algo_chooser.cpp 31.7 KB
Newer Older
1 2 3 4
/**
 * \file src/opr/impl/search_policy/algo_chooser.cpp
 * MegEngine is Licensed under the Apache License, Version 2.0 (the "License")
 *
5
 * Copyright (c) 2014-2021 Megvii Inc. All rights reserved.
6 7 8 9 10 11 12 13
 *
 * Unless required by applicable law or agreed to in writing,
 * software distributed under the License is distributed on an
 * "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or
 * implied.
 */

#include "megbrain/opr/search_policy/algo_chooser.h"
14
#include <limits>
15 16
#include <unordered_set>
#include "megbrain/opr/dnn/convolution.h"
17
#include "megbrain/opr/internal/megdnn_opr_wrapper.h"
18
#include "megbrain/opr/search_policy/algo_chooser_helper.h"
19 20 21 22 23 24 25 26
#include "megbrain/opr/search_policy/profiler.h"

#include "../internal/invoke.h"
#include "../internal/megdnn_opr_wrapper.inl"
#include "./workspace_need_limit_getter.inl"

//! TODO: here has to be know some megdnn::opr when there is produced midout.h
//! fix it if there is another graceful way.
27
#include "megdnn/opr_param_defs.h"
28
#include "megdnn/oprs.h"
29
#include "megdnn/oprs/base.h"
30 31 32 33 34 35 36 37
#include "midout.h"
MIDOUT_DECL(megbrain_opr_algo_chooser)
#define MIDOUT_B(...) MIDOUT_BEGIN(megbrain_opr_algo_chooser, __VA_ARGS__) {
#define MIDOUT_E \
    }            \
    MIDOUT_END();

using mgb::opr::intl::WorkspaceLimitGetter;
38 39
using namespace megdnn;
using namespace mgb;
40 41 42 43 44 45 46 47

#define APPLY(statement, ...)                                  \
    mgb::apply([&](const auto&... args) { return statement; }, \
               std::tuple_cat(__VA_ARGS__))

// timeout delta to be added with fastest known algorithm for new algos
constexpr double TIMEOUT_TOLERANCE = 2;

48
#define CACHE_KEY_VERSION "v4"
49 50 51 52 53 54 55 56 57 58

namespace {
template <typename Opr>
std::string profile_name(Opr* opr) {
    std::string ret =
            std::string(MegDNNOpr2MGBOpr<Opr>::MGBOpr::typeinfo()->name) +
            CACHE_KEY_VERSION;
    ret.append(opr->get_algorithm_set_name());
    return ret;
}
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83

template <typename Opr>
std::string format_fixlayouts(
        const typename opr::AlgoChooser<Opr>::FixedTensorLayouts& layouts,
        size_t arity_in, size_t arity_out) {
    std::string ret;
    ret.append(": tensor layouts(");
    for (size_t i = 0; i < arity_in; ++i) {
        if (i) {
            ret.append(", ");
        }
        ret.append(layouts[i].to_string() + " ");
        ret.append(layouts[i].dtype.name());
    }
    ret.append(") -> (");
    for (size_t i = 0; i < arity_out; ++i) {
        if (i) {
            ret.append(", ");
        }
        ret.append(layouts[i + arity_in].to_string() + " ");
        ret.append(layouts[i + arity_in].dtype.name());
    }
    return ret;
}

84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
/**
 * \brief Check if the sub opr list has circular dependence.
 */
class CircularDepsChecker {
    struct SearchItemStorage {
        std::string data_hold;
        size_t hash = 0;

        SearchItemStorage(const Algorithm::SearchItem& item) {
            Algorithm::serialize_write_pod(item.opr_type, data_hold);
            for (auto&& layout : item.layouts) {
                data_hold += layout.serialize();
            }
            data_hold += item.param;
        }

        SearchItemStorage& init_hash() {
            hash = XXHash64CT::hash(data_hold.data(), data_hold.size(),
                                    20201225);
            return *this;
        }

        bool operator==(const SearchItemStorage& rhs) const {
            return data_hold == rhs.data_hold;
        }

        struct Hash {
            size_t operator()(const SearchItemStorage& s) const {
                return s.hash;
            }
        };
    };
    std::unordered_set<SearchItemStorage, SearchItemStorage::Hash> m_set;

public:
    void put(const megdnn::Algorithm::SearchItem& key) {
        SearchItemStorage key_storage(key);
        key_storage.init_hash();
        mgb_assert(m_set.find(key_storage) == m_set.end(),
                   "Circular dependency during flatten search space");
        auto ret = m_set.insert(std::move(key_storage));
        mgb_assert(ret.second);
    }
    void remove(const megdnn::Algorithm::SearchItem& key) {
        SearchItemStorage key_storage(key);
        key_storage.init_hash();
        auto&& iter = m_set.find(key_storage);
        mgb_assert(iter != m_set.end());
        m_set.erase(iter);
    }
};

136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
///////////////// OprTypeTrait /////////////////////////////
template <megdnn::Algorithm::OprType>
struct OprFromOprTypeTrait;

template <typename Opr>
struct OprTypeFromOprTrait;

#define cb(_opr_type, _opr)                                             \
    template <>                                                         \
    struct OprFromOprTypeTrait<megdnn::Algorithm::OprType::_opr_type> { \
        using Opr = megdnn::_opr;                                       \
    };                                                                  \
    template <>                                                         \
    struct OprTypeFromOprTrait<megdnn::_opr> {                          \
        constexpr static megdnn::Algorithm::OprType opr_type =          \
                megdnn::Algorithm::OprType::_opr_type;                  \
    }

cb(MATRIX_MUL_FORWARD, MatrixMulForward);
cb(BATCHED_MATRIX_MUL_FORWARD, BatchedMatrixMulForward);
cb(CONVOLUTION_FORWARD, ConvolutionForward);
cb(CONVOLUTION_BACKWARD_DATA, ConvolutionBackwardData);
cb(CONVOLUTION_BACKWARD_FILTER, ConvolutionBackwardFilter);
cb(CONVOLUTION3D_FORWARD, Convolution3DForward);
cb(CONVOLUTION3D_BACKWARD_DATA, Convolution3DBackwardData);
cb(CONVOLUTION3D_BACKWARD_FILTER, Convolution3DBackwardFilter);
cb(LOCAL_SHARE_FORWARD, LocalShareForward);
cb(LOCAL_SHARE_BACKWARD_DATA, LocalShareBackwardData);
cb(LOCAL_SHARE_BACKWARD_FILTER, LocalShareBackwardFilter);
cb(DEFORMABLE_CONV_FORWARD, DeformableConvForward);
cb(DEFORMABLE_CONV_BACKWARD_DATA, DeformableConvBackwardData);
cb(DEFORMABLE_CONV_BACKWARD_FILTER, DeformableConvBackwardFilter);
cb(BATCH_CONV_FORWARD, BatchConvBiasForward);
cb(CONVBIAS_FORWARD, ConvBiasForward);

#undef cb

// clang-format off
#define FOREACH_OPR_TYPE_WITH_STMT(cb, stmt)  \
    cb(MATRIX_MUL_FORWARD, stmt)              \
    cb(BATCHED_MATRIX_MUL_FORWARD, stmt)      \
    cb(CONVOLUTION_FORWARD, stmt)             \
    cb(CONVOLUTION_BACKWARD_DATA, stmt)       \
    cb(CONVOLUTION_BACKWARD_FILTER, stmt)     \
    cb(CONVOLUTION3D_FORWARD, stmt)           \
    cb(CONVOLUTION3D_BACKWARD_DATA, stmt)     \
    cb(CONVOLUTION3D_BACKWARD_FILTER, stmt)   \
    cb(LOCAL_SHARE_FORWARD, stmt)             \
    cb(LOCAL_SHARE_BACKWARD_DATA, stmt)       \
    cb(LOCAL_SHARE_BACKWARD_FILTER, stmt)     \
    cb(DEFORMABLE_CONV_FORWARD, stmt)         \
    cb(DEFORMABLE_CONV_BACKWARD_DATA, stmt)   \
    cb(DEFORMABLE_CONV_BACKWARD_FILTER, stmt) \
    cb(BATCH_CONV_FORWARD, stmt)              \
    cb(CONVBIAS_FORWARD, stmt)
// clang-format on

#define _OPR_TYPE_CASE(_opr_type, _stmt)             \
    case Algorithm::OprType::_opr_type: {            \
        using _Opr = typename OprFromOprTypeTrait<   \
                Algorithm::OprType::_opr_type>::Opr; \
        _stmt;                                       \
        break;                                       \
    }

#define FOREACH_OPR_TYPE_DISPATCH(_search_items, _stmt)          \
    for (size_t _item_idx = 0; _item_idx < _search_items.size(); \
         _item_idx++) {                                          \
        auto&& _item = _search_items[_item_idx];                 \
        switch (_item.opr_type) {                                \
            FOREACH_OPR_TYPE_WITH_STMT(_OPR_TYPE_CASE, _stmt)    \
            default:                                             \
                mgb_throw(MegBrainError, "unknown opr_type");    \
        }                                                        \
    }

template <typename Opr>
TensorLayoutArray to_layout_array(
        const typename opr::AlgoChooser<Opr>::FixedTensorLayouts& layouts) {
    TensorLayoutArray ret;
    for (auto&& layout : layouts) {
        ret.push_back(layout);
    }
    return ret;
220 221
}

222 223 224 225 226 227 228 229 230 231 232 233
template <typename Opr>
typename opr::AlgoChooser<Opr>::FixedTensorLayouts to_fixed_layouts(
        const TensorLayoutArray& layouts) {
    typename opr::AlgoChooser<Opr>::FixedTensorLayouts ret;
    mgb_assert(ret.size() == layouts.size());
    size_t idx = 0;
    for (auto&& layout : layouts) {
        ret[idx++] = layout;
    }
    return ret;
}

234 235 236 237 238 239 240 241 242 243 244 245
/**
 * flatten search space in postorder traversal
 * The subopr search construct a search tree
 *
 *           A
 *        /    \
 *       B1B2   C
 *      /     \
 *     D1D2D3   E
 * We use postorder traverse the search tree.
 * D1 -> D2 -> D3 -> E -> B1 -> B2 -> C -> A
 */
246
template <typename Opr>
247 248 249 250 251 252 253
std::vector<megdnn::Algorithm::SearchItem> flatten_search_space(
        const typename opr::AlgoChooser<Opr>::ExeContext& ctx,
        CircularDepsChecker& checker) {
    auto&& search_item = megdnn::Algorithm::SearchItem{
            OprTypeFromOprTrait<Opr>::opr_type, ctx.param(),
            to_layout_array<Opr>(ctx.layouts())};
    checker.put(search_item);
254 255 256 257 258 259 260 261 262
    std::vector<megdnn::Algorithm::SearchItem> ret;
    for (auto algo_info : ctx.get_all_candidates()) {
        megdnn::Algorithm* algo = ctx.get_algorithm_from_desc(algo_info.desc);
        mgb_assert(algo, "Unknown algo description");
        std::vector<megdnn::Algorithm::SearchItem>&& sub_items =
                algo->get_subopr_list(to_layout_array<Opr>(ctx.layouts()),
                                      ctx.megdnn_opr());

        FOREACH_OPR_TYPE_DISPATCH(sub_items, {
263 264
            auto&& megdnn_opr =
                    opr::intl::create_megdnn_opr<_Opr>(ctx.comp_node());
265 266 267
            megdnn_opr->param() =
                    Algorithm::deserialize_read_pod<typename _Opr::Param>(
                            _item.param);
268
            typename opr::AlgoChooser<_Opr>::ExeContext sub_ctx(
269 270 271
                    to_fixed_layouts<_Opr>(_item.layouts), megdnn_opr.get(),
                    _item.param, ctx.mgb_opr(), ctx.comp_node(),
                    ctx.execution_policy(), ctx.allow_weight_preprocess());
272
            auto space = flatten_search_space<_Opr>(sub_ctx, checker);
273 274
            ret.insert(ret.end(), space.begin(), space.end());
        });
275
    }
276 277
    ret.push_back(search_item);
    checker.remove(search_item);
278 279
    return ret;
}
280

281 282 283 284 285
}  // namespace

namespace mgb {
namespace opr {

286 287 288 289
template <typename Opr>
void AlgoChooser<Opr>::profile(ExeContext& ctx, bool require_reproducible) {
    if (ctx.get_profile_result_from_cache(require_reproducible).valid())
        return;
290 291 292 293 294 295
    AlgoChooserProfileCache::Result prof_rst;

    std::string str_on_inp_shape = ssprintf(
            "on input layouts (%s, %s)", ctx.layouts()[0].to_string().c_str(),
            ctx.layouts()[1].to_string().c_str());
    double cur_timeout = 0;
296 297 298 299

    auto workspace_limit = WorkspaceLimitGetter::get_workspace_limit(
            ctx.owner_graph(), ctx.comp_node(),
            ctx.execution_policy().workspace_limit);
300
    RealTimer timer;
301
    for (auto algo : ctx.get_all_candidates()) {
302 303 304
        Maybe<AlgoChooserProfileCache::ResultEntry> cur_rst;
        std::string msg = ssprintf("profiling %s algorithm %s %s",
                                   ctx.mgb_opr()->dyn_typeinfo()->name,
305
                                   algo.name.c_str(), str_on_inp_shape.c_str());
306 307
        ImplExecutionPolicy policy;
        policy.algo = algo.desc;
308
        ctx.construct_execution_policy(require_reproducible, policy);
309 310 311
        if (ctx.get_workspace_size_bytes(policy) >= workspace_limit)
            continue;

312
        timer.reset();
313
        MGB_TRY { cur_rst = ctx.profile_single_algo(policy, cur_timeout); }
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341
        MGB_CATCH(std::exception & exc, {
            mgb_log_warn("caught exception during %s: %s", msg.c_str(),
                         exc.what());
            continue;
        })
        MGB_CATCH(..., {
            mgb_log_warn("caught exception during %s", msg.c_str());
            continue;
        })
        if (!cur_rst.valid()) {
            mgb_log_warn("timeout when %s; timeout setting: %.3fsec",
                         msg.c_str(), cur_timeout);
            continue;
        }
        if (!cur_timeout) {
            cur_timeout = timer.get_secs() + TIMEOUT_TOLERANCE;
        } else {
            cur_timeout =
                    std::min(cur_timeout, timer.get_secs() + TIMEOUT_TOLERANCE);
        }
        auto&& rst = cur_rst.val();
        mgb_log_debug("%s: workspace: %zu; time: %.3gsec", msg.c_str(),
                      rst.workspace, rst.time);
        prof_rst.push_back(rst);
    }
    mgb_assert(!prof_rst.empty(), "no usable convolution algorithm %s",
               str_on_inp_shape.c_str());

342 343 344 345 346 347 348 349
    FixedTensorLayouts origin_layouts = ctx.layouts();
    typename Opr::Param origin_param = ctx.megdnn_opr()->param();
    AlgoChooserProfileCache::Key cache_key{origin_layouts.data(),
                                           origin_layouts.size(), &origin_param,
                                           sizeof(origin_param)};

    AlgoChooserProfileCache cache(ctx.comp_node(),
                                  profile_name(ctx.megdnn_opr()).c_str());
350 351 352 353
    cache.put(cache_key, prof_rst);
}

template <typename Opr>
354 355 356
typename AlgoChooser<Opr>::ImplExecutionPolicy
AlgoChooser<Opr>::choose_by_profile(ExeContext& ctx, bool require_reproducible,
                                    bool enable_update) {
357
    MIDOUT_B(Opr, midout_iv(MGB_HASH_STR("AlgoChooser::choose_by_profile")))
358 359 360 361
    if (ctx.owner_graph()->options().no_profiling_on_shape_change) {
        auto policy = ctx.megdnn_opr()->execution_policy();
        if (policy.algo.valid())
            return policy;
362 363
    }

364
    if (enable_update) {
365 366 367
        CircularDepsChecker circular_deps_checker;
        auto&& search_items =
                flatten_search_space<Opr>(ctx, circular_deps_checker);
368 369 370 371 372 373 374 375 376 377 378
        FOREACH_OPR_TYPE_DISPATCH(search_items, {
            auto&& megdnn_opr = intl::create_megdnn_opr<_Opr>(ctx.comp_node());
            megdnn_opr->param() =
                    Algorithm::deserialize_read_pod<typename _Opr::Param>(
                            _item.param);
            typename AlgoChooser<_Opr>::ExeContext sub_ctx(
                    to_fixed_layouts<_Opr>(_item.layouts), megdnn_opr.get(),
                    _item.param, ctx.mgb_opr(), ctx.comp_node(),
                    ctx.execution_policy(), ctx.allow_weight_preprocess());
            AlgoChooser<_Opr>::profile(sub_ctx, require_reproducible);
        });
379
    }
380
    typename AlgoChooser<Opr>::ImplExecutionPolicy policy;
381
    ctx.construct_execution_policy(require_reproducible, policy);
382
    return policy;
383 384 385 386
    MIDOUT_E
}

template <typename Opr>
387
size_t AlgoChooser<Opr>::setup_algo(const FixedTensorLayouts& layouts,
388 389 390 391 392 393
                                    Opr* megdnn_opr, const MGBOpr* mgb_opr,
                                    bool allow_weight_preprocess) {
    if (WorkspaceLimitGetter::is_prealloc_run(mgb_opr->owner_graph())) {
        return 0;
    }

394 395 396 397 398
    std::string param_str;
    Algorithm::serialize_write_pod(megdnn_opr->param(), param_str);
    ExeContext ctx(layouts, megdnn_opr, param_str, mgb_opr,
                   mgb_opr->comp_node(), mgb_opr->execution_policy(),
                   allow_weight_preprocess);
399

400
    ImplExecutionPolicy policy;
401
    if (auto algo_choose_hook = mgb_opr->algo_chooser()) {
402
        policy = algo_choose_hook(mgb_opr);
403 404 405 406 407
        ctx.construct_execution_policy(
                mgb_opr->execution_policy().strategy ==
                        mixin::AlgoChooserHelper::ExecutionPolicy::Strategy::
                                HEURISTIC_REPRODUCIBLE,
                policy, false);
408
    }
409 410
    if (!policy.algo.valid()) {
        policy = get_policy(ctx);
411
    }
412
    size_t workspace = ctx.get_workspace_size_bytes(policy);
M
Megvii Engine Team 已提交
413 414 415

    std::string ret;
    ret.append(mgb_opr->dyn_typeinfo()->name);
416 417 418 419
    ret += format_fixlayouts<Opr>(layouts, arity_in, arity_out);
    Algorithm* palgo = megdnn_opr->get_algorithm_from_desc(policy.algo);
    mgb_assert(palgo, "Unknown algo description");
    ret.append("): algo=" + std::string(palgo->name()));
M
Megvii Engine Team 已提交
420
    ret.append(ssprintf(" workspace=%.2fMiB reproducible=%d",
421
                        workspace / (1024 * 1024.0), palgo->is_reproducible()));
M
Megvii Engine Team 已提交
422 423
    mgb_log_debug("%s", ret.c_str());

424
    megdnn_opr->execution_policy() = policy;
425 426 427 428
    return workspace;
}

template <typename Opr>
429
typename AlgoChooser<Opr>::ImplExecutionPolicy AlgoChooser<Opr>::get_policy(
430
        ExeContext& ctx) {
431
    using S = mixin::AlgoChooserHelper::ExecutionPolicy::Strategy;
432
    MGB_MARK_USED_VAR(TIMEOUT_TOLERANCE);
433
    switch (ctx.execution_policy().strategy) {
434 435 436 437 438
        case S::HEURISTIC:
            return ctx.choose_by_heuristic();
        case S::HEURISTIC_REPRODUCIBLE:
            return ctx.choose_by_heuristic(true);
        case S::PROFILE_HEURISTIC: {
439 440 441 442
            ImplExecutionPolicy policy = choose_by_profile(ctx, false, false);
            if (!policy.algo.valid())
                policy = ctx.choose_by_heuristic();
            return policy;
443 444 445 446 447 448 449 450 451 452 453 454 455
        }
#if MGB_ENABLE_FASTRUN
        case S::PROFILE:
            return choose_by_profile(ctx, false);
        case S::PROFILE_REPRODUCIBLE:
            return choose_by_profile(ctx, true);
#endif
        default:
            mgb_throw(GraphError, "bad convolution ExecutionPolicy strategy");
    }
}

#define INST(Opr)                                                            \
456 457 458 459 460
    template AlgoChooser<megdnn::Opr>::ImplExecutionPolicy                   \
    AlgoChooser<megdnn::Opr>::get_policy(ExeContext& ctx);                   \
    template void AlgoChooser<megdnn::Opr>::profile(                         \
            ExeContext& ctx, bool require_reproducible);                     \
    template AlgoChooser<megdnn::Opr>::ImplExecutionPolicy                   \
461 462 463
    AlgoChooser<megdnn::Opr>::choose_by_profile(                             \
            ExeContext& ctx, bool require_reproducible, bool enable_update); \
    template size_t AlgoChooser<megdnn::Opr>::setup_algo(                    \
464
            const FixedTensorLayouts& layouts, megdnn::Opr* megdnn_opr,      \
465
            const MGBOpr* mgb_opr, bool allow_weight_preprocess);            \
466 467 468 469 470 471

MGB_FOREACH_FASTRUN_OPR(INST)

#undef INST

//////////////////////////////// ExeContext /////////////////////////////
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
template <typename Opr>
AlgoChooser<Opr>::ExeContext::ExeContext(
        const FixedTensorLayouts& layouts, Opr* megdnn_opr,
        const std::string& param_str, const cg::OperatorNodeBase* mgb_opr,
        const CompNode& cn,
        const megdnn::param::ExecutionPolicy& execution_policy,
        bool allow_weight_preprocess)
        : m_layouts{layouts},
          m_megdnn_opr{megdnn_opr},
          m_param{param_str},
          m_base_mgb_opr{mgb_opr},
          m_cn{cn},
          m_execution_policy{execution_policy},
          m_allow_weight_preprocess{allow_weight_preprocess} {
    mgb_assert(m_layouts.size() == layouts.size());
    static_assert(std::tuple_size<FixedTensorLayouts>::value == 3 ||
                          std::tuple_size<FixedTensorLayouts>::value == 5 ||
                          std::tuple_size<FixedTensorLayouts>::value == 8,
                  "Convolution AlgoChooser assumes arity = 3 , 5 or 8 (for "
                  "deformable conv)");
}
493 494 495

template <typename Opr>
typename AlgoChooser<Opr>::ImplAlgo
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
AlgoChooser<Opr>::ExeContext::get_profile_result_from_cache(
        bool require_reproducible) const {
    MIDOUT_B(Opr,
             midout_iv(MGB_HASH_STR(
                     "AlgoChooser::ExeContext::get_profile_result_from_cache")))
    AlgoChooserProfileCache cache(m_cn,
                                  profile_name(m_megdnn_opr).c_str());

    typename Opr::Param origin_param = m_megdnn_opr->param();
    AlgoChooserProfileCache::Key cache_key{m_layouts.data(), m_layouts.size(),
                                           &origin_param, sizeof(origin_param)};
    auto&& rst = cache.get(cache_key);
    if (!rst.valid())
        return {};

    auto&& prof = rst.val();
    std::unordered_map<std::string, ImplAlgo> algo_map;
    for (auto i : get_all_candidates()) {
        auto ins = algo_map.emplace(i.name.c_str(), i);
        mgb_assert(ins.second, "duplicated algo name: %s", i.name.c_str());
    }

    if (prof.empty())
        return {};
    for (auto&& i : prof) {
        if ((!require_reproducible || i.reproducible)) {
            auto iter = algo_map.find(i.algo);
            mgb_assert(iter != algo_map.end(),
                       "algorithm %s exists in "
                       "profiling result but not in algo_map; please "
                       "report this "
                       "bug; opr: %s{%s}, layouts: %s ",
                       i.algo.c_str(), m_base_mgb_opr->cname(),
                       m_base_mgb_opr->dyn_typeinfo()->name,
                       format_fixlayouts<Opr>(m_layouts, arity_in, arity_out)
                               .c_str());
            return iter->second;
        }
    }

    mgb_log_error(
            "Workspace requirement (%zu) could not be satisfied. Abort now "
            "to "
            "avoid further problems",
            WorkspaceLimitGetter::get_workspace_limit(
                    m_base_mgb_opr->owner_graph(), m_cn,
                    m_execution_policy.workspace_limit));
    mgb_trap();
    MIDOUT_E
}

template <typename Opr>
typename AlgoChooser<Opr>::ImplExecutionPolicy
549
AlgoChooser<Opr>::ExeContext::choose_by_heuristic(bool reproducible) const {
550 551 552 553 554 555 556
    if (m_execution_policy.workspace_limit !=
        std::numeric_limits<decltype(
                m_execution_policy.workspace_limit)>::max()) {
        mgb_log_warn(
                "workspace_limit should not be setted if choose algo by "
                "heuristic");
    }
557
    auto workspace_limit = WorkspaceLimitGetter::get_workspace_limit(
558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
            owner_graph(), m_cn, m_execution_policy.workspace_limit);
    ImplExecutionPolicy policy;
    policy.algo = APPLY(m_megdnn_opr->get_algorithm_info_heuristic(
                                args..., workspace_limit, reproducible),
                        m_layouts).desc;

    Algorithm* algo = m_megdnn_opr->get_algorithm_from_desc(policy.algo);
    mgb_assert(algo, "Unknown algo description");
    std::vector<Algorithm::SearchItem>&& sub_items = algo->get_subopr_list(
            to_layout_array<Opr>(m_layouts), m_megdnn_opr);

    FOREACH_OPR_TYPE_DISPATCH(sub_items, {
        auto&& megdnn_opr = intl::create_megdnn_opr<_Opr>(m_cn);
        megdnn_opr->param() =
                Algorithm::deserialize_read_pod<typename _Opr::Param>(
                        _item.param);
        typename AlgoChooser<_Opr>::ExeContext sub_ctx(
                to_fixed_layouts<_Opr>(_item.layouts), megdnn_opr.get(),
                _item.param, m_base_mgb_opr, m_cn, m_execution_policy,
                m_allow_weight_preprocess);
        policy.sub_policy.push_back(sub_ctx.choose_by_heuristic(reproducible));
    });

    return policy;
582 583 584 585 586 587
}

template <typename Opr>
std::vector<typename AlgoChooser<Opr>::ImplAlgo>
AlgoChooser<Opr>::ExeContext::get_all_candidates() const {
    auto heu = choose_by_heuristic();
588 589
    auto&& ret =
            APPLY(m_megdnn_opr->get_all_algorithms_info(args...), m_layouts);
590 591
    bool found = false;
    for (size_t i = 0; i < ret.size(); ++i) {
592
        if (ret[i].desc == heu.algo) {
593 594 595 596 597
            found = true;
            std::swap(ret[i], ret[0]);
            break;
        }
    }
598 599 600

    Algorithm* palgo = m_megdnn_opr->get_algorithm_from_desc(heu.algo);
    mgb_assert(palgo, "Unknown algo description");
601 602
    mgb_assert(found,
               "algo %s got by heuristic not found in "
603
               "candidate list",
604
               palgo->name());
605 606 607 608
    return std::move(ret);
}

template <typename Opr>
609
void AlgoChooser<Opr>::ExeContext::construct_execution_policy(
610
        bool require_reproducible,
611 612
        typename AlgoChooser<Opr>::ImplExecutionPolicy& policy,
        bool retrive_from_cache) const {
613
    if (!policy.algo.valid()) {
614 615 616 617 618 619 620 621 622 623 624 625
        if (retrive_from_cache) {
            policy.algo =
                    get_profile_result_from_cache(require_reproducible).desc;
        } else {
            auto workspace_limit = WorkspaceLimitGetter::get_workspace_limit(
                    owner_graph(), m_cn, m_execution_policy.workspace_limit);
            policy.algo = APPLY(m_megdnn_opr->get_algorithm_info_heuristic(
                                        args..., workspace_limit,
                                        require_reproducible),
                                m_layouts)
                                  .desc;
        }
626
        mgb_assert(policy.algo.valid(),
627 628
                   "No algo found from cache or heuristic, maybe some error "
                   "occured");
629
    }
630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645

    Algorithm* algo = m_megdnn_opr->get_algorithm_from_desc(policy.algo);
    mgb_assert(algo, "Unknown algo description");
    std::vector<Algorithm::SearchItem>&& sub_items = algo->get_subopr_list(
            to_layout_array<Opr>(m_layouts), m_megdnn_opr);

    FOREACH_OPR_TYPE_DISPATCH(sub_items, {
        auto&& megdnn_opr = intl::create_megdnn_opr<_Opr>(m_cn);
        megdnn_opr->param() =
                Algorithm::deserialize_read_pod<typename _Opr::Param>(
                        _item.param);
        typename AlgoChooser<_Opr>::ExeContext sub_ctx(
                to_fixed_layouts<_Opr>(_item.layouts), megdnn_opr.get(),
                _item.param, m_base_mgb_opr, m_cn, m_execution_policy,
                m_allow_weight_preprocess);
        policy.sub_policy.push_back({});
646 647 648
        sub_ctx.construct_execution_policy(require_reproducible,
                                           policy.sub_policy.back(),
                                           retrive_from_cache);
649 650 651
    });

    return;
652 653 654 655
}

template <typename Opr>
size_t AlgoChooser<Opr>::ExeContext::get_workspace_size_bytes(
656 657
        const ImplExecutionPolicy& policy) const {
    m_megdnn_opr->execution_policy() = policy;
658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680
    size_t result;
    if_constexpr<opr_supports_preprocess<Opr>()>(
            [&](auto _) {
                auto&& opr = _(m_megdnn_opr);
                auto prep = this->construct_fake_preprocess_filter();
                PreprocessFilter<Opr>* prep_ptr =
                        prep.valid() ? &prep.val() : nullptr;
                result = std::max(
                        APPLY(opr->get_preprocess_workspace_in_bytes(args...),
                              m_layouts),
                        APPLY(opr->get_workspace_in_bytes(args..., prep_ptr),
                              m_layouts));
            },
            /* else */
            [&](auto _) {
                result = APPLY(_(m_megdnn_opr)->get_workspace_in_bytes(args...),
                               m_layouts);
            });
    return result;
}

template <typename Opr>
Maybe<AlgoChooserProfileCache::ResultEntry>
681 682
AlgoChooser<Opr>::ExeContext::profile_single_algo(
        const ImplExecutionPolicy& policy, double& timeout) const {
683 684
    typename TimedProfiler<Opr>::Param param;
    // force check copy size <= dest len-1 from gcc8 for safe
685 686 687
    param.execution_policy =
            TimedProfiler<Opr>::Param::ExecutionPolicyBlob::serialize(policy);
    param.workspace = get_workspace_size_bytes(policy);
688 689 690 691 692 693 694 695 696 697
    for (int i = 0; i < arity; ++i) {
        auto&& src = m_layouts[i];
        mgb_assert(src.format.is_default() &&
                           (src.dtype.category() == DTypeCategory::FLOAT ||
                            src.dtype.category() == DTypeCategory::INT ||
                            src.dtype.category() == DTypeCategory::QUANTIZED),
                   "unsupported layout in profiling: %s",
                   src.to_string().c_str());
        param.dtypes[i] = src.dtype.enumv();
    }
698
    param.comp_node_loc = m_cn.locator();
699 700 701 702 703 704
    mgb_assert(param.shapes.size() == m_layouts.size());
    for (size_t i = 0; i < param.shapes.size(); ++i)
        param.shapes[i] = m_layouts[i];
    param.opr_param = m_megdnn_opr->param();
    param.allow_weight_preprocess = m_allow_weight_preprocess;

705 706
    Algorithm* palgo = m_megdnn_opr->get_algorithm_from_desc(policy.algo);
    mgb_assert(palgo, "Unknown algo description");
707 708 709 710
    auto rst = TimedProfiler<Opr>::profile(param, timeout);
    // MIOpen conv profiles all available algos when a specfic shape is
    // provided for the first time, which probably adds to the result time.
    // Therefore, a second profile execution is needed.
711
    if (strncmp(palgo->name(), "MIOpen", 6) == 0)
712 713 714 715
        rst = TimedProfiler<Opr>::profile(param, timeout);
    if (!rst.valid())
        return None;
    return AlgoChooserProfileCache::ResultEntry{
716
            palgo->name(), palgo->is_reproducible(), rst.val().time,
717 718 719 720 721 722 723 724 725 726 727
            param.workspace};
}

template <typename Opr>
Maybe<PreprocessFilter<Opr>>
AlgoChooser<Opr>::ExeContext::construct_fake_preprocess_filter() const {
    Maybe<PreprocessFilter<Opr>> result = None;
    if_constexpr<opr_supports_preprocess<Opr>()>([&](auto _) {
        if (!m_allow_weight_preprocess)
            return;
        auto opr = _(m_megdnn_opr);
728 729 730 731
        auto layouts = APPLY(opr->deduce_preprocessed_filter_layout(args...),
                             m_layouts);
        //! No preprocess layout means no need weight preprocess
        if (layouts.empty()) {
732
            return;
733 734 735 736 737 738 739 740 741 742 743 744
        }
        //! all layouts arm empty means no need weight preprocess
        bool layout_valid = false;
        for (auto&& layout : layouts) {
            if (!layout.is_empty()) {
                layout_valid = true;
            }
        }
        if (!layout_valid) {
            return;
        }

745 746 747
        result = PreprocessFilter<Opr>{};
        auto& res = result.val();
        res.algorithm_id = nullptr;
748 749 750
        res.tensors.resize(layouts.size());
        for (size_t i = 0; i < layouts.size(); i++) {
            res.tensors[i] = megdnn::TensorND(nullptr, layouts[i]);
751 752 753 754 755 756
        }
    });
    return result;
}

#define INST(Opr)                                                              \
757 758 759 760 761 762 763
    template AlgoChooser<megdnn::Opr>::ExeContext::ExeContext(                 \
            const FixedTensorLayouts& layouts, megdnn::Opr* megdnn_opr,        \
            const std::string& param_str, const cg::OperatorNodeBase* mgb_opr, \
            const CompNode& cn,                                                \
            const megdnn::param::ExecutionPolicy& execution_policy,            \
            bool allow_weight_preprocess);                                     \
    template typename AlgoChooser<megdnn::Opr>::ImplExecutionPolicy            \
764 765
    AlgoChooser<megdnn::Opr>::ExeContext::choose_by_heuristic(                 \
            bool reproducible) const;                                          \
766 767 768
    template typename AlgoChooser<megdnn::Opr>::ImplAlgo                       \
    AlgoChooser<megdnn::Opr>::ExeContext::get_profile_result_from_cache(       \
            bool require_reproducible) const;                                  \
769 770 771 772
    template std::vector<typename AlgoChooser<megdnn::Opr>::ImplAlgo>          \
    AlgoChooser<megdnn::Opr>::ExeContext::get_all_candidates() const;          \
    template size_t                                                            \
    AlgoChooser<megdnn::Opr>::ExeContext::get_workspace_size_bytes(            \
773 774
            const typename AlgoChooser<megdnn::Opr>::ImplExecutionPolicy&      \
                    policy) const;                                             \
775 776 777 778 779
    template void                                                              \
    AlgoChooser<megdnn::Opr>::ExeContext::construct_execution_policy(          \
            bool require_reproducible,                                         \
            typename AlgoChooser<megdnn::Opr>::ImplExecutionPolicy& policy,    \
            bool retrive_from_cache) const;                                    \
780 781
    template Maybe<AlgoChooserProfileCache::ResultEntry>                       \
    AlgoChooser<megdnn::Opr>::ExeContext::profile_single_algo(                 \
782 783 784
            const typename AlgoChooser<megdnn::Opr>::ImplExecutionPolicy&      \
                    policy,                                                    \
            double& timeout) const;
785 786 787 788 789 790 791 792

MGB_FOREACH_FASTRUN_OPR(INST)

#undef INST
}  // namespace opr
}  // namespace mgb

// vim: syntax=cpp.doxygen foldmethod=marker foldmarker=f{{{,f}}}