Documentation

NVML temp code updated (driver version ~555 and up)

README.md

@@ -36,7 +36,7 @@ Access to the `/dev/i2c` device files, which means either:
 Add the make flag `USE_NVML=1` and the it will also display the main GPU temperature ("GPU1") as reported by the NVIDIA driver.  It will also display the performance cap/clock reason and memory controller utilization.  This requires the NVIDIA management library (NVML) to be installed.
 
 ### VRAM and Hotspot temperature
-Run as root and you can also read the VRAM and "hotspot" temperatures.  These require access the BAR0 through the `/sys/bus/pci/devices/*/resource0` device files.  These sensors are **extremely** undocumented so I can't say anything about their accuracy. If you have a Blackwell (50 series) card you can read all 4 die temp sensors, a 'system' temperature, and every individual VRAM chip temp (add your card in blackwell-temps.c).
+Run as root and you can also read the VRAM and "hotspot" temperatures.  These require access the BAR0 through the `/sys/bus/pci/devices/*/resource0` device files.  These sensors are **extremely** undocumented so I can't say anything about their accuracy. If you have a Blackwell (50 series) card you can read all 4 die temp sensors, a 'system' temperature, and every individual VRAM chip temp (add your card in blackwell-temps.c). You will probably require the kernel parameter `iomem=relaxed`.
 
 ### Datacenter GPU Manager metrics
 Add the make flag `USE_DCGM=1` and it can display real-time metrics from supported Nvidia GPUs like SM occupancy, integer/fp pipe usage, tensor usage, and DRAM bandwidth utilization. This requires a Quadro/Workstation or a Tesla/Datacenter card, it does not work on GeForce cards.

blackwell-sensors.c

@@ -3,6 +3,7 @@
 #include <dirent.h>
 #include <stdint.h>
 #include <math.h>
+#include <stdbool.h>
 
 #include "blackwell-sensors.h"
 

card.h

@@ -36,6 +36,12 @@ struct card_info {
 	void *nvml_device;					/* Pointer to nvmlDevice_t for use in NVML calls */
 };
 
+struct clock_reason {
+	unsigned long long mask;
+	char *short_name;
+	char *long_name;
+};
+
 unsigned short read_pci_id(char *device_path, char *field);
 char *read_nvidia_pci_address(char *device_path);
 void mmap_bar0(struct card_info *card);

nvidia-dcgm.c

@@ -78,8 +78,8 @@ void print_dcgm(int compact, int overwrite)
 	else
 		printf("Clock reasons: ");
 	
-	for (int i = 0; i < (sizeof(clock_reason_names) / sizeof(struct clock_reason)); i++) {
+	for (int i = 0; i < (sizeof(dcgm_clock_reason_names) / sizeof(struct clock_reason)); i++) {
-		if (clock_reasons & clock_reason_names[i].mask) {
+		if (clock_reasons & dcgm_clock_reason_names[i].mask) {
 			if (!single_reason) {
 				if (compact)
 					printf(",");
@@ -89,9 +89,9 @@ void print_dcgm(int compact, int overwrite)
 			single_reason = 0;
 			
 			if (compact)
-				printf("%s", clock_reason_names[i].short_name);
+				printf("%s", dcgm_clock_reason_names[i].short_name);
 			else
-				printf("%s", clock_reason_names[i].long_name);
+				printf("%s", dcgm_clock_reason_names[i].long_name);
 		}
 	}
 	

nvidia-dcgm.h

@@ -26,13 +26,7 @@ static struct metric_reading metric_readings[] =
 	{DCGM_FI_PROF_DRAM_ACTIVE,				"DRAM",		"Memory BW Utilization"}
 };
 
-struct clock_reason {
+static struct clock_reason dcgm_clock_reason_names[] = 
-	unsigned long long mask;
-	char *short_name;
-	char *long_name;
-};
-
-static struct clock_reason clock_reason_names[] = 
 {
 	{DCGM_CLOCKS_EVENT_REASON_GPU_IDLE,			"Idle", 	"GPU idle"},
 	{DCGM_CLOCKS_EVENT_REASON_CLOCKS_SETTING,	"AppClk",	"Application clocks"},

nvidia-sensors.c

@@ -25,8 +25,8 @@ void print_nvml_clock_reason(int compact, struct card_info *card)
 	unsigned long long reasons = get_nvml_clock_reasons(card);
 	int single_reason = 1;
 		
-	for (int i = 0; i < (sizeof(clock_reason_names) / sizeof(struct clock_reason)); i++) {
+	for (int i = 0; i < (sizeof(nvml_clock_reason_names) / sizeof(struct clock_reason)); i++) {
-		if (reasons & clock_reason_names[i].mask) {
+		if (reasons & nvml_clock_reason_names[i].mask) {
 			if (!single_reason) {
 				if (compact)
 					printf(",");
@@ -36,9 +36,9 @@ void print_nvml_clock_reason(int compact, struct card_info *card)
 			single_reason = 0;
 			
 			if (compact)
-				printf("%s", clock_reason_names[i].short_name);
+				printf("%s", nvml_clock_reason_names[i].short_name);
 			else
-				printf("%s", clock_reason_names[i].long_name);
+				printf("%s", nvml_clock_reason_names[i].long_name);
 		}
 	}
 	
@@ -56,13 +56,15 @@ unsigned int get_nvml_temp(struct card_info *card)
 	if (card->nvml_device == NULL)
 		return 0;
 	
-	unsigned int temp;
+	nvmlTemperature_t temp;
-	nvmlReturn_t result = nvmlDeviceGetTemperature(*(nvmlDevice_t*)(card->nvml_device), NVML_TEMPERATURE_GPU, &temp); 
+	temp.version = nvmlTemperature_v1;
+	temp.sensorType = NVML_TEMPERATURE_GPU;
+	nvmlReturn_t result = nvmlDeviceGetTemperatureV(*(nvmlDevice_t*)(card->nvml_device), &temp); 
 	if (result != NVML_SUCCESS) {
 		printf("Failed to get temperature for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
 		return 0;
 	}
-	return temp;
+	return temp.temperature;
 }
 
 unsigned long long get_nvml_clock_reasons(struct card_info *card)

nvidia-sensors.h

@@ -2,13 +2,7 @@
 
 #include "evga-card.h"
 
-struct clock_reason {
+static struct clock_reason nvml_clock_reason_names[] = 
-	unsigned long long mask;
-	char *short_name;
-	char *long_name;
-};
-
-static struct clock_reason clock_reason_names[] = 
 {
 	{nvmlClocksEventReasonGpuIdle,						"Idle", 	"GPU idle"},
 	{nvmlClocksEventReasonApplicationsClocksSetting,	"AppClk",	"Application clocks"},