On Mon, 23 Mar 2020 at 22:12, Bill Huffman <huffman@...> wrote:

On 3/23/20 12:37 PM, Krste Asanovic wrote:
> EXTERNAL MAIL
>
>
>
> Date: 2020/3/20
> Task Group: Vector Extension
> Chair: Krste Asanovic
> Number of Attendees: ~20
> Current issues on github: https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_riscv_riscv-2Dv-2Dspec&d=DwIBAg&c=aUq983L2pue2FqKFoP6PGHMJQyoJ7kl3s3GZ-_haXqY&r=AYJ4kbebphYpRw2lYDUDCk5w5Qa3-DR3bQnFjLVmM80&m=3-SR1O6HM8Iu6jjTQmNtVVm49J8CuL1gmyiHKoiiA6Q&s=zZlo9CtETrWKKf8ZSBOIloLoVQ0w6upesOs7M7WK44k&e=
>
...
>
> #393 Fractional LMUL additional registers
>
> Proposal to add fractional LMUL to support a greater number of usable
> architecture registers in presence of mixed-width operands.
>
> There was discussion about the new proposal on mapping fractional LMUL
> elements when systems have SLEN<VLEN, as it appeared to break SLEN
> wiring optimization.  A scheme that appears to address the datapath
> wiring concern spreads out fractional LMUL elements along the register
> was discussed in meeting
>
>      VLEN=128b, SLEN=64b
>      F E D C B A 9 8 7 6 5 4 3 2 1 0  Byte number
>
>      X X X X X X X X X X X X 0 0 0 0  LMUL=1/4, SEW=32b
>      X X X X X X X X 0 0 0 0 0 0 0 0  LMUL=1/2, SEW=64b
>
>      X X X X 1 1 1 1 X X X X 0 0 0 0  LMUL=1/2, SEW=32b
>      1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0  LMUL=1, SEW=64b
>
> Discussion to continue on mailing list.

I assume that:

ELEN=64b here.  It can't be smaller because SEW=64b is listed.  It can't
be larger because SLEN=64b

LMUL=1/8, SEW=32b   is illegal
LMUL=1/4, SEW=64b   is illegal

LMUL=1/8, SEW=64b   is illegal

The compiler knows these are illegal because it knows ELEN.

      Bill

Note that ELEN > SLEN has long been allowed in the spec (although it's expected that such implementations would be rare and unusual). But yes, if ELEN=64 then the natural constraint we independently proposed (LMUL >= SEW/ELEN in your words) would mean those combinations of LMUL and SEW are illegal.

Conversely, if ELEN=128 (so that LMUL=1/4, SEW=32 would be legal legal), then we're looking at an implementation with ELEN > SLEN. Such implementations are somewhat odd and the usual intuition about SLEN's effects on the datapath doesn't apply to them even at LMUL = 1.

Best,

Hanna

>
>
>
>