Allow per-device fan ranges to be set and use them in auto-fan mode.

README

@@ -117,7 +117,7 @@ Options for both config file and command line:
 --failover-only     Don't leak work to backup pools when primary pool is lagging
 --gpu-threads|-g <arg> Number of threads per GPU (1 - 10) (default: 2)
 --gpu-engine <arg>  GPU engine (over)clock range in Mhz - one value, range and/or comma separated list (e.g. 850-900,900,750-850)
---gpu-fan <arg>     Set the GPU fan percentage - one value for all or separate by commas for per card.
+--gpu-fan <arg>     GPU fan percentage range - one value, range and/or comma separated list (e.g. 25-85,85,65)
 --gpu-memclock <arg> Set the GPU memory (over)clock in Mhz - one value for all or separate by commas for per card.
 --gpu-powertune <arg> Set the GPU powertune percentage - one value for all or separate by commas for per card.
 --gpu-vddc <arg>    Set the GPU voltage in Volts - one value for all or separate by commas for per card.
@@ -358,8 +358,9 @@ temperature. By default this is set to 75 degrees C but can be changed with:
 --temp-target
 
 AUTO FAN:
-i.e.
+e.g.
---auto-fan
+--auto-fan (implies 85% upper limit)
+--gpu-fan 25-85,65 --auto-fan
 
 Fan control in auto fan works off the theory that the minimum possible fan
 required to maintain an optimal temperature will use less power, make less

adl.c

@@ -264,6 +264,7 @@ void init_adl(int nDevs)
 			ga->maxspeed = setengine;
 			if (gpus[gpu].min_engine)
 				ga->minspeed = gpus[gpu].min_engine * 100;
+			ga->managed = true;
 		}
 		if (gpus[gpu].gpu_memclock) {
 			int setmem = gpus[gpu].gpu_memclock * 100;
@@ -275,6 +276,7 @@ void init_adl(int nDevs)
 			lpOdPerformanceLevels->aLevels[lev].iMemoryClock = setmem;
 			applog(LOG_INFO, "Setting GPU %d memory clock to %d", gpu, gpus[gpu].gpu_memclock);
 			ADL_Overdrive5_ODPerformanceLevels_Set(iAdapterIndex, lpOdPerformanceLevels);
+			ga->managed = true;
 		}
 		if (gpus[gpu].gpu_vddc) {
 			int setv = gpus[gpu].gpu_vddc * 1000;
@@ -286,16 +288,15 @@ void init_adl(int nDevs)
 			lpOdPerformanceLevels->aLevels[lev].iVddc = setv;
 			applog(LOG_INFO, "Setting GPU %d voltage to %.3f", gpu, gpus[gpu].gpu_vddc);
 			ADL_Overdrive5_ODPerformanceLevels_Set(iAdapterIndex, lpOdPerformanceLevels);
+			ga->managed = true;
 		}
 		ADL_Overdrive5_ODPerformanceLevels_Get(iAdapterIndex, 0, lpOdPerformanceLevels);
 		ga->iEngineClock = lpOdPerformanceLevels->aLevels[lev].iEngineClock;
 		ga->iMemoryClock = lpOdPerformanceLevels->aLevels[lev].iMemoryClock;
 		ga->iVddc = lpOdPerformanceLevels->aLevels[lev].iVddc;
 
-		if (ADL_Overdrive5_FanSpeedInfo_Get(iAdapterIndex, 0, &ga->lpFanSpeedInfo) != ADL_OK) {
+		if (ADL_Overdrive5_FanSpeedInfo_Get(iAdapterIndex, 0, &ga->lpFanSpeedInfo) != ADL_OK)
 			applog(LOG_INFO, "Failed to ADL_Overdrive5_FanSpeedInfo_Get");
-			continue;
-		}
 
 		/* Save the fanspeed values as defaults in case we reset later */
 		ADL_Overdrive5_FanSpeed_Get(ga->iAdapterIndex, 0, &ga->DefFanSpeedValue);
@@ -306,7 +307,9 @@ void init_adl(int nDevs)
 			ga->lpFanSpeedValue.iSpeedType = ADL_DL_FANCTRL_SPEED_TYPE_PERCENT;
 			applog(LOG_INFO, "Setting GPU %d fan speed to %d%%", gpu, gpus[gpu].gpu_fan);
 			ADL_Overdrive5_FanSpeed_Set(ga->iAdapterIndex, 0, &ga->lpFanSpeedValue);
-		}
+			ga->managed = true;
+		} else
+			gpus[gpu].gpu_fan = 85; /* Set a nominal upper limit of 85% */
 
 		/* Not fatal if powercontrol get fails */
 		if (ADL_Overdrive5_PowerControl_Get(ga->iAdapterIndex, &ga->iPercentage, &dummy) != ADL_OK)
@@ -315,13 +318,8 @@ void init_adl(int nDevs)
 		if (gpus[gpu].gpu_powertune) {
 			ADL_Overdrive5_PowerControl_Set(ga->iAdapterIndex, gpus[gpu].gpu_powertune);
 			ADL_Overdrive5_PowerControl_Get(ga->iAdapterIndex, &ga->iPercentage, &dummy);
-		}
-
-		/* Flag if the values of this GPU were ever modified to tell us
-		 * we need to reset the values upon exiting */
-		if (gpus[gpu].gpu_engine || gpus[gpu].gpu_memclock || gpus[gpu].gpu_vddc ||
-		    gpus[gpu].gpu_fan || gpus[gpu].gpu_powertune)
 			ga->managed = true;
+		}
 
 		/* Set some default temperatures for autotune when enabled */
 		ga->targettemp = opt_targettemp;
@@ -856,19 +854,22 @@ void gpu_autotune(int gpu, bool *enable)
 	newengine = engine = gpu_engineclock(gpu) * 100;
 
 	if (temp && fanpercent >= 0 && ga->autofan) {
+		int top = gpus[gpu].gpu_fan;
+		int bot = gpus[gpu].min_fan;
+
 		if (temp > ga->overtemp && fanpercent < 100) {
 			applog(LOG_WARNING, "Overheat detected, increasing fan to 100%");
 			newpercent = 100;
-		} else if (temp > ga->targettemp && fanpercent < 85) {
+		} else if (temp > ga->targettemp && fanpercent < top) {
 			if (opt_debug)
 				applog(LOG_DEBUG, "Temperature over target, increasing fanspeed");
 			if (temp > ga->targettemp + opt_hysteresis)
 				newpercent = fanpercent + 10;
 			else
 				newpercent = fanpercent + 5;
-			if (newpercent > 85)
+			if (newpercent > top)
-				newpercent = 85;
+				newpercent = top;
-		} else if (fanpercent && temp < ga->targettemp - opt_hysteresis) {
+		} else if (fanpercent > bot && temp < ga->targettemp - opt_hysteresis) {
 			if (opt_debug)
 				applog(LOG_DEBUG, "Temperature %d degrees below target, decreasing fanspeed", opt_hysteresis);
 			newpercent = fanpercent - 1;

main.c

@@ -1162,27 +1162,36 @@ static char *set_gpu_engine(char *arg)
 
 static char *set_gpu_fan(char *arg)
 {
-	int i, val = 0, device = 0;
+	int i, val1 = 0, val2 = 0, device = 0;
 	char *nextptr;
 
 	nextptr = strtok(arg, ",");
 	if (nextptr == NULL)
 		return "Invalid parameters for set gpu fan";
-	val = atoi(nextptr);
+	get_intrange(nextptr, &val1, &val2);
-	if (val < 0 || val > 100)
+	if (val1 < 0 || val1 > 100 || val2 < 0 || val2 > 100)
 		return "Invalid value passed to set_gpu_fan";
 
-	gpus[device++].gpu_fan = val;
+	gpus[device].min_fan = val1;
+	gpus[device].gpu_fan = val2;
+	device++;
 
 	while ((nextptr = strtok(NULL, ",")) != NULL) {
-		val = atoi(nextptr);
+		get_intrange(nextptr, &val1, &val2);
-		if (val < 0 || val > 100)
+		if (val1 < 0 || val1 > 100 || val2 < 0 || val2 > 100)
 			return "Invalid value passed to set_gpu_fan";
 
-		gpus[device++].gpu_fan = val;
+		gpus[device].min_fan = val1;
+		gpus[device].gpu_fan = val2;
+		device++;
+	}
+
+	if (device == 1) {
+		for (i = 1; i < 16; i++) {
+			gpus[i].min_fan = gpus[0].min_fan;
+			gpus[i].gpu_fan = gpus[0].gpu_fan;
+		}
 	}
-	for (i = device; i < 16; i++)
-		gpus[i].gpu_fan = val;
 
 	return NULL;
 }
@@ -1337,7 +1346,7 @@ static struct opt_table opt_config_table[] = {
 		     "GPU engine (over)clock range in Mhz - one value, range and/or comma separated list (e.g. 850-900,900,750-850)"),
 	OPT_WITH_ARG("--gpu-fan",
 		     set_gpu_fan, NULL, NULL,
-		     "Set the GPU fan percentage - one value for all or separate by commas for per card."),
+		     "GPU fan percentage range - one value, range and/or comma separated list (e.g. 0-85,85,65)"),
 	OPT_WITH_ARG("--gpu-memclock",
 		     set_gpu_memclock, NULL, NULL,
 		     "Set the GPU memory (over)clock in Mhz - one value for all or separate by commas for per card."),