From 5857bf5d1e5a5ffe5ae51a38514ee55495c0cc69 Mon Sep 17 00:00:00 2001 From: Sjoerd Meijer Date: Thu, 30 May 2019 08:07:06 +0000 Subject: [PATCH] [ARM] Add an MVE execution domain MVE architecturally specifies a 'beat' system in which a vector instruction executed now will complete its actual operation over the next four cycles, so it can overlap with the execution of the previous and next MVE instruction. This makes it generally an advantage to avoid moving values back and forth between MVE registers and anywhere else, if there's any sensible way to do the same processing in whatever register type the values already occupied. That's just what the 'execution domain' system is supposed to achieve. So here we add a new execution domain which will contain all the MVE vector instructions when they are added. Patch by: Simon Tatham Differential Revision: https://reviews.llvm.org/D60703 llvm-svn: 362068 --- llvm/lib/Target/ARM/ARMInstrFormats.td | 9 +++++---- llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h | 5 +++-- 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/llvm/lib/Target/ARM/ARMInstrFormats.td b/llvm/lib/Target/ARM/ARMInstrFormats.td index 00a7202..b09aa8a 100644 --- a/llvm/lib/Target/ARM/ARMInstrFormats.td +++ b/llvm/lib/Target/ARM/ARMInstrFormats.td @@ -120,14 +120,15 @@ def IndexModePost : IndexMode<2>; def IndexModeUpd : IndexMode<3>; // Instruction execution domain. -class Domain val> { - bits<3> Value = val; +class Domain val> { + bits<4> Value = val; } def GenericDomain : Domain<0>; def VFPDomain : Domain<1>; // Instructions in VFP domain only def NeonDomain : Domain<2>; // Instructions in Neon domain only def VFPNeonDomain : Domain<3>; // Instructions in both VFP & Neon domains def VFPNeonA8Domain : Domain<5>; // Instructions in VFP & Neon under A8 +def MVEDomain : Domain<8>; // Instructions in MVE and ARMv8.1m //===----------------------------------------------------------------------===// // ARM special operands. @@ -324,8 +325,8 @@ class InstTemplate