[PATCH] D124869: [RISCV] Hoist VSETVLI out of idiomatic fixed length vector loops

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp:1293
+
+  if (RISCVII::LMUL_1 != Info.getVLMUL())
+    return false;
----------------
reames wrote:
> craig.topper wrote:
> > What prevents supporting other LMULs?
> Nothing, we just need to adjust the formula below for it.  Mind if I do this in a separate change?  I'm still not 100% trusting my mental model of L:MUL, and would prefer to have that reviewed on its own.  
> 
> I will add a TODO.  
Separate review is fine.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124869/new/

https://reviews.llvm.org/D124869