feat(lite): add disable configure by model info interface

GitOrigin-RevId: cd155a1fcf8bf6b845fa9118a6a791d662b2b624

lite/include/lite/network.h

@@ -117,6 +117,17 @@ struct LITE_API Config {
     Options options = {};
 };
 
+/*!
+ * \brief Extra Configuration for a network
+ *
+ * \param disable_configure_by_model_info disable the configuration dumped with model,
+ * if set true, all configuration in the model will not apply, users should configure
+ * the network.
+ */
+struct LITE_API ExtraConfig {
+    bool disable_configure_by_model_info = false;
+};
+
 /*!
  * \brief config the network input and output item
  *
@@ -275,6 +286,12 @@ public:
     //! get static peak memory info showed by Graph visualization
     void get_static_memory_alloc_info(const std::string& log_dir = "logs/test") const;
 
+    /** @brief the extra configuration
+     *
+     * @param extra_config the extra configuration to set into the network
+     */
+    void extra_configure(const ExtraConfig& extra_config);
+
 public:
     friend class NetworkHelper;
 
@@ -288,6 +305,7 @@ private:
 private:
     bool m_loaded = false;
     Config m_config;
+    ExtraConfig m_extra_config;
     NetworkIO m_network_io;
     std::unique_ptr<NetworkImplBase> m_impl;
     std::string m_extra_info;

lite/lite-c/include/lite-c/network_c.h

@@ -113,6 +113,17 @@ typedef struct LiteConfig {
 //! get default config
 LITE_API LiteConfig* default_config();
 
+/*!
+ * \brief Exetra Configuration for a network
+ *
+ * \param disable_configure_by_model_info disable the configuration dumped with model,
+ * if set true, all configuration in the model will not apply, users should configure
+ * the network.
+ */
+typedef struct LiteExtraConfig {
+    int disable_configure_by_model_info;
+} LiteExtraConfig;
+
 /*!
  * \brief config the network input and output item
  *
@@ -599,6 +610,12 @@ LITE_API int LITE_get_model_io_info_by_memory(
         const void* model_mem, size_t size, const LiteConfig config,
         LiteNetworkIO* ios);
 
+/** @brief the extra configuration
+ *
+ * @param extra_config the extra configuration to set into the network
+ */
+LITE_API int LITE_extra_configure(LiteNetwork network, LiteExtraConfig extra_config);
+
 #ifdef __cplusplus
 }
 #endif

lite/lite-c/src/network.cpp

@@ -181,6 +181,12 @@ InnerIO convert_to_inner_io(const lite::NetworkIO& network_io) {
     return innner_io;
 }
 
+lite::ExtraConfig convert_extra_config(const LiteExtraConfig& extra_config) {
+    lite::ExtraConfig ret;
+    ret.disable_configure_by_model_info = extra_config.disable_configure_by_model_info;
+    return ret;
+}
+
 int LITE_make_default_network(LiteNetwork* network) {
     LITE_CAPI_BEGIN();
     LITE_ASSERT(network, "The network pass to LITE api is null");
@@ -734,4 +740,12 @@ int LITE_get_model_io_info_by_memory(
     LITE_CAPI_END();
 }
 
+LITE_API int LITE_extra_configure(LiteNetwork network, LiteExtraConfig extra_config) {
+    LITE_CAPI_BEGIN();
+    LITE_ASSERT(network, "The network pass to LITE api is null");
+    static_cast<lite::Network*>(network)->extra_configure(
+            convert_extra_config(extra_config));
+    LITE_CAPI_END();
+}
+
 // vim: syntax=cpp.doxygen foldmethod=marker foldmarker=f{{{,f}}}

lite/pylite/megenginelite/network.py

@@ -134,6 +134,31 @@ class LiteConfig(Structure):
         return data.__repr__()
 
 
+class LiteExtraConfig(Structure):
+    """
+    Extra configuration when load and compile the graph
+
+    disable_configure_by_model_info: disable the configuration dumped with
+    model, if set true, all configuration in the model will not apply, users
+    should configure the network.
+    """
+
+    _fields_ = [
+        ("disable_configure_by_model_info", c_int),
+    ]
+
+    def __init__(self, disable_model_config=False):
+        self.disable_configure_by_model_info = disable_model_config
+
+    def __repr__(self):
+        data = {
+            "disable_configure_by_model_info": bool(
+                self.disable_configure_by_model_info
+            ),
+        }
+        return data.__repr__()
+
+
 class LiteIO(Structure):
     """
     config the network input and output item
@@ -365,6 +390,7 @@ class _NetworkAPI(_LiteCObjBase):
             "LITE_get_model_io_info_by_memory",
             [c_char_p, c_size_t, LiteConfig, POINTER(_LiteNetworkIO)],
         ),
+        ("LITE_extra_configure", [_Cnetwork, LiteExtraConfig]),
     ]
 
 
@@ -541,6 +567,12 @@ class LiteNetwork(object):
             ret_name = [names[i].decode("utf-8") for i in range(nr_output.value)]
             return ret_name
 
+    def extra_configure(self, extra_config):
+        """
+        Extra Configuration to the network.
+        """
+        self._api.LITE_extra_configure(self._network, extra_config)
+
     def share_weights_with(self, src_network):
         """
         share weights with the loaded network

lite/pylite/test/test_network.py

@@ -112,6 +112,13 @@ class TestNetwork(TestShuffleNet):
         network.load(model_path)
         self.do_forward(network)
 
+    def test_disable_model_config(self):
+        model_path = os.path.join(self.source_dir, "test_packed_model_rc4.lite")
+        network = LiteNetwork()
+        network.extra_configure(LiteExtraConfig(True))
+        network.load(model_path)
+        self.do_forward(network)
+
     def test_pack_cache_to_model(self):
         model_path = os.path.join(self.source_dir, "test_pack_cache_to_model.lite")
         network = LiteNetwork()

lite/src/mge/network_impl.cpp

@@ -31,7 +31,6 @@ using namespace mgb;
 LITE_DYN_TYPE_OBJ_FINAL_IMPL(NetworkImplDft);
 
 void NetworkImplDft::set_config(const Config& config) {
-    m_user_config = std::make_unique<Config>();
     *m_user_config = config;
     m_compnode_locator = to_compnode_locator(m_user_config->device_type);
     m_compnode_locator.device = config.device_id;
@@ -428,8 +427,11 @@ void NetworkImplDft::load_model(
 
     global_layout_transform();
 
+    //! some optimization option maybe invalid in some case, so here just
+    //! auto determine whether some options will apply.
     adapt_option_valid();
 
+    //! find how many compnode the model has, this should call before update_io
     cross_compnode_model_detect();
 
     //! update the IO of the network
@@ -496,7 +498,6 @@ void NetworkImplDft::finish() const {
 }
 
 void NetworkImplDft::set_io(const NetworkIO& network_io) {
-    m_network_io = std::make_unique<NetworkIOInner>();
     for (auto&& in : network_io.inputs) {
         m_network_io->inputs.emplace_back(in);
     }

lite/src/mge/network_impl.h

@@ -29,7 +29,11 @@ class NetworkImplDft final : public Network::NetworkImplBase {
     LITE_DYN_TYPE_OBJ_FINAL_DECL;
 
 public:
-    NetworkImplDft() { m_load_config.comp_graph = mgb::ComputingGraph::make(); }
+    NetworkImplDft() {
+        m_load_config.comp_graph = mgb::ComputingGraph::make();
+        m_user_config = std::make_unique<Config>();
+        m_network_io = std::make_unique<NetworkIOInner>();
+    }
     using S = megdnn::param::ExecutionPolicy::Strategy;
     using Var = mgb::cg::SymbolVar;
     //! set the config of the network, include:

lite/src/network.cpp

@@ -80,15 +80,18 @@ void Network::prase_model(std::shared_ptr<void> model_data, size_t size) {
     ModelParser model_parser(model_data, size);
     //! parse the model info
     if (model_parser.parse_model_info(
-                m_config, m_network_io, separate_config_map, m_extra_info)) {
+                m_config, m_network_io, separate_config_map, m_extra_info,
+                !m_extra_config.disable_configure_by_model_info)) {
         if (m_config.backend == LiteBackend::LITE_DEFAULT &&
             m_impl->get_backend_type() != LiteBackend::LITE_DEFAULT) {
             m_impl.reset(try_call_func<NetworkImplDft, lite::Network::NetworkImplBase*>(
                     "parse_model"));
         }
+        if (!m_extra_config.disable_configure_by_model_info) {
             m_impl->set_config(m_config);
             m_impl->set_io(m_network_io);
         }
+    }
     //! decryption the model
     size_t model_length;
     auto&& model_shared_ptr = model_parser.parse_model(model_length, m_config);
@@ -290,6 +293,18 @@ void Network::get_static_memory_alloc_info(const std::string& log_dir) const {
     LITE_ERROR_HANDLER_END
 }
 
+void Network::extra_configure(const ExtraConfig& extra_config) {
+    LITE_ERROR_HANDLER_BEGIN
+    if (!extra_config.disable_configure_by_model_info) {
+        LITE_ASSERT(
+                !m_loaded,
+                "disable_configure_by_model_info should be configured before model "
+                "loaded.");
+    }
+    m_extra_config = extra_config;
+    LITE_ERROR_HANDLER_END
+}
+
 /*********************** MGE special network function ***************/
 
 void Runtime::set_cpu_threads_number(

lite/src/parse_model/model_parser.cpp

@@ -43,7 +43,7 @@ void ModelParser::parse_header() {
 bool ModelParser::parse_model_info(
         Config& network_config, NetworkIO& network_io,
         std::unordered_map<std::string, LiteAny>& isolated_config_map,
-        std::string& extra_info) const {
+        std::string& extra_info, bool configure_valid) const {
     //! no model info, no parse, direct return
     if (m_is_bare_model || !m_info) {
         return false;
@@ -78,7 +78,7 @@ bool ModelParser::parse_model_info(
         }
     }
     //! parse ModelInfo::algo_policy
-    if (m_info->algo_policy()) {
+    if (m_info->algo_policy() && configure_valid) {
         size_t cache_length = m_info->algo_policy()->size();
         const uint8_t* cache = m_info->algo_policy()->Data();
         if (m_info_cache_parse_func_name == "LITE_parse_cache") {
@@ -93,6 +93,10 @@ bool ModelParser::parse_model_info(
             } else {
                 LITE_THROW("opencl binary cache is not given");
             }
+        } else {
+            LITE_THROW(ssprintf(
+                    "model cache parse function of  %s is not defined.",
+                    m_info_cache_parse_func_name.c_str()));
         }
     }
     return true;

lite/src/parse_model/model_parser.h

@@ -25,7 +25,7 @@ public:
     bool parse_model_info(
             Config& network_config, NetworkIO& network_io,
             std::unordered_map<std::string, LiteAny>& isolated_config_map,
-            std::string& extra_info) const;
+            std::string& extra_info, bool configure_valid) const;
 
     //! parse the model and decrypt the model
     std::shared_ptr<void> parse_model(size_t& model_length, const Config& config) const;

lite/test/test_network_options.cpp

@@ -7,6 +7,8 @@
 #include "lite/global.h"
 
 #include "megbrain/tensor.h"
+#include "megbrain/utils/infile_persistent_cache.h"
+#include "megbrain/utils/persistent_cache.h"
 #include "test_common.h"
 
 #include <string.h>
@@ -173,6 +175,29 @@ TEST(TestNetWorkOptions, test_cache) {
     compare_lite_tensor<float>(output_tensor, result_mgb);
 }
 
+TEST(TestNetWorkOptions, DisableModelInfo) {
+    //! clear the cache set by other test
+    mgb::PersistentCache::inst().set_impl(
+            std::make_shared<mgb::InMemoryPersistentCache>());
+    Config config;
+    auto tensor = get_input_data("./input_data.npy");
+    std::string model_path = "./test_pack_cache_to_model.lite";
+    std::string model_path2 = "./test_pack_cache_to_model.lite";
+    std::string input_name = "data";
+    std::shared_ptr<Network> network = std::make_shared<Network>(config);
+    network->extra_configure({true});
+    Runtime::set_cpu_inplace_mode(network);
+    network->load_model(model_path);
+    //! the fast-run cache will not configure, so it is not support dump
+    ASSERT_EQ(mgb::PersistentCache::inst().support_dump_cache(), false);
+    ASSERT_EQ(Runtime::is_cpu_inplace_mode(network), true);
+
+    std::shared_ptr<Network> network2 = std::make_shared<Network>(config);
+    network2->load_model(model_path2);
+    //! the fast-run cache is configured by the model information
+    ASSERT_EQ(mgb::PersistentCache::inst().support_dump_cache(), true);
+}
+
 TEST(TestNetWorkOptions, FastRunIgnorBatch) {
     Config config;
     auto tensor = get_input_data("./input_data.npy");