[PATCH] [AArch64 NEON] Fix pattern match failed on FP_ROUND from v1f128 to v1f64.
Weiming Zhao
weimingz at codeaurora.org
Fri Jan 24 10:47:55 PST 2014
Hi Kevin,
This patch looks good to me.
Actually, we're just going to submit a patch for fp_round, which is almost the same as yours.
Thanks for the patch.
Weiming
-----Original Message-----
From: llvm-commits-bounces at cs.uiuc.edu [mailto:llvm-commits-bounces at cs.uiuc.edu] On Behalf Of Kevin Qin
Sent: Friday, January 24, 2014 2:38 AM
To: kevinqindev at gmail.com; t.p.northover at gmail.com
Cc: llvm-commits at cs.uiuc.edu
Subject: [PATCH] [AArch64 NEON] Fix pattern match failed on FP_ROUND from v1f128 to v1f64.
Hi,
Please review, thanks.
http://llvm-reviews.chandlerc.com/D2613
Files:
lib/CodeGen/SelectionDAG/LegalizeTypes.h
lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
test/CodeGen/AArch64/neon-fpround_f128.ll
Index: lib/CodeGen/SelectionDAG/LegalizeTypes.h
===================================================================
--- lib/CodeGen/SelectionDAG/LegalizeTypes.h
+++ lib/CodeGen/SelectionDAG/LegalizeTypes.h
@@ -541,6 +541,7 @@
SDValue ScalarizeVecOp_CONCAT_VECTORS(SDNode *N);
SDValue ScalarizeVecOp_EXTRACT_VECTOR_ELT(SDNode *N);
SDValue ScalarizeVecOp_STORE(StoreSDNode *N, unsigned OpNo);
+ SDValue ScalarizeVecOp_FP_ROUND(SDNode *N, unsigned OpNo);
//===--------------------------------------------------------------------===//
// Vector Splitting Support: LegalizeVectorTypes.cpp
Index: lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
+++ lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
@@ -384,6 +384,9 @@
case ISD::STORE:
Res = ScalarizeVecOp_STORE(cast<StoreSDNode>(N), OpNo);
break;
+ case ISD::FP_ROUND:
+ Res = ScalarizeVecOp_FP_ROUND(N, OpNo);
+ break;
}
}
@@ -467,6 +470,15 @@
N->getOriginalAlignment(), N->getTBAAInfo()); }
+/// ScalarizeVecOp_FP_ROUND - If the value to round is a vector that
+needs /// to be scalarized, it must be <1 x ty>. Convert the element instead.
+SDValue DAGTypeLegalizer::ScalarizeVecOp_FP_ROUND(SDNode *N, unsigned
+OpNo) {
+ SDValue Elt = GetScalarizedVector(N->getOperand(0));
+ SDValue Res = DAG.getNode(ISD::FP_ROUND, SDLoc(N),
+ N->getValueType(0).getVectorElementType(), Elt,
+ N->getOperand(1));
+ return DAG.getNode(ISD::SCALAR_TO_VECTOR, SDLoc(N),
+N->getValueType(0), Res); }
//===----------------------------------------------------------------------===//
// Result Vector Splitting
Index: test/CodeGen/AArch64/neon-fpround_f128.ll
===================================================================
--- /dev/null
+++ test/CodeGen/AArch64/neon-fpround_f128.ll
@@ -0,0 +1,18 @@
+; RUN: llc < %s -verify-machineinstrs -mtriple=aarch64-none-linux-gnu
+-mattr=+neon -fp-contract=fast | FileCheck %s
+
+define <1 x double> @test_fpround_v1f128(<1 x fp128>* %a) { ;
+CHECK-LABEL: test_fpround_v1f128:
+; CHECK: bl __trunctfdf2
+ %b = load <1 x fp128>* %a
+ %c = fptrunc <1 x fp128> %b to <1 x double>
+ ret <1 x double> %c
+}
+
+define <2 x double> @test_fpround_v2f128(<2 x fp128>* %a) { ;
+CHECK-LABEL: test_fpround_v2f128:
+; CHECK: bl __trunctfdf2
+; CHECK: bl __trunctfdf2
+ %b = load <2 x fp128>* %a
+ %c = fptrunc <2 x fp128> %b to <2 x double>
+ ret <2 x double> %c
+}
More information about the llvm-commits
mailing list