diff --git a/src/fft.c b/src/fft.c
index c9eeb96..5a6427c 100755
--- a/src/fft.c
+++ b/src/fft.c
@@ -105,7 +105,7 @@
 	int i, j, step, k=0;
 	unsigned int avg;
 
-	step = inLen/(max);
+	step = inLen/(max)-2;
 	for(i=0; i<inLen; i=i+step){
 		avg = 0;
 		for(j=0; j<step; j++){
diff --git a/src/music_visualizer.c b/src/music_visualizer.c
index b2f875c..c99bfef 100755
--- a/src/music_visualizer.c
+++ b/src/music_visualizer.c
@@ -82,6 +82,7 @@
     STATUS* status = NULL;
     fd_set set;
     int ret;
+	int cnt = 0; // used to set resolution (wether to skip / not to skip a read)
 
     // open mpd fifo
     while((fifo = open(MPD_FIFO, O_RDONLY)) == -1);
@@ -112,7 +113,11 @@
             // read data when avaiable
             read(fifo, (uint16_t*)buf, 2*N_SAMPLES);
             // process read buffer
-            process_fifo(buf, fftBuf, fftAvg);
+			if(cnt == NICENESS) {
+            	process_fifo(buf, fftBuf, fftAvg);
+				cnt = 0;
+			}
+			cnt++;
         }
         // refresh status at SIGALRM
         if(getstatus) {
@@ -125,9 +130,10 @@
         }
         if (ret > 0) {
             // clear screen for printing (only if new data on fifo)
-            erase();
-            print_visual(fftBuf, fftAvg);
-            // print mpd status (if not refreshed, print old one)
+			if(cnt == NICENESS) {
+            	erase();
+           		print_visual(fftBuf, fftAvg);
+			}
         }
         // print mpd status even if no new data is avaiable
         print_mpd_status(status, maxC, maxR/2+maxR/6);
diff --git a/src/settings.h b/src/settings.h
index a853f2d..e36c4f3 100644
--- a/src/settings.h
+++ b/src/settings.h
@@ -3,3 +3,9 @@
 static char defDBlocation[] = "/home/francesco/.mpd/mpd.db"; // the location of the database to be used with the custom search
 #define TIMEOUT 1000 // timeout for connection, define at your own risk. Good values range between 1000 and 5000
 #define STATUS_REFRESH 5 // #seconds between each mpd status refresh
+#define NICENESS 0 // nuber of buffers whose computation is skipped (cyclically).
+				   // More skips mean less precision, nicer on older cpus
+				   // 0 -> do not skip
+				   // 1 -> process one out of two
+				   // 2 -> process one out of three
+				   // etc