Re: [PATCH V3 2/4] RISC-V: Add vector related pipelines

[Edwin Lu <ewlu@rivosinc.com>]

On 1/25/2024 9:06 AM, Robin Dapp wrote:
> Thanks, that looks better IMHO.
>
>> +;; Copyright (C) 2011-2024 Free Software Foundation, Inc.
>> +;; Contributed by Andrew Waterman (andrew@sifive.com).
>> +;; Based on MIPS target for GNU compiler.
> You might want to change that, as well as the date.  While at
> it you can also fix the broken date in my original file ;)


Completely forgot about this. I'll update it :)

>> +(define_insn_reservation "vec_load" 6
>> +  (and (eq_attr "is_inorder" "no")
>> +       (eq_attr "type" "vlde,vldm,vlds,vldux,vldox,vldff,vldr"))
>> +  "vxu_ooo_issue,vxu_ooo_alu")
> I would rather ditch the is_inorder attribute for now and define
> "low" latencies as well as reservations explicitly once we're
> sure rather than falling back to scheduler defaults.


I think removing the is_inorder attribute should be ok. I added it 
because I wanted to avoid having two matching insn reservations defined 
since matching solely on the type attribute should also match on all 
subsets as well (i.e. if eventually we add an insn reservation checking 
for type "vlde" and tune "generic-ooo", any "vlde" insn would map to 
both reservations)

For now I should just remove the is_inorder attribute. We will update 
the latencies and add new reservations after we know what they should 
be. Is that correct?

> OK with those changes.
>
> Regards
>   Robin