1. posterid:1677298
  2. RISC-V Vector Task Group: fractional LMUL

Re: RISC-V Vector Task Group: fractional LMUL

David Horner
 

I  left the encoding unspecified in the proposal.

That was intentional as I saw various tradeoffs.

However, I now recommend the codes be in order of increasing VLMAX value as so:


vlmul

mnemonic

LMUL

VLMAX

#groups

Grouped registers

0

0

0

       lf16

1/16

VLEN/SEW/16

       32

vn (single register, lower 1/16th)

0

0

1

       lf8

1/8

VLEN/SEW/8

       32

vn (single register, lower 1/8th)

0

1

0

       lf4

1/4

VLEN/SEW/4

       32

vn (single register, lower 1/4th)

0

1

1

       lf2

1/2

VLEN/SEW/2

       32

vn (single register, lower 1/2th)

1

0

0

       l1

1

VLEN/SEW

       32

vn (single register in group)

1

0

1

       l2

2

2*VLEN/SEW

       16

vn, vn+1

1

1

0

       l4

4

4*VLEN/SEW

       8

vn, …​, vn+3

1

1

1

       l8

8

8*VLEN/SEW

       4

vn, …​, vn+7


This allows for a quick "andi and bz" check of the "special" LMUL 1/16

It also allows the formula
        new LMUL = ( new sew / old SEW )  * old LMUL
to be implemented as
      vlmul = new vsew - old vsew + old vlmul

The first column could be inverted to allow the original code values with no appreciable hardware cost, but it complicates software for the two above cases.



On 2020-02-09 12:44 a.m., David Horner via Lists.Riscv.Org wrote:

Special lmul code in vsetvl{i} to derive LML from existing SEW/LMUL and provided sew code.

As mentioned in the TG we suggested widen LMUL to 3 bits with 7 (explicit) states.

I suggest we name them l1,l2,l4,l8 as before and lf2, lf4, lf8 the fractional 1/2, 1/4, 1/8 respectively.

lf16 may be useful in larger machines, but already lf8 cannot be used on minimal RV32 machines.

( I recommend we use lf2, lf4 and lf8 rather than f2, f4, f8 to avoid confusion with float and to associate with lmul as a fraction value)

I propose the remaining value be used to derive LMUL from the existing SEW/LMUL and provided sew code.

Specifically, new LMUL = ( new sew / old SEW )  * old LMUL which retains the initial SEW/LMUL ratio.
Join tech-vector-ext@lists.riscv.org to automatically receive all group messages.