Lack of asm volatile qualifier (explicitly) again.

Guillaume Piolat first.name at gmail.com
Tue Jul 28 12:56:55 UTC 2020


On Tuesday, 28 July 2020 at 06:57:36 UTC, Cecil Ward wrote:
>
> What do others think? If others agree, how could a very small 
> DIP be set in motion ?
>

Hello,

LDC lets you do optimizable assembly with ldc.llvmasm.__asm
Better yet, you can also create IR directly with 
ldc.llvmasm.__ir_pure
This will yield results mroe portable and with optimal efficiency 
in a lot of cases.

GDC let's you do optimizable assembly if you can understand its 
arcane syntax!

But all this isn't very useful since doing assembly directly 
rarely lead to fastest results provided you use a modern backend.





More information about the Digitalmars-d-learn mailing list