CN103455356A - Concurrence loading and rendering method of 3D (three-dimensional) models on multi-core mobile device - Google Patents

Abstract

The invention relates to a concurrence loading and rendering method of 3D (three-dimensional) models on a multi-core mobile device, comprising the following steps of taking a CPU (Central Processing Unit) core number of the current device as the capacity of a thread pool to create the thread pool; respectively creating 3D model data loading tasks for each 3D model file stored on a secondary storage device; respectively creating a rendering task for each 3D model; submitting all data loading tasks to a buffering queue of the thread pool, and carrying out concurrent execution on the data loading tasks by the thread pool; after the data loading tasks are all executed, submitting the rendering tasks into the buffering queue of the thread pool and carrying out concurrent execution on the rendering tasks by the thread pool; finally, after all rendering tasks are completed, destroying the thread pool and releasing system resources.

Description

Concurrent loading and the rendering intent of 3D model on the multinuclear mobile device

Technical field

The invention belongs to field of embedded software.Relate in particular to a kind of concurrent loading of 3D model and rendering intent on the multinuclear mobile device.

Background technology

Along with the development of embedded technology, with the mobile device headed by smart mobile phone, panel computer, obtained swift and violent development.The central processing unit of mobile device from before core go so far as present double-core, four cores or even eight cores.In the 3D application facet, the multinuclear mobile device not only can be showed complicated 3D figure, even can move large-scale 3D game.

3D application for complex scene, need to load and play up a plurality of 3D models.Existing method is that first by all 3D model files, the mode with serial is loaded in internal memory, then in the mode of serial, plays up one by one.Yet, in the multinuclear mobile device, the advantage that the 3D disposal route of this serial can not be brought into play polycaryon processor improves the 3D processing speed.

Summary of the invention

The technical matters that will solve required for the present invention is to provide a kind of concurrent loading of 3D model and rendering intent on the multinuclear mobile device.

The technical scheme that technical solution problem of the present invention is taked is as follows:

First judge the core cpu number of current mobile device before loading and playing up the 3D model.The core cpu number of current device of usining creates thread pool as the capacity of thread pool.Then be respectively each the 3D model file be stored on external equipment and create respectively 3D model data loading tasks.The data loading tasks is used for loading the vertex data, normal vector data, texture coordinate data, face data, material quality data of 3D model file to internal memory.Create respectively rendering task for each 3D model again.Finally more all data loading tasks are submitted in the buffer queue of thread pool, by the concurrent executing data loading tasks of thread pool.After all completing etc. the data loading tasks, rendering task is submitted in the buffer queue of thread pool, by the concurrent execution rendering task of thread pool.After finally waiting all rendering tasks to be finished, destroy the thread pool free system resources.

The present invention has following beneficial effect with respect to prior art: the present invention creates thread pool according to the central processing unit core number of mobile device.Thread pool comes the asynchronous 3D model file of concurrent execution data loading tasks, asynchronous 3D model rendering task with the several worker threads of central processing unit core.Whole process has been given full play to the handling property of multinuclear mobile device.And each task is carried out with asynchronous system, has eliminated the synchronization overhead between each task.Therefore can greatly improve the speed that the multinuclear mobile device loaded and played up the 3D model.

The accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.

As shown in Figure 1, on the multinuclear mobile device, concurrent loading and the rendering intent of 3D model comprises the steps:

Step 1: obtain the core cpu number of current mobile device, specifically:

The operating system of using according to mobile device, the API directly provided by operating system or obtain the core cpu number of current mobile device by the number of files of depositing CPU information in the judgement file system.Operating system includes but not limited to IOS system and Android system.

For the mobile device of IOS operating system, can use " sysctlbyname " function to obtain the core number of CPU;

For the mobile device of Android operating system, obtain the core number of CPU by the number of files that records CPU information under judgement "/sys/devices/system/cpu/ " catalogue.The number of files that records CPU information is the core number of CPU;

Step 2: the thread pool that the establishment capacity is the core cpu number, specifically:

Create the thread pool number variable nThread of an integer, the core cpu number that will be obtained by the first step is assigned to variable nThread;

Thread pool class in the application development kit of the operating system of using mobile device to use creates the thread pool that capacity is nThread, and operating system includes but not limited to IOS system and Android system;

For the mobile device of IOS operating system, use " NSOperation " in IOS application development kit to create thread pool, use

The capacity that " queue setMaxConcurrentOperationCount:nThread " statement arranges thread pool is nThread;

For the mobile device of Android operating system, can use " Executors " class in Android application development kit to help create thread pool.Concrete use " Executors.newFixedThreadPool (nThread) " carrys out the thread pool that the establishment capacity is nThread;

Step 3: for each 3D model file creates respectively the asynchronous data loading tasks, specifically:

First traversal is stored in the 3D model file on external equipment, obtains the path of depositing of each model file, and is placed in array;

Create the asynchronous data loading tasks, the specific tasks flow process is again:

Flow process one: according to the file path that is stored in the 3D model file in the array I/O that opens file;

Described 3D model file form comprises OBJ form and 3DS form;

Flow process two: extract the data that are stored in the 3D model file on external equipment, comprising: vertex data, normal vector data, texture coordinate data, face data, material quality data;

Flow process three: the 3D model data of extraction is packaged into separately to a data object;

The data loading tasks number altogether created equals the number of 3D model file;

Step 4: for each 3D model creates respectively asynchronous rendering task, the rendering task flow process specifically:

Flow process one: the 3D model data in the 3D model data object that uses step 3 to encapsulate is paintbrush binding vertex data, the normal vector data, and texture coordinate data, face data, and the material characteristic is set;

Flow process two: the gl.glDrawElements function that calls OpenGl ES is played up the 3D model.

The rendering task number altogether created equals the number of 3D model file;

Step 5: all asynchronous data loading tasks by the concurrent execution of thread pool, specifically:

All asynchronous data loading tasks that will be created by step 3 are submitted in the buffer queue of the thread pool created by step 2;

Start thread pool, carry out concomitantly all asynchronous data loading tasks in buffer queue.

Step 6: all asynchronous rendering task by the concurrent execution of thread pool, specifically:

After all being finished etc. the asynchronous data loading tasks of step 5, all rendering tasks are submitted in the buffer queue of the thread pool created by step 2;

Start thread pool, carry out concomitantly all asynchronous rendering tasks in buffer queue;

After all asynchronous rendering tasks are finished, destroy the thread pool free system resources.

Claims (1)

1. concurrent loading and the rendering intent of 3D model on the multinuclear mobile device, its feature comprises the following steps:

Step 1: obtain the core cpu number of current mobile device, specifically:

The operating system of using according to mobile device, the API directly provided by operating system or obtain the core cpu number of current mobile device by the number of files of depositing CPU information in the judgement file system.

Step 2: the thread pool that the establishment capacity is the core cpu number, specifically:

Create the thread pool number variable of integer, the core cpu number that the value of this variable is set to be obtained by step 1;

Using the thread pool number variable of integer as the capacity parameter of thread pool, and the thread pool class in the application development kit of the operating system of using with mobile device is carried out the thread pool that the establishment capacity is the core cpu number.

Step 3: for each 3D model file creates respectively the asynchronous data loading tasks, specifically:

First traversal is stored in the 3D model file on external equipment, obtains the path of depositing of each model file, and is placed in array;

Create the asynchronous data loading tasks, the tool flow of task is again:

Flow process one: according to the file path that is stored in the 3D model file in the array I/O that opens file;

Flow process two: extract the data that are stored in the 3D model file on external equipment, comprising: vertex data, normal vector data, texture coordinate data, face data, material quality data;

Flow process three: the 3D model data of extraction is packaged into separately to a data object.

Step 4: for each 3D model creates respectively asynchronous rendering task, the rendering task flow process specifically:

Flow process one: the 3D model data in the 3D model data object that uses step 3 to encapsulate is paintbrush binding vertex data, the normal vector data, and texture coordinate data, face data, and the material characteristic is set;

Flow process two: the gl.glDrawElements function that calls OpenGl ES is played up the 3D model.

Step 5: all asynchronous data loading tasks by the concurrent execution of thread pool, specifically:

All asynchronous data loading tasks that will be created by step 3 are submitted in the buffer queue of the thread pool created by step 2;

Start thread pool, carry out concomitantly all asynchronous data loading tasks in buffer queue.

Step 6: all asynchronous rendering task by the concurrent execution of thread pool, specifically:

After all being finished etc. the asynchronous data loading tasks of step 5, all rendering tasks are submitted in the buffer queue of the thread pool created by step 2;

Start thread pool, carry out concomitantly all asynchronous rendering tasks in buffer queue;

After all asynchronous rendering tasks are finished, destroy the thread pool free system resources.

Images