Implementation With the User Multithread Model

Sample Directory

/CRIWARE/SDK/pc/samples/criatomex/user_multithread_framework

Sample Description

Perform the following procedure to run the Atom library's server processing in a separate thread created by the user.

• When initializing the library, specify the user multithread model.
• Create a thread, and then call the server processing function at a regular interval on that threads.

The following is a code sample that demonstrates this procedure.

#define USER_THREAD_PRIORITY_AUDIO      (8)
#define USER_THREAD_PRIORITY_FILEACCESS (12)
#define USER_THREAD_PRIORITY_DATADECOMP (24)
 
                :
 
/* Various server threads used in the user multithread model of ADX */
static void app_audio_process(void *arg)
{
    (void)arg;
    /* Audio processing by CRI ADX */
    criAtomEx_ExecuteAudioProcess();
}
 
static void app_file_access_process(void *arg)
{
    (void)arg;
    /* File processing (read-in) by CRI File System */
    /* The load completion callback function is invoked from this thread. */
    criFs_ExecuteFileAccess();
}
 
static void app_data_decomp_process(void *arg)
{
    (void)arg;
    /* Data decompression processing by CRI File System*/
    criFs_ExecuteDataDecompression();
}
                
                :
                
void Sample_Initialize(void)
{
                :
                :
                
    /* Specify user multithread model */
    criAtomEx_SetDefaultConfig(&criatomex_config);
    criatomex_config.thread_model = CRIATOMEX_THREAD_MODEL_USER_MULTI;
    
    /* Library initialization */
    smputl_initialize(&criatomex_config, NULL, 0);
                
                :
                :
                
    /* Create a D-Bas, load the ACF file and register it            */
    /* Load ACB files and create ACB handles            */
    /* Create the voice pool, then create the player and get its status */
                
                :
                :
                
    /* Create a thread for each processing */
    {
        CriUint32 thread_work_size;
        thread_work_size = userThread_CalculateWorkSize();
 
        /* Allocate thread work area */
        user_audio_thread_work          = user_alloc_func(NULL, thread_work_size);
        user_file_access_thread_work    = user_alloc_func(NULL, thread_work_size);
        user_data_decomp_thread_work    = user_alloc_func(NULL, thread_work_size);
 
        /* Run audio processing on the same thread as CRI Atom library initialization */
        user_audio_thread = userThread_Create(
            user_audio_thread_work, thread_work_size,
            app_audio_process, app_obj, USER_THREAD_PRIORITY_AUDIO, 0);
        user_file_access_thread = userThread_Create(
            user_file_access_thread_work, thread_work_size,
            app_file_access_process, app_obj, USER_THREAD_PRIORITY_FILEACCESS, THREAD_AFFINITY_CORE2);
        user_data_decomp_thread = userThread_Create(
            user_data_decomp_thread_work, thread_work_size,
            app_data_decomp_process, app_obj, USER_THREAD_PRIORITY_DATADECOMP, THREAD_AFFINITY_CORE2);
    }
}

Attention

When you use the criAtomEx_ExecuteMain function to execute the server processing, it will wait for file reading within the function and therefore the processing will be locked for a number of V cycles.
If you use the user multithread model, run only the criAtomEx_ExecuteAudioProcess function on the main thread. Then, run other functions such as criFs_ExecuteFileAccess or criFs_ExecuteDataDecompression on a separate thread.

<b