:  ldarg.0 
> 	IL_0076:  callvirt instance void class [System]'System.Text.RegularExpressions.CILMachineBase'::'Backtrack'()
> 	IL_007b:  br IL_00c9
> 
> //**** Eval Character 1
> 	IL_0080:  ldarg.0 

Ok, the arragnement here is really weird. If you move the block 6a-70 to
right above 80, you can avoid a branch.

For debugging purposes, some of your code is really complex and hard to
understand. I realize it is not designed to be read, but you are going
ot make it hard on yourself if you emit code like:

> 	IL_00a7:  ldarg.0 
> 	IL_00a8:  dup 
> 	IL_00a9:  ldfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::scan_ptr
> 	IL_00ae:  ldc.i4.1 
> 	IL_00af:  add 
> 	IL_00b0:  stfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::scan_ptr

It is much easier to read

> 	IL_00a7:  ldarg.0 
> 	IL_00a9:  ldfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::scan_ptr
> 	IL_00ae:  ldc.i4.1 
> 	IL_00af:  add 
> 		  ldarg.0
> 	IL_00b0:  stfld  int32 [System]System.Text.RegularExpressions.CILMachineBase::scan_ptr
> 


> IL_00da:  br IL_0128
> ...
> IL_0128:  br IL_005f

That can be simplified.


> 	IL_002a:  blt IL_0068
> 	IL_0063:  br IL_0028
> 
> 	IL_0068:  ldc.i4.0 
> 	IL_0069:  ret

The only branch to 68 is the one in 2a. So you can just move the code up there.

I think this shoudl get you on your way. Once you get these done, the
code should be alot cleaner and easier to optimize.

-- ben