Fixed #10. Supported GOTO_NUM_THREADS & GOTO_THREADS_TIMEOUT environment variables.

Author: xianyi

Makefile.rule

@@ -70,7 +70,7 @@ VERSION = 0.1
 # time out to improve performance. This number should be from 4 to 30
 # which corresponds to (1 << n) cycles. For example, if you set to 26,
 # thread will be running for (1 << 26) cycles(about 25ms on 3.0GHz
-# system). Also you can control this mumber by GOTO_THREAD_TIMEOUT
+# system). Also you can control this mumber by THREAD_TIMEOUT
 # CCOMMON_OPT	+= -DTHREAD_TIMEOUT=26
 
 # Using special device driver for mapping physically contigous memory

README

@@ -32,11 +32,19 @@ MIPS64:
 4.Usages
 Link with libopenblas.a or -lopenblas for shared library.
 
-Set the number of threads. for example,
+4.1 Set the number of threads with environment variables. for example,
 export OPENBLAS_NUM_THREADS=4
+ or
+export GOTO_NUM_THREADS=4
  or 
 export OMP_NUM_THREADS=4
-OPENBLAS_NUM_THREAD is prior to OMP_NUM_THREADS.
+
+The priorities are OPENBLAS_NUM_THREAD > GOTO_NUM_THREADS > OMP_NUM_THREADS.
+
+4.2 Set the number of threads with calling functions. for example,
+void goto_set_num_threads(int num_threads);
+or
+void openblas_set_num_threads(int num_threads);
 
 5.Report Bugs
 Please add a issue in https://github.com/xianyi/OpenBLAS/issues

driver/others/blas_server.c

@@ -525,7 +525,16 @@ int blas_thread_init(void){
       if (thread_timeout <  4) thread_timeout =  4;
       if (thread_timeout > 30) thread_timeout = 30;
       thread_timeout = (1 << thread_timeout);
-    }
+    }else{
+		p = getenv("GOTO_THREAD_TIMEOUT");
+		if (p) {
+			thread_timeout = atoi(p);
+			if (thread_timeout <  4) thread_timeout =  4;
+			if (thread_timeout > 30) thread_timeout = 30;
+			thread_timeout = (1 << thread_timeout);
+		}
+	}
+	
 
     for(i = 0; i < blas_num_threads - 1; i++){
 
@@ -790,6 +799,11 @@ void goto_set_num_threads(int num_threads) {
 
 }
 
+void openblas_set_num_threads(int num_threads) {
+	goto_set_num_threads(num_threads);
+	
+}
+
 /* Compatible function with pthread_create / join */
 
 int gotoblas_pthread(int numthreads, void *function, void *args, int stride) {

driver/others/init.c

@@ -581,6 +581,7 @@ void gotoblas_affinity_init(void) {
   numprocs = 0;
 #else
   numprocs = readenv("OPENBLAS_NUM_THREADS");
+  if (numprocs == 0) numprocs = readenv("GOTO_NUM_THREADS");
 #endif
 
   if (numprocs == 0) numprocs = readenv("OMP_NUM_THREADS");
@@ -666,7 +667,7 @@ void gotoblas_affinity_init(void) {
 
     setup_mempolicy();
 
-    if (readenv("OPENBLAS_MAIN_FREE")) {
+    if (readenv("OPENBLAS_MAIN_FREE") || readenv("GOTOBLAS_MAIN_FREE")) {
       sched_setaffinity(0, sizeof(cpu_orig_mask), &cpu_orig_mask[0]);
     }
 

driver/others/memory.c

@@ -231,6 +231,13 @@ int blas_get_cpu_number(void){
   p = getenv("OPENBLAS_NUM_THREADS");
   if (p) blas_goto_num = atoi(p);
   if (blas_goto_num < 0) blas_goto_num = 0;
+
+  if (blas_goto_num == 0) {
+	    p = getenv("GOTO_NUM_THREADS");
+		if (p) blas_goto_num = atoi(p);
+		if (blas_goto_num < 0) blas_goto_num = 0;
+  }
+  
 #endif
 
   blas_omp_num = 0;