[PATCH] D47374: [RFC][patch 3/3] Add support for variant scheduling classes in llvm-mca.

[Andrea Di Biagio via Phabricator via llvm-commits]

andreadb added inline comments.


================
Comment at: CodeGen/X86/sse-schedule.ll:6228
 ; BTVER2-SSE:       # %bb.0:
-; BTVER2-SSE-NEXT:    xorps %xmm0, %xmm0 # sched: [1:0.50]
+; BTVER2-SSE-NEXT:    xorps %xmm0, %xmm0 # sched: [0:?]
 ; BTVER2-SSE-NEXT:    #APP
----------------
RKSimon wrote:
> These are unfortunate - please can you raise an upstream bugzilla about throughput printing defaulting to issues width.
Makes sense. I will raise a bug for it.


================
Comment at: llvm-mca/InstrBuilder.cpp:388
+  if (!SchedClassID)
+    llvm::report_fatal_error("unable to resolve this variant class.");
 
----------------
RKSimon wrote:
> So !SchedClassID can only occur here due to variant resolution failing?
Good point.
Strictly speaking, the scheduling class ID associated with MCI *should* always be valid. But I cannot guarantee it.. I will add an if-stmt to guard against invalid classes.


https://reviews.llvm.org/D47374