diff options
author | Justin Holewinski <jholewinski@nvidia.com> | 2012-05-04 20:18:50 +0000 |
---|---|---|
committer | Justin Holewinski <jholewinski@nvidia.com> | 2012-05-04 20:18:50 +0000 |
commit | 49683f3c961379fbc088871a5d6304950f1f1cbc (patch) | |
tree | 830fa1ee9c992ef4645863d128be912ce2bfc987 /lib/Target/NVPTX/NVPTXLowerAggrCopies.h | |
parent | 2c7e5c714c8675f757c4936a3a2132c2466a626c (diff) | |
download | llvm-49683f3c961379fbc088871a5d6304950f1f1cbc.tar.gz llvm-49683f3c961379fbc088871a5d6304950f1f1cbc.tar.bz2 llvm-49683f3c961379fbc088871a5d6304950f1f1cbc.tar.xz |
This patch adds a new NVPTX back-end to LLVM which supports code generation for NVIDIA PTX 3.0. This back-end will (eventually) replace the current PTX back-end, while maintaining compatibility with it.
The new target machines are:
nvptx (old ptx32) => 32-bit PTX
nvptx64 (old ptx64) => 64-bit PTX
The sources are based on the internal NVIDIA NVPTX back-end, and
contain more functionality than the current PTX back-end currently
provides.
NV_CONTRIB
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@156196 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Target/NVPTX/NVPTXLowerAggrCopies.h')
-rw-r--r-- | lib/Target/NVPTX/NVPTXLowerAggrCopies.h | 47 |
1 files changed, 47 insertions, 0 deletions
diff --git a/lib/Target/NVPTX/NVPTXLowerAggrCopies.h b/lib/Target/NVPTX/NVPTXLowerAggrCopies.h new file mode 100644 index 0000000000..ac7f1509f2 --- /dev/null +++ b/lib/Target/NVPTX/NVPTXLowerAggrCopies.h @@ -0,0 +1,47 @@ +//===-- llvm/lib/Target/NVPTX/NVPTXLowerAggrCopies.h ------------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the declaration of the NVIDIA specific lowering of +// aggregate copies +// +//===----------------------------------------------------------------------===// + +#ifndef NVPTX_LOWER_AGGR_COPIES_H +#define NVPTX_LOWER_AGGR_COPIES_H + +#include "llvm/Pass.h" +#include "llvm/CodeGen/MachineFunctionAnalysis.h" +#include "llvm/Target/TargetData.h" + +namespace llvm { + +// actual analysis class, which is a functionpass +struct NVPTXLowerAggrCopies : public FunctionPass { + static char ID; + + NVPTXLowerAggrCopies() : FunctionPass(ID) {} + + void getAnalysisUsage(AnalysisUsage &AU) const { + AU.addRequired<TargetData>(); + AU.addPreserved<MachineFunctionAnalysis>(); + } + + virtual bool runOnFunction(Function &F); + + static const unsigned MaxAggrCopySize = 128; + + virtual const char *getPassName() const { + return "Lower aggregate copies/intrinsics into loops"; + } +}; + +extern FunctionPass *createLowerAggrCopies(); +} + +#endif |