[PATCH] D93594: [X86] Pass to transform amx intrinsics to scalar operation.

[Bing Yu via Phabricator via cfe-commits]

yubing added inline comments.


================
Comment at: llvm/lib/Target/X86/X86LowerAMXIntrinsics.cpp:311
+  Value *ResElt = B.CreateAdd(EltC, SubVecR);
+  Value *NewVecC = B.CreateInsertElement(VecCPhi, ResElt, IdxC);
+  Value *NewVecD = B.CreateInsertElement(VecDPhi, ResElt, IdxC);
----------------
LuoYuanke wrote:
> yubing wrote:
> > pengfei wrote:
> > > yubing wrote:
> > > > pengfei wrote:
> > > > > Is it necessary to insert the ResElt to VecC?
> > > > Yes, it is necessary since you should use updated eltC(aka, Cij) when you are doing matrix dotproduct:
> > > > Cij =Cij+Ai1.*B1j
> > > > Cij =Cij+Ai2.*B2j
> > > > ....
> > > > Cij =Cij+AiK.*BKj
> > > But you don't need to update both C and D. Something like the psudo code should enough:
> > > ```
> > > for (k : K)
> > >   Dij += Aik * Bkj;
> > > Dij += Cij
> > > ```
> > I change code into the following style, and it can also reduce inner loop's size:
> > ```
> > for (k : K)
> >   Cij += Aik * Bkj;
> > Dij = Cij
> > ```
> > Besides, I hoist the procedure of calculating (i,j)'s linear index above inner loops.
> It seems keeping vector C unchanged is simpler. We can eliminate the phi, extract and insert instruction for vector C.
But your solution  still need to update D so D's phi will be kept in the inner loops.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D93594/new/

https://reviews.llvm.org/D93594