llvm dev - Sep 2017 - Question about quad-register

Hi All,

  If the target supports quad-register R0:R1:R2:R3 (Rn is 32-bit
register), is it possible mapping quad-register
to v4i32 so that the following example work?

    typedef int v4si __attribute__ ((vector_size (16)));

    void foo(v4si i) {
      v4si j = i;
    }

I don't know how to write CallingConv.td to represent the concept of
occupying quad-register R0:R1:R2:R3
once seeing v4i32. Any example that I can refer to?

Thanks.

Regards,
chenwj

-- 
Wei-Ren Chen (陳韋任)
Homepage: https://people.cs.nctu.edu.tw/~chenwj

The standard trick is to define the "quad register" that would alias
r0/r1/r2/r3. See e.g. ARM backend.

On Sun, Sep 10, 2017 at 6:17 AM, 陳韋任 via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> Hi All,
>
>   If the target supports quad-register R0:R1:R2:R3 (Rn is 32-bit
> register), is it possible mapping quad-register
> to v4i32 so that the following example work?
>
>     typedef int v4si __attribute__ ((vector_size (16)));
>
>     void foo(v4si i) {
>       v4si j = i;
>     }
>
> I don't know how to write CallingConv.td to represent the concept of
> occupying quad-register R0:R1:R2:R3
> once seeing v4i32. Any example that I can refer to?
>
> Thanks.
>
> Regards,
> chenwj
>
> --
> Wei-Ren Chen (陳韋任)
> Homepage: https://people.cs.nctu.edu.tw/~chenwj
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev


-- 
With best regards, Anton Korobeynikov
Department of Statistical Modelling, Saint Petersburg State University

Hi Anton,

  I guess you mean something like this in RegisterInfo.td?

    def DPR : RegisterClass<"ARM", [f64, v8i8, v4i16, v2i32, v1i64,
v2f32, v4f16], 64,
                                             (sequence "D%u", 0,
31)> {


    def Tuples2DSpc : RegisterTuples<[dsub_0, dsub_2],
                                                            [(shl DPR, 0),
                                                             (shl DPR, 2)]>;

or this?

  let SubRegIndices = [dsub_0, dsub_1] in {
    def Q0  : ARMReg< 0,  "q0", [D0,   D1]>;
    def Q1  : ARMReg< 1,  "q1", [D2,   D3]>;
  }


Regards,
chenwj


2017-09-11 3:07 GMT+08:00 Anton Korobeynikov <anton at
korobeynikov.info>:
> The standard trick is to define the "quad register" that would
alias
> r0/r1/r2/r3. See e.g. ARM backend.
>
> On Sun, Sep 10, 2017 at 6:17 AM, 陳韋任 via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
>> Hi All,
>>
>>   If the target supports quad-register R0:R1:R2:R3 (Rn is 32-bit
>> register), is it possible mapping quad-register
>> to v4i32 so that the following example work?
>>
>>     typedef int v4si __attribute__ ((vector_size (16)));
>>
>>     void foo(v4si i) {
>>       v4si j = i;
>>     }
>>
>> I don't know how to write CallingConv.td to represent the concept
of
>> occupying quad-register R0:R1:R2:R3
>> once seeing v4i32. Any example that I can refer to?
>>
>> Thanks.
>>
>> Regards,
>> chenwj
>>
>> --
>> Wei-Ren Chen (陳韋任)
>> Homepage: https://people.cs.nctu.edu.tw/~chenwj
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
>
>
> --
> With best regards, Anton Korobeynikov
> Department of Statistical Modelling, Saint Petersburg State University


-- 
Wei-Ren Chen (陳韋任)
Homepage: https://people.cs.nctu.edu.tw/~chenwj