...@rivosinc.com
Subject: Re: [PATCH V3] RISC-V: Optimize codegen of VLA SLP
On 6/20/23 03:01, Robin Dapp wrote:
> LGTM.
Likewise -- that V2/V3 is a nice improvement over the original V1 approach.
jeff
On 6/20/23 03:01, Robin Dapp wrote:
LGTM.
Likewise -- that V2/V3 is a nice improvement over the original V1 approach.
jeff
LGTM.
Regards
Robin
Add comments for Robin:
We want to create a pattern where value[ix] = floor (ix / NPATTERNS).
As NPATTERNS is always a power of two we can rewrite this as
= ix & -NPATTERNS.
`
Recently, I figure out a better approach in case of codegen for VLA stepped
vector.
Here is the detail descriptions:
Cas