llvm dev - Nov 2012 - [LLVMdev] Question about llvm.ctpop.*

Hi,

    Following is excerpted from 
http://llvm.org/releases/3.1/docs/LangRef.html#int_ctpop.
How come the return type needs to be consistent with parameter type?  
i64/i128 seems to be overkill,
and i8, i16 are inconvenient.

-----------------------------------
   declare i8 @llvm.ctpop.i8(i8  <src>)
   declare i16 @llvm.ctpop.i16(i16 <src>)
   declare i32 @llvm.ctpop.i32(i32 <src>)
   declare i64 @llvm.ctpop.i64(i64 <src>)
   declare i256 @llvm.ctpop.i256(i256 <src>)
   declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
-----------------------------------------------

Thanks
Shuxin

On Tue, Nov 13, 2012 at 4:53 PM, Shuxin Yang <shuxin.llvm at gmail.com>
wrote:
> Hi,
>
>    Following is excerpted from
> http://llvm.org/releases/3.1/docs/LangRef.html#int_ctpop.
> How come the return type needs to be consistent with parameter type?
> i64/i128 seems to be overkill,
> and i8, i16 are inconvenient.
>
> -----------------------------------
>   declare i8 @llvm.ctpop.i8(i8  <src>)
>   declare i16 @llvm.ctpop.i16(i16 <src>)
>   declare i32 @llvm.ctpop.i32(i32 <src>)
>   declare i64 @llvm.ctpop.i64(i64 <src>)
>   declare i256 @llvm.ctpop.i256(i256 <src>)
>   declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
It's basically just for the sake of simplicity... why should the
intrinsic be overloaded over two types if only one is necessary? If
there's a compelling use-case, it wouldn't be difficult to make the
return type independent of the operand type, but I'm not sure why it
matters.

-Eli

On Nov 13, 2012, at 4:53 PM, Shuxin Yang <shuxin.llvm at gmail.com> wrote:
> Hi,
> 
>   Following is excerpted from
http://llvm.org/releases/3.1/docs/LangRef.html#int_ctpop.
> How come the return type needs to be consistent with parameter type? 
i64/i128 seems to be overkill,
> and i8, i16 are inconvenient.
It isn't required at the machine code level, it is just to make clients of
the IR simpler, because ctpop (and other bithack intrinsics) are simpler that
way.  Also, picking any fixed result type won't work: you have to support
vectors.  Allowing it to be arbitrary just adds complexity.

-Chris

> 
> -----------------------------------
>  declare i8 @llvm.ctpop.i8(i8  <src>)
>  declare i16 @llvm.ctpop.i16(i16 <src>)
>  declare i32 @llvm.ctpop.i32(i32 <src>)
>  declare i64 @llvm.ctpop.i64(i64 <src>)
>  declare i256 @llvm.ctpop.i256(i256 <src>)
>  declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
> -----------------------------------------------
> 
> Thanks
> Shuxin
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev