| 1 | /***************************************************************************** |
| 2 | * Copyright (C) 2013 x265 project |
| 3 | * |
| 4 | * Authors: Steve Borho <steve@borho.org> |
| 5 | * |
| 6 | * This program is free software; you can redistribute it and/or modify |
| 7 | * it under the terms of the GNU General Public License as published by |
| 8 | * the Free Software Foundation; either version 2 of the License, or |
| 9 | * (at your option) any later version. |
| 10 | * |
| 11 | * This program is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | * GNU General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU General Public License |
| 17 | * along with this program; if not, write to the Free Software |
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. |
| 19 | * |
| 20 | * This program is also available under a commercial proprietary license. |
| 21 | * For more information, contact us at license @ x265.com |
| 22 | *****************************************************************************/ |
| 23 | |
| 24 | #include "threadpool.h" |
| 25 | #include "threading.h" |
| 26 | #include "wavefront.h" |
| 27 | #include "common.h" |
| 28 | |
| 29 | namespace x265 { |
| 30 | // x265 private namespace |
| 31 | |
| 32 | bool WaveFront::init(int numRows) |
| 33 | { |
| 34 | m_numRows = numRows; |
| 35 | |
| 36 | m_numWords = (numRows + 63) >> 6; |
| 37 | m_internalDependencyBitmap = X265_MALLOC(uint64_t, m_numWords); |
| 38 | if (m_internalDependencyBitmap) |
| 39 | memset((void*)m_internalDependencyBitmap, 0, sizeof(uint64_t) * m_numWords); |
| 40 | |
| 41 | m_externalDependencyBitmap = X265_MALLOC(uint64_t, m_numWords); |
| 42 | if (m_externalDependencyBitmap) |
| 43 | memset((void*)m_externalDependencyBitmap, 0, sizeof(uint64_t) * m_numWords); |
| 44 | |
| 45 | return m_internalDependencyBitmap && m_externalDependencyBitmap; |
| 46 | } |
| 47 | |
| 48 | WaveFront::~WaveFront() |
| 49 | { |
| 50 | x265_free((void*)m_internalDependencyBitmap); |
| 51 | x265_free((void*)m_externalDependencyBitmap); |
| 52 | } |
| 53 | |
| 54 | void WaveFront::clearEnabledRowMask() |
| 55 | { |
| 56 | memset((void*)m_externalDependencyBitmap, 0, sizeof(uint64_t) * m_numWords); |
| 57 | } |
| 58 | |
| 59 | void WaveFront::enqueueRow(int row) |
| 60 | { |
| 61 | // thread safe |
| 62 | uint64_t bit = 1LL << (row & 63); |
| 63 | |
| 64 | X265_CHECK(row < m_numRows, "invalid row\n"); |
| 65 | ATOMIC_OR(&m_internalDependencyBitmap[row >> 6], bit); |
| 66 | if (m_pool) m_pool->pokeIdleThread(); |
| 67 | } |
| 68 | |
| 69 | void WaveFront::enableRow(int row) |
| 70 | { |
| 71 | // thread safe |
| 72 | uint64_t bit = 1LL << (row & 63); |
| 73 | |
| 74 | X265_CHECK(row < m_numRows, "invalid row\n"); |
| 75 | ATOMIC_OR(&m_externalDependencyBitmap[row >> 6], bit); |
| 76 | } |
| 77 | |
| 78 | void WaveFront::enableAllRows() |
| 79 | { |
| 80 | memset((void*)m_externalDependencyBitmap, ~0, sizeof(uint64_t) * m_numWords); |
| 81 | } |
| 82 | |
| 83 | bool WaveFront::checkHigherPriorityRow(int curRow) |
| 84 | { |
| 85 | int fullwords = curRow >> 6; |
| 86 | uint64_t mask = (1LL << (curRow & 63)) - 1; |
| 87 | |
| 88 | // Check full bitmap words before curRow |
| 89 | for (int i = 0; i < fullwords; i++) |
| 90 | { |
| 91 | if (m_internalDependencyBitmap[i] & m_externalDependencyBitmap[i]) |
| 92 | return true; |
| 93 | } |
| 94 | |
| 95 | // check the partially masked bitmap word of curRow |
| 96 | if (m_internalDependencyBitmap[fullwords] & m_externalDependencyBitmap[fullwords] & mask) |
| 97 | return true; |
| 98 | return false; |
| 99 | } |
| 100 | |
| 101 | bool WaveFront::dequeueRow(int row) |
| 102 | { |
| 103 | uint64_t oldval, newval; |
| 104 | |
| 105 | oldval = m_internalDependencyBitmap[row >> 6]; |
| 106 | newval = oldval & ~(1LL << (row & 63)); |
| 107 | return ATOMIC_CAS(&m_internalDependencyBitmap[row >> 6], oldval, newval) == oldval; |
| 108 | } |
| 109 | |
| 110 | bool WaveFront::findJob(int threadId) |
| 111 | { |
| 112 | unsigned long id; |
| 113 | |
| 114 | // thread safe |
| 115 | for (int w = 0; w < m_numWords; w++) |
| 116 | { |
| 117 | uint64_t oldval = m_internalDependencyBitmap[w]; |
| 118 | while (oldval & m_externalDependencyBitmap[w]) |
| 119 | { |
| 120 | uint64_t mask = oldval & m_externalDependencyBitmap[w]; |
| 121 | |
| 122 | CTZ64(id, mask); |
| 123 | |
| 124 | uint64_t newval = oldval & ~(1LL << id); |
| 125 | if (ATOMIC_CAS(&m_internalDependencyBitmap[w], oldval, newval) == oldval) |
| 126 | { |
| 127 | // we cleared the bit, process row |
| 128 | processRow(w * 64 + id, threadId); |
| 129 | return true; |
| 130 | } |
| 131 | // some other thread cleared the bit, try another bit |
| 132 | oldval = m_internalDependencyBitmap[w]; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | // made it through the bitmap without finding any enqueued rows |
| 137 | return false; |
| 138 | } |
| 139 | } |