[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: I think register coalescer is pretty much the same thing as gcc's early_remat. The llvm's register coalescer decide to remat this, because the return register is been used. And gcc's early_remat decide not to remat this. I put part of the gcc's log in here: cast_and_load_1.c.31

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: ping https://github.com/llvm/llvm-project/pull/163047 ___ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: @arsenm @preames @lukel97 @aengelke ping https://github.com/llvm/llvm-project/pull/163047 ___ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

@@ -202,13 +202,13 @@ define { <4 x i8>, <4 x i1> } @always_usub_const_vector() nounwind { ; SSE-LABEL: always_usub_const_vector: ; SSE: # %bb.0: ; SSE-NEXT:pcmpeqd %xmm0, %xmm0 -; SSE-NEXT:pcmpeqd %xmm1, %xmm1 +; SSE-NEXT:movdqa %xmm0, %xmm1

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: ping https://github.com/llvm/llvm-project/pull/163047 ___ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

@@ -202,13 +202,13 @@ define { <4 x i8>, <4 x i1> } @always_usub_const_vector() nounwind { ; SSE-LABEL: always_usub_const_vector: ; SSE: # %bb.0: ; SSE-NEXT:pcmpeqd %xmm0, %xmm0 -; SSE-NEXT:pcmpeqd %xmm1, %xmm1 +; SSE-NEXT:movdqa %xmm0, %xmm1

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: > From the motivation case, do you know why MachineCSE fails to optimize? Machine CSE is before this register coalescer, maybe add a CSE after this remat will work. https://github.com/llvm/llvm-project/pull/163047 ___ cfe-commits maili

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

@@ -202,13 +202,13 @@ define { <4 x i8>, <4 x i1> } @always_usub_const_vector() nounwind { ; SSE-LABEL: always_usub_const_vector: ; SSE: # %bb.0: ; SSE-NEXT:pcmpeqd %xmm0, %xmm0 -; SSE-NEXT:pcmpeqd %xmm1, %xmm1 +; SSE-NEXT:movdqa %xmm0, %xmm1

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: > > Machine CSE is before this register coalescer, Machine CSE will see > > something like: > > ``` > > li a2, 42 > > a0 = copy a2 > > ``` > > > > > > > > > > > > > > > > > > > > > > > > So Machine CSE will not eliminate this

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

@@ -202,13 +202,13 @@ define { <4 x i8>, <4 x i1> } @always_usub_const_vector() nounwind { ; SSE-LABEL: always_usub_const_vector: ; SSE: # %bb.0: ; SSE-NEXT:pcmpeqd %xmm0, %xmm0 -; SSE-NEXT:pcmpeqd %xmm1, %xmm1 +; SSE-NEXT:movdqa %xmm0, %xmm1

[clang] [llvm] [RegisterCoalescer] Improve register allocation for return values by limiting rematerialization (PR #163047)

rez5427 wrote: > > Machine CSE is before this register coalescer, Machine CSE will see > > something like: > > ``` > > li a2, 42 > > a0 = copy a2 > > ``` > > > > > > > > > > > > > > > > > > > > > > > > So Machine CSE will not eliminate this