gem5 v24.0.0.0
Loading...
Searching...
No Matches
scalar_register_file.cc
Go to the documentation of this file.
1/*
2 * Copyright (c) 2015-2017 Advanced Micro Devices, Inc.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions are met:
7 *
8 * 1. Redistributions of source code must retain the above copyright notice,
9 * this list of conditions and the following disclaimer.
10 *
11 * 2. Redistributions in binary form must reproduce the above copyright notice,
12 * this list of conditions and the following disclaimer in the documentation
13 * and/or other materials provided with the distribution.
14 *
15 * 3. Neither the name of the copyright holder nor the names of its
16 * contributors may be used to endorse or promote products derived from this
17 * software without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
20 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
23 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
33
34#include "base/logging.hh"
35#include "debug/GPUSRF.hh"
38#include "gpu-compute/shader.hh"
40#include "params/ScalarRegisterFile.hh"
41
42namespace gem5
43{
44
45ScalarRegisterFile::ScalarRegisterFile(const ScalarRegisterFileParams &p)
47{
48 regFile.resize(numRegs(), 0);
49}
50
51bool
53{
54 for (const auto& srcScalarOp : ii->srcScalarRegOperands()) {
55 for (const auto& physIdx : srcScalarOp.physIndices()) {
56 if (regBusy(physIdx)) {
57 DPRINTF(GPUSRF, "RAW stall: WV[%d]: %s: physReg[%d]\n",
58 w->wfDynId, ii->disassemble(), physIdx);
59 w->stats.numTimesBlockedDueRAWDependencies++;
60 return false;
61 }
62 }
63 }
64
65 for (const auto& dstScalarOp : ii->dstScalarRegOperands()) {
66 for (const auto& physIdx : dstScalarOp.physIndices()) {
67 if (regBusy(physIdx)) {
68 DPRINTF(GPUSRF, "WAX stall: WV[%d]: %s: physReg[%d]\n",
69 w->wfDynId, ii->disassemble(), physIdx);
70 w->stats.numTimesBlockedDueWAXDependencies++;
71 return false;
72 }
73 }
74 }
75
76 return true;
77}
78
79void
81{
82 for (const auto& dstScalarOp : ii->dstScalarRegOperands()) {
83 for (const auto& physIdx : dstScalarOp.physIndices()) {
84 // mark the destination scalar register as busy
85 markReg(physIdx, true);
86 }
87 }
88}
89
90void
92{
93 stats.registerReads += ii->numSrcScalarDWords();
94
95 if (!ii->isLoad() && !(ii->isAtomic() || ii->isMemSync())) {
97 Tick tickDelay = computeUnit->cyclesToTicks(delay);
98
99 for (const auto& dstScalarOp : ii->dstScalarRegOperands()) {
100 for (const auto& physIdx : dstScalarOp.physIndices()) {
101 enqRegFreeEvent(physIdx, tickDelay);
102 }
103 }
104
105 stats.registerWrites += ii->numDstScalarDWords();
106 }
107}
108
109void
111 GPUDynInstPtr ii)
112{
113 assert(ii->isLoad() || ii->isAtomicRet());
114 for (const auto& dstScalarOp : ii->dstScalarRegOperands()) {
115 for (const auto& physIdx : dstScalarOp.physIndices()) {
117 }
118 }
119
120 stats.registerWrites += ii->numDstScalarDWords();
121}
122
123} // namespace gem5
#define DPRINTF(x,...)
Definition trace.hh:210
Tick cyclesToTicks(Cycles c) const
Tick clockPeriod() const
int scalarPipeLength() const
Cycles is a wrapper class for representing cycle counts, i.e.
Definition types.hh:79
gem5::RegisterFile::RegisterFileStats stats
virtual bool regBusy(int idx) const
virtual void markReg(int regIdx, bool value)
virtual void enqRegFreeEvent(uint32_t regIdx, uint64_t delay)
ComputeUnit * computeUnit
virtual void scheduleWriteOperandsFromLoad(Wavefront *w, GPUDynInstPtr ii) override
ScalarRegisterFile(const ScalarRegisterFileParams &p)
virtual bool operandsReady(Wavefront *w, GPUDynInstPtr ii) const override
virtual void scheduleWriteOperands(Wavefront *w, GPUDynInstPtr ii) override
std::vector< ScalarRegU32 > regFile
virtual void waveExecuteInst(Wavefront *w, GPUDynInstPtr ii) override
Bitfield< 0 > p
Bitfield< 0 > w
Copyright (c) 2024 - Pranith Kumar Copyright (c) 2020 Inria All rights reserved.
Definition binary32.hh:36
std::shared_ptr< GPUDynInst > GPUDynInstPtr
Definition misc.hh:49
uint64_t Tick
Tick count type.
Definition types.hh:58

Generated on Tue Jun 18 2024 16:24:04 for gem5 by doxygen 1.11.0