Merge amd-dev into amd-master 20241010
Signed-off-by: Zhang Ava <niandong.zhang@amd.com> Change-Id: I70496500788e8890924d8d3169b296d9dde34a41
这个提交包含在：
Zhang Ava
+276
-270
@@ -2091,7 +2091,7 @@ class AMDSMICommands():
|
||||
if "throttle" in current_platform_args:
|
||||
if args.throttle:
|
||||
throttle_status = {
|
||||
# gpu metric values
|
||||
# violation status values - counter/accumulated
|
||||
'accumulation_counter': "N/A",
|
||||
'prochot_accumulated': "N/A",
|
||||
'ppt_accumulated': "N/A",
|
||||
@@ -2114,20 +2114,15 @@ class AMDSMICommands():
|
||||
'hbm_thermal_violation_percent': "N/A"
|
||||
}
|
||||
|
||||
try:
|
||||
throttle_status['accumulation_counter'] = gpu_metric['accumulation_counter']
|
||||
throttle_status['prochot_accumulated'] = gpu_metric['prochot_residency_acc']
|
||||
throttle_status['ppt_accumulated'] = gpu_metric['ppt_residency_acc']
|
||||
throttle_status['socket_thermal_accumulated'] = gpu_metric['socket_thm_residency_acc']
|
||||
throttle_status['vr_thermal_accumulated'] = gpu_metric['vr_thm_residency_acc']
|
||||
throttle_status['hbm_thermal_accumulated'] = gpu_metric['hbm_thm_residency_acc']
|
||||
|
||||
except Exception as e:
|
||||
values_dict['throttle'] = throttle_status
|
||||
logging.debug("Failed to get gpu metric information for throttle status' for gpu %s | %s", gpu_id, e)
|
||||
|
||||
try:
|
||||
violation_status = amdsmi_interface.amdsmi_get_violation_status(args.gpu)
|
||||
throttle_status['accumulation_counter'] = violation_status['acc_counter']
|
||||
throttle_status['prochot_accumulated'] = violation_status['acc_prochot_thrm']
|
||||
throttle_status['ppt_accumulated'] = violation_status['acc_ppt_pwr']
|
||||
throttle_status['socket_thermal_accumulated'] = violation_status['acc_socket_thrm']
|
||||
throttle_status['vr_thermal_accumulated'] = violation_status['acc_vr_thrm']
|
||||
throttle_status['hbm_thermal_accumulated'] = violation_status['acc_hbm_thrm']
|
||||
|
||||
throttle_status['prochot_violation_active'] = violation_status['active_prochot_thrm']
|
||||
throttle_status['ppt_violation_active'] = violation_status['active_ppt_pwr']
|
||||
throttle_status['socket_thermal_violation_active'] = violation_status['active_socket_thrm']
|
||||
@@ -3908,19 +3903,24 @@ class AMDSMICommands():
|
||||
args.gpu = device_handle
|
||||
|
||||
# Error if no subcommand args are passed
|
||||
if not any([args.fan is not None,
|
||||
args.perf_level,
|
||||
args.profile,
|
||||
args.compute_partition,
|
||||
args.memory_partition,
|
||||
args.perf_determinism is not None,
|
||||
args.power_cap is not None,
|
||||
args.soc_pstate is not None,
|
||||
args.xgmi_plpd is not None,
|
||||
args.process_isolation is not None,
|
||||
args.clk_limit is not None]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
if self.helpers.is_baremetal():
|
||||
if not any([args.fan is not None,
|
||||
args.perf_level,
|
||||
args.profile,
|
||||
args.compute_partition,
|
||||
args.memory_partition,
|
||||
args.perf_determinism is not None,
|
||||
args.power_cap is not None,
|
||||
args.soc_pstate is not None,
|
||||
args.xgmi_plpd is not None,
|
||||
args.clk_limit is not None,
|
||||
args.process_isolation is not None]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
else:
|
||||
if not any([args.process_isolation is not None]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
|
||||
# Build GPU string for errors
|
||||
try:
|
||||
@@ -3934,100 +3934,113 @@ class AMDSMICommands():
|
||||
gpu_string = f"GPU ID: {gpu_id} BDF:{gpu_bdf}"
|
||||
|
||||
# Handle args
|
||||
if isinstance(args.fan, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_fan_speed(args.gpu, 0, args.fan)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set fan speed {args.fan} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'fan', f"Successfully set fan speed {args.fan}")
|
||||
if args.perf_level:
|
||||
perf_level = amdsmi_interface.AmdSmiDevPerfLevel[args.perf_level]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, perf_level)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set performance level {args.perf_level} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'perflevel', f"Successfully set performance level {args.perf_level}")
|
||||
if args.profile:
|
||||
self.logger.store_output(args.gpu, 'profile', "Not Yet Implemented")
|
||||
if isinstance(args.perf_determinism, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_determinism_mode(args.gpu, args.perf_determinism)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set performance determinism and clock frequency to {args.perf_determinism} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'perfdeterminism', f"Successfully enabled performance determinism and set GFX clock frequency to {args.perf_determinism}")
|
||||
if args.compute_partition:
|
||||
compute_partition = amdsmi_interface.AmdSmiComputePartitionType[args.compute_partition]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_compute_partition(args.gpu, compute_partition)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set compute partition to {args.compute_partition} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'computepartition', f"Successfully set compute partition to {args.compute_partition}")
|
||||
if args.memory_partition:
|
||||
memory_partition = amdsmi_interface.AmdSmiMemoryPartitionType[args.memory_partition]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_memory_partition(args.gpu, memory_partition)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set memory partition to {args.memory_partition} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'memorypartition', f"Successfully set memory partition to {args.memory_partition}")
|
||||
if isinstance(args.power_cap, int):
|
||||
try:
|
||||
power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
|
||||
logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
|
||||
min_power_cap = power_cap_info["min_power_cap"]
|
||||
min_power_cap = self.helpers.convert_SI_unit(min_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
max_power_cap = power_cap_info["max_power_cap"]
|
||||
max_power_cap = self.helpers.convert_SI_unit(max_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
current_power_cap = power_cap_info["power_cap"]
|
||||
current_power_cap = self.helpers.convert_SI_unit(current_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
raise ValueError(f"Unable to get power cap info from {gpu_string}") from e
|
||||
|
||||
if args.power_cap == current_power_cap:
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap is already set to {args.power_cap}")
|
||||
elif args.power_cap >= min_power_cap and args.power_cap <= max_power_cap:
|
||||
if self.helpers.is_baremetal():
|
||||
if isinstance(args.fan, int):
|
||||
try:
|
||||
new_power_cap = self.helpers.convert_SI_unit(args.power_cap, AMDSMIHelpers.SI_Unit.BASE,
|
||||
AMDSMIHelpers.SI_Unit.MICRO)
|
||||
amdsmi_interface.amdsmi_set_power_cap(args.gpu, 0, new_power_cap)
|
||||
amdsmi_interface.amdsmi_set_gpu_fan_speed(args.gpu, 0, args.fan)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set power cap to {args.power_cap} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Successfully set power cap to {args.power_cap}")
|
||||
else:
|
||||
# setting power cap to 0 will return the current power cap so the technical minimum value is 1
|
||||
if min_power_cap == 0:
|
||||
min_power_cap = 1
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap must be between {min_power_cap} and {max_power_cap}")
|
||||
if isinstance(args.soc_pstate, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_soc_pstate(args.gpu, args.soc_pstate)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set dpm soc pstate policy to {args.soc_pstate} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'socpstate', f"Successfully soc pstate dpm policy to id {args.soc_pstate}")
|
||||
if isinstance(args.xgmi_plpd, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_xgmi_plpd(args.gpu, args.xgmi_plpd)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set XGMI policy to {args.xgmi_plpd} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'xgmiplpd', f"Successfully set per-link power down policy to id {args.xgmi_plpd}")
|
||||
raise ValueError(f"Unable to set fan speed {args.fan} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'fan', f"Successfully set fan speed {args.fan}")
|
||||
if args.perf_level:
|
||||
perf_level = amdsmi_interface.AmdSmiDevPerfLevel[args.perf_level]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, perf_level)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set performance level {args.perf_level} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'perflevel', f"Successfully set performance level {args.perf_level}")
|
||||
if args.profile:
|
||||
self.logger.store_output(args.gpu, 'profile', "Not Yet Implemented")
|
||||
if isinstance(args.perf_determinism, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_determinism_mode(args.gpu, args.perf_determinism)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set performance determinism and clock frequency to {args.perf_determinism} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'perfdeterminism', f"Successfully enabled performance determinism and set GFX clock frequency to {args.perf_determinism}")
|
||||
if args.compute_partition:
|
||||
compute_partition = amdsmi_interface.AmdSmiComputePartitionType[args.compute_partition]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_compute_partition(args.gpu, compute_partition)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set compute partition to {args.compute_partition} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'computepartition', f"Successfully set compute partition to {args.compute_partition}")
|
||||
if args.memory_partition:
|
||||
memory_partition = amdsmi_interface.AmdSmiMemoryPartitionType[args.memory_partition]
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_memory_partition(args.gpu, memory_partition)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set memory partition to {args.memory_partition} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'memorypartition', f"Successfully set memory partition to {args.memory_partition}")
|
||||
if isinstance(args.power_cap, int):
|
||||
try:
|
||||
power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
|
||||
logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
|
||||
min_power_cap = power_cap_info["min_power_cap"]
|
||||
min_power_cap = self.helpers.convert_SI_unit(min_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
max_power_cap = power_cap_info["max_power_cap"]
|
||||
max_power_cap = self.helpers.convert_SI_unit(max_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
current_power_cap = power_cap_info["power_cap"]
|
||||
current_power_cap = self.helpers.convert_SI_unit(current_power_cap, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
raise ValueError(f"Unable to get power cap info from {gpu_string}") from e
|
||||
|
||||
if args.power_cap == current_power_cap:
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap is already set to {args.power_cap}")
|
||||
elif args.power_cap >= min_power_cap and args.power_cap <= max_power_cap:
|
||||
try:
|
||||
new_power_cap = self.helpers.convert_SI_unit(args.power_cap, AMDSMIHelpers.SI_Unit.BASE,
|
||||
AMDSMIHelpers.SI_Unit.MICRO)
|
||||
amdsmi_interface.amdsmi_set_power_cap(args.gpu, 0, new_power_cap)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set power cap to {args.power_cap} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Successfully set power cap to {args.power_cap}")
|
||||
else:
|
||||
# setting power cap to 0 will return the current power cap so the technical minimum value is 1
|
||||
if min_power_cap == 0:
|
||||
min_power_cap = 1
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap must be between {min_power_cap} and {max_power_cap}")
|
||||
if isinstance(args.soc_pstate, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_soc_pstate(args.gpu, args.soc_pstate)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set dpm soc pstate policy to {args.soc_pstate} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'socpstate', f"Successfully soc pstate dpm policy to id {args.soc_pstate}")
|
||||
if isinstance(args.xgmi_plpd, int):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_xgmi_plpd(args.gpu, args.xgmi_plpd)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set XGMI policy to {args.xgmi_plpd} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'xgmiplpd', f"Successfully set per-link power down policy to id {args.xgmi_plpd}")
|
||||
if isinstance(args.clk_limit, tuple):
|
||||
try:
|
||||
clk_type = args.clk_limit.clk_type
|
||||
lim_type = args.clk_limit.lim_type
|
||||
val = args.clk_limit.val
|
||||
amdsmi_interface.amdsmi_set_gpu_clk_limit(args.gpu, clk_type, lim_type, val)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set {args.clk_limit.lim_type} of {args.clk_limit.clk_type} to {args.clk_limit.val} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'clk_limit', f"Successfully changed {args.clk_limit.lim_type} of {args.clk_limit.clk_type} to {args.clk_limit.val}")
|
||||
|
||||
if isinstance(args.process_isolation, int):
|
||||
status_string = "Enabled" if args.process_isolation else "Disabled"
|
||||
result = f"Requested process isolation to {status_string}" # This should not print out
|
||||
@@ -4044,17 +4057,6 @@ class AMDSMICommands():
|
||||
raise ValueError(f"Unable to set process isolation to {status_string} on {gpu_string}") from e
|
||||
|
||||
self.logger.store_output(args.gpu, 'process_isolation', result)
|
||||
if isinstance(args.clk_limit, tuple):
|
||||
try:
|
||||
clk_type = args.clk_limit.clk_type
|
||||
lim_type = args.clk_limit.lim_type
|
||||
val = args.clk_limit.val
|
||||
amdsmi_interface.amdsmi_set_gpu_clk_limit(args.gpu, clk_type, lim_type, val)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to set {args.clk_limit.lim_type} of {args.clk_limit.clk_type} to {args.clk_limit.val} on {gpu_string}") from e
|
||||
self.logger.store_output(args.gpu, 'clk_limit', f"Successfully changed {args.clk_limit.lim_type} of {args.clk_limit.clk_type} to {args.clk_limit.val}")
|
||||
|
||||
if multiple_devices:
|
||||
self.logger.store_multiple_device_output()
|
||||
@@ -4267,159 +4269,163 @@ class AMDSMICommands():
|
||||
gpu_id = self.helpers.get_gpu_id_from_device_handle(args.gpu)
|
||||
|
||||
# Error if no subcommand args are passed
|
||||
if not any([args.gpureset, args.clocks, args.fans, args.profile, args.xgmierr, \
|
||||
args.perf_determinism, args.compute_partition, args.memory_partition, \
|
||||
args.power_cap, args.clean_local_data]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
if self.helpers.is_baremetal():
|
||||
if not any([args.gpureset, args.clocks, args.fans, args.profile, args.xgmierr, \
|
||||
args.perf_determinism, args.compute_partition, args.memory_partition, \
|
||||
args.power_cap, args.clean_local_data]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
else:
|
||||
if not any([args.clean_local_data]):
|
||||
command = " ".join(sys.argv[1:])
|
||||
raise AmdSmiRequiredCommandException(command, self.logger.format)
|
||||
|
||||
if args.gpureset:
|
||||
if self.helpers.is_amd_device(args.gpu):
|
||||
if self.helpers.is_baremetal():
|
||||
if args.gpureset:
|
||||
if self.helpers.is_amd_device(args.gpu):
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu(args.gpu)
|
||||
result = 'Successfully reset GPU'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "Failed to reset GPU"
|
||||
else:
|
||||
result = 'Unable to reset non-amd GPU'
|
||||
self.logger.store_output(args.gpu, 'gpu_reset', result)
|
||||
if args.clocks:
|
||||
reset_clocks_results = {'overdrive': '',
|
||||
'clocks': '',
|
||||
'performance': ''}
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu(args.gpu)
|
||||
result = 'Successfully reset GPU'
|
||||
amdsmi_interface.amdsmi_set_gpu_overdrive_level(args.gpu, 0)
|
||||
reset_clocks_results['overdrive'] = 'Overdrive set to 0'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "Failed to reset GPU"
|
||||
else:
|
||||
result = 'Unable to reset non-amd GPU'
|
||||
reset_clocks_results['overdrive'] = "N/A"
|
||||
logging.debug("Failed to reset overdrive on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
self.logger.store_output(args.gpu, 'gpu_reset', result)
|
||||
if args.clocks:
|
||||
reset_clocks_results = {'overdrive': '',
|
||||
'clocks': '',
|
||||
'performance': ''}
|
||||
try:
|
||||
amdsmi_interface.amdsmi_set_gpu_overdrive_level(args.gpu, 0)
|
||||
reset_clocks_results['overdrive'] = 'Overdrive set to 0'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_clocks_results['overdrive'] = "N/A"
|
||||
logging.debug("Failed to reset overdrive on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_clocks_results['clocks'] = 'Successfully reset clocks'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_clocks_results['clocks'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_clocks_results['performance'] = 'Performance level reset to auto'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_clocks_results['performance'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
self.logger.store_output(args.gpu, 'reset_clocks', reset_clocks_results)
|
||||
if args.fans:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_fan(args.gpu, 0)
|
||||
result = 'Successfully reset fan speed to driver control'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset fans on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
self.logger.store_output(args.gpu, 'reset_fans', result)
|
||||
if args.profile:
|
||||
reset_profile_results = {'power_profile' : '',
|
||||
'performance_level': ''}
|
||||
try:
|
||||
power_profile_mask = amdsmi_interface.AmdSmiPowerProfilePresetMasks.BOOTUP_DEFAULT
|
||||
amdsmi_interface.amdsmi_set_gpu_power_profile(args.gpu, 0, power_profile_mask)
|
||||
reset_profile_results['power_profile'] = 'Successfully reset Power Profile'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_profile_results['power_profile'] = "N/A"
|
||||
logging.debug("Failed to reset power profile on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_profile_results['performance_level'] = 'Successfully reset Performance Level'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_profile_results['performance_level'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
self.logger.store_output(args.gpu, 'reset_profile', reset_profile_results)
|
||||
if args.xgmierr:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_xgmi_error(args.gpu)
|
||||
result = 'Successfully reset XGMI Error count'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset xgmi error count on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_xgmi_err', result)
|
||||
if args.perf_determinism:
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
result = 'Successfully disabled performance determinism'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to set perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_perf_determinism', result)
|
||||
if args.compute_partition:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_compute_partition(args.gpu)
|
||||
result = 'Successfully reset compute partition'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset compute partition on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_compute_partition', result)
|
||||
if args.memory_partition:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_memory_partition(args.gpu)
|
||||
result = 'Successfully reset memory partition'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset memory partition on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_memory_partition', result)
|
||||
if args.power_cap:
|
||||
try:
|
||||
power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
|
||||
logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
|
||||
default_power_cap_in_w = power_cap_info["default_power_cap"]
|
||||
default_power_cap_in_w = self.helpers.convert_SI_unit(default_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
current_power_cap_in_w = power_cap_info["power_cap"]
|
||||
current_power_cap_in_w = self.helpers.convert_SI_unit(current_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
raise ValueError(f"Unable to get power cap info from {gpu_id}") from e
|
||||
|
||||
if current_power_cap_in_w == default_power_cap_in_w:
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap is already set to {default_power_cap_in_w}")
|
||||
else:
|
||||
try:
|
||||
default_power_cap_in_uw = self.helpers.convert_SI_unit(default_power_cap_in_w,
|
||||
AMDSMIHelpers.SI_Unit.BASE,
|
||||
AMDSMIHelpers.SI_Unit.MICRO)
|
||||
amdsmi_interface.amdsmi_set_power_cap(args.gpu, 0, default_power_cap_in_uw)
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_clocks_results['clocks'] = 'Successfully reset clocks'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to reset power cap to {default_power_cap_in_w} on GPU {gpu_id}") from e
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Successfully set power cap to {default_power_cap_in_w}")
|
||||
reset_clocks_results['clocks'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_clocks_results['performance'] = 'Performance level reset to auto'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_clocks_results['performance'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
self.logger.store_output(args.gpu, 'reset_clocks', reset_clocks_results)
|
||||
if args.fans:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_fan(args.gpu, 0)
|
||||
result = 'Successfully reset fan speed to driver control'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset fans on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_fans', result)
|
||||
if args.profile:
|
||||
reset_profile_results = {'power_profile' : '',
|
||||
'performance_level': ''}
|
||||
try:
|
||||
power_profile_mask = amdsmi_interface.AmdSmiPowerProfilePresetMasks.BOOTUP_DEFAULT
|
||||
amdsmi_interface.amdsmi_set_gpu_power_profile(args.gpu, 0, power_profile_mask)
|
||||
reset_profile_results['power_profile'] = 'Successfully reset Power Profile'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_profile_results['power_profile'] = "N/A"
|
||||
logging.debug("Failed to reset power profile on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
reset_profile_results['performance_level'] = 'Successfully reset Performance Level'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
reset_profile_results['performance_level'] = "N/A"
|
||||
logging.debug("Failed to reset perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_profile', reset_profile_results)
|
||||
if args.xgmierr:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_xgmi_error(args.gpu)
|
||||
result = 'Successfully reset XGMI Error count'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset xgmi error count on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_xgmi_err', result)
|
||||
if args.perf_determinism:
|
||||
try:
|
||||
level_auto = amdsmi_interface.AmdSmiDevPerfLevel.AUTO
|
||||
amdsmi_interface.amdsmi_set_gpu_perf_level(args.gpu, level_auto)
|
||||
result = 'Successfully disabled performance determinism'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to set perf level on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_perf_determinism', result)
|
||||
if args.compute_partition:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_compute_partition(args.gpu)
|
||||
result = 'Successfully reset compute partition'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset compute partition on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_compute_partition', result)
|
||||
if args.memory_partition:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_reset_gpu_memory_partition(args.gpu)
|
||||
result = 'Successfully reset memory partition'
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
result = "N/A"
|
||||
logging.debug("Failed to reset memory partition on gpu %s | %s", gpu_id, e.get_error_info())
|
||||
self.logger.store_output(args.gpu, 'reset_memory_partition', result)
|
||||
if args.power_cap:
|
||||
try:
|
||||
power_cap_info = amdsmi_interface.amdsmi_get_power_cap_info(args.gpu)
|
||||
logging.debug(f"Power cap info for gpu {gpu_id} | {power_cap_info}")
|
||||
default_power_cap_in_w = power_cap_info["default_power_cap"]
|
||||
default_power_cap_in_w = self.helpers.convert_SI_unit(default_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
current_power_cap_in_w = power_cap_info["power_cap"]
|
||||
current_power_cap_in_w = self.helpers.convert_SI_unit(current_power_cap_in_w, AMDSMIHelpers.SI_Unit.MICRO)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
raise ValueError(f"Unable to get power cap info from {gpu_id}") from e
|
||||
|
||||
if current_power_cap_in_w == default_power_cap_in_w:
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Power cap is already set to {default_power_cap_in_w}")
|
||||
else:
|
||||
try:
|
||||
default_power_cap_in_uw = self.helpers.convert_SI_unit(default_power_cap_in_w,
|
||||
AMDSMIHelpers.SI_Unit.BASE,
|
||||
AMDSMIHelpers.SI_Unit.MICRO)
|
||||
amdsmi_interface.amdsmi_set_power_cap(args.gpu, 0, default_power_cap_in_uw)
|
||||
except amdsmi_exception.AmdSmiLibraryException as e:
|
||||
if e.get_error_code() == amdsmi_interface.amdsmi_wrapper.AMDSMI_STATUS_NO_PERM:
|
||||
raise PermissionError('Command requires elevation') from e
|
||||
raise ValueError(f"Unable to reset power cap to {default_power_cap_in_w} on GPU {gpu_id}") from e
|
||||
self.logger.store_output(args.gpu, 'powercap', f"Successfully set power cap to {default_power_cap_in_w}")
|
||||
|
||||
if args.clean_local_data:
|
||||
try:
|
||||
amdsmi_interface.amdsmi_clean_gpu_local_data(args.gpu)
|
||||
|
||||
@@ -1077,8 +1077,8 @@ class AMDSMIParser(argparse.ArgumentParser):
|
||||
set_value_parser.add_argument('-C', '--compute-partition', action='store', choices=self.helpers.get_compute_partition_types(), type=str.upper, required=False, help=set_compute_partition_help, metavar='PARTITION')
|
||||
set_value_parser.add_argument('-M', '--memory-partition', action='store', choices=self.helpers.get_memory_partition_types(), type=str.upper, required=False, help=set_memory_partition_help, metavar='PARTITION')
|
||||
set_value_parser.add_argument('-o', '--power-cap', action='store', type=self._positive_int, required=False, help=set_power_cap_help, metavar='WATTS')
|
||||
set_value_parser.add_argument('-p', '--soc-pstate', action='store', required=False, type=self._not_negative_int, help=set_soc_pstate_help, metavar='POLICY_ID')
|
||||
set_value_parser.add_argument('-x', '--xgmi-plpd', action='store', required=False, type=self._not_negative_int, help=set_xgmi_plpd_help, metavar='POLICY_ID')
|
||||
set_value_parser.add_argument('-p', '--soc-pstate', action='store', required=False, type=self._not_negative_int, help=set_soc_pstate_help, metavar='POLICY_ID')
|
||||
set_value_parser.add_argument('-x', '--xgmi-plpd', action='store', required=False, type=self._not_negative_int, help=set_xgmi_plpd_help, metavar='POLICY_ID')
|
||||
set_value_parser.add_argument('-L', '--clk-limit', action=self._limit_select(), nargs=3, required=False, help=set_clk_limit_help, metavar=('CLK_TYPE', 'LIM_TYPE', 'VALUE'))
|
||||
|
||||
set_value_parser.add_argument('-R', '--process-isolation', action='store', choices=[0,1], type=self._not_negative_int, required=False, help=set_process_isolation_help, metavar='STATUS')
|
||||
|
||||
+7
-2
@@ -528,11 +528,16 @@ typedef struct {
|
||||
} amdsmi_vram_usage_t;
|
||||
/**
|
||||
* @brief This structure hold violation status information.
|
||||
* Note: for MI3x asics and higher, older ASICs will show unsupported.
|
||||
*/
|
||||
typedef struct {
|
||||
uint64_t reference_timestamp; //!< Represents CPU timestamp in microseconds (uS)
|
||||
uint64_t violation_timestamp; //!< Violation time in milliseconds (ms)
|
||||
uint64_t acc_counter; //!< Current accumulated counter; Max uint64 means unsupported
|
||||
uint64_t acc_prochot_thrm; //!< Current accumulated processor hot violation count; Max uint64 means unsupported
|
||||
uint64_t acc_ppt_pwr; //!< PVIOL; Current accumulated Package Power Tracking (PPT) count; Max uint64 means unsupported
|
||||
uint64_t acc_socket_thrm; //!< TVIOL; Current accumulated Socket thermal count; Max uint64 means unsupported
|
||||
uint64_t acc_vr_thrm; //!< Current accumulated voltage regulator count; Max uint64 means unsupported
|
||||
uint64_t acc_hbm_thrm; //!< Current accumulated High Bandwidth Memory (HBM) thermal count; Max uint64 means unsupported
|
||||
uint64_t per_prochot_thrm; //!< Processor hot violation % (greater than 0% is a violation); Max uint64 means unsupported
|
||||
uint64_t per_ppt_pwr; //!< PVIOL; Package Power Tracking (PPT) violation % (greater than 0% is a violation); Max uint64 means unsupported
|
||||
uint64_t per_socket_thrm; //!< TVIOL; Socket thermal violation % (greater than 0% is a violation); Max uint64 means unsupported
|
||||
@@ -543,7 +548,7 @@ typedef struct {
|
||||
uint8_t active_socket_thrm; //!< Socket thermal violation; 1 = active 0 = not active; Max uint8 means unsupported
|
||||
uint8_t active_vr_thrm; //!< Voltage regulator violation; 1 = active 0 = not active; Max uint8 means unsupported
|
||||
uint8_t active_hbm_thrm; //!< High Bandwidth Memory (HBM) thermal violation; 1 = active 0 = not active; Max uint8 means unsupported
|
||||
uint64_t reserved[24]; // Reserved for new violation info
|
||||
uint64_t reserved[30]; // Reserved for new violation info
|
||||
} amdsmi_violation_status_t;
|
||||
typedef struct {
|
||||
amdsmi_range_t supported_freq_range;
|
||||
|
||||
@@ -2020,6 +2020,12 @@ def amdsmi_get_violation_status(
|
||||
return {
|
||||
"reference_timestamp": _validate_if_max_uint(violation_status.reference_timestamp, MaxUIntegerTypes.UINT64_T),
|
||||
"violation_timestamp": _validate_if_max_uint(violation_status.violation_timestamp, MaxUIntegerTypes.UINT64_T),
|
||||
"acc_counter": _validate_if_max_uint(violation_status.acc_counter, MaxUIntegerTypes.UINT64_T),
|
||||
"acc_prochot_thrm": _validate_if_max_uint(violation_status.acc_prochot_thrm, MaxUIntegerTypes.UINT64_T),
|
||||
"acc_ppt_pwr": _validate_if_max_uint(violation_status.acc_ppt_pwr, MaxUIntegerTypes.UINT64_T), #PVIOL
|
||||
"acc_socket_thrm": _validate_if_max_uint(violation_status.acc_socket_thrm, MaxUIntegerTypes.UINT64_T), #TVIOL
|
||||
"acc_vr_thrm": _validate_if_max_uint(violation_status.acc_vr_thrm, MaxUIntegerTypes.UINT64_T),
|
||||
"acc_hbm_thrm": _validate_if_max_uint(violation_status.acc_hbm_thrm, MaxUIntegerTypes.UINT64_T),
|
||||
"per_prochot_thrm": _validate_if_max_uint(violation_status.per_prochot_thrm, MaxUIntegerTypes.UINT64_T, isActivity=True),
|
||||
"per_ppt_pwr": _validate_if_max_uint(violation_status.per_ppt_pwr, MaxUIntegerTypes.UINT64_T, isActivity=True), #PVIOL
|
||||
"per_socket_thrm": _validate_if_max_uint(violation_status.per_socket_thrm, MaxUIntegerTypes.UINT64_T, isActivity=True), #TVIOL
|
||||
|
||||
+20
-14
@@ -734,6 +734,12 @@ struct_amdsmi_violation_status_t._pack_ = 1 # source:False
|
||||
struct_amdsmi_violation_status_t._fields_ = [
|
||||
('reference_timestamp', ctypes.c_uint64),
|
||||
('violation_timestamp', ctypes.c_uint64),
|
||||
('acc_counter', ctypes.c_uint64),
|
||||
('acc_prochot_thrm', ctypes.c_uint64),
|
||||
('acc_ppt_pwr', ctypes.c_uint64),
|
||||
('acc_socket_thrm', ctypes.c_uint64),
|
||||
('acc_vr_thrm', ctypes.c_uint64),
|
||||
('acc_hbm_thrm', ctypes.c_uint64),
|
||||
('per_prochot_thrm', ctypes.c_uint64),
|
||||
('per_ppt_pwr', ctypes.c_uint64),
|
||||
('per_socket_thrm', ctypes.c_uint64),
|
||||
@@ -745,7 +751,7 @@ struct_amdsmi_violation_status_t._fields_ = [
|
||||
('active_vr_thrm', ctypes.c_ubyte),
|
||||
('active_hbm_thrm', ctypes.c_ubyte),
|
||||
('PADDING_0', ctypes.c_ubyte * 3),
|
||||
('reserved', ctypes.c_uint64 * 24),
|
||||
('reserved', ctypes.c_uint64 * 30),
|
||||
]
|
||||
|
||||
amdsmi_violation_status_t = struct_amdsmi_violation_status_t
|
||||
@@ -798,19 +804,6 @@ amdsmi_card_form_factor_t = ctypes.c_uint32 # enum
|
||||
class struct_amdsmi_pcie_info_t(Structure):
|
||||
pass
|
||||
|
||||
class struct_pcie_static_(Structure):
|
||||
pass
|
||||
|
||||
struct_pcie_static_._pack_ = 1 # source:False
|
||||
struct_pcie_static_._fields_ = [
|
||||
('max_pcie_width', ctypes.c_uint16),
|
||||
('PADDING_0', ctypes.c_ubyte * 2),
|
||||
('max_pcie_speed', ctypes.c_uint32),
|
||||
('pcie_interface_version', ctypes.c_uint32),
|
||||
('slot_type', amdsmi_card_form_factor_t),
|
||||
('reserved', ctypes.c_uint64 * 10),
|
||||
]
|
||||
|
||||
class struct_pcie_metric_(Structure):
|
||||
pass
|
||||
|
||||
@@ -831,6 +824,19 @@ struct_pcie_metric_._fields_ = [
|
||||
('reserved', ctypes.c_uint64 * 12),
|
||||
]
|
||||
|
||||
class struct_pcie_static_(Structure):
|
||||
pass
|
||||
|
||||
struct_pcie_static_._pack_ = 1 # source:False
|
||||
struct_pcie_static_._fields_ = [
|
||||
('max_pcie_width', ctypes.c_uint16),
|
||||
('PADDING_0', ctypes.c_ubyte * 2),
|
||||
('max_pcie_speed', ctypes.c_uint32),
|
||||
('pcie_interface_version', ctypes.c_uint32),
|
||||
('slot_type', amdsmi_card_form_factor_t),
|
||||
('reserved', ctypes.c_uint64 * 10),
|
||||
]
|
||||
|
||||
struct_amdsmi_pcie_info_t._pack_ = 1 # source:False
|
||||
struct_amdsmi_pcie_info_t._fields_ = [
|
||||
('pcie_static', struct_pcie_static_),
|
||||
|
||||
+17
-1
@@ -637,6 +637,14 @@ amdsmi_status_t amdsmi_get_violation_status(amdsmi_processor_handle processor_ha
|
||||
|
||||
violation_status->reference_timestamp = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->violation_timestamp = std::numeric_limits<uint64_t>::max();
|
||||
|
||||
violation_status->acc_counter = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->acc_prochot_thrm = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->acc_ppt_pwr = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->acc_socket_thrm = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->acc_vr_thrm = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->acc_hbm_thrm = std::numeric_limits<uint64_t>::max();
|
||||
|
||||
violation_status->per_prochot_thrm = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->per_ppt_pwr = std::numeric_limits<uint64_t>::max();
|
||||
violation_status->per_socket_thrm = std::numeric_limits<uint64_t>::max();
|
||||
@@ -702,6 +710,14 @@ amdsmi_status_t amdsmi_get_violation_status(amdsmi_processor_handle processor_ha
|
||||
return status;
|
||||
}
|
||||
|
||||
// Insert current accumulator counters into struct
|
||||
violation_status->acc_counter = metric_info_b.accumulation_counter;
|
||||
violation_status->acc_prochot_thrm = metric_info_b.prochot_residency_acc;
|
||||
violation_status->acc_ppt_pwr = metric_info_b.ppt_residency_acc;
|
||||
violation_status->acc_socket_thrm = metric_info_b.socket_thm_residency_acc;
|
||||
violation_status->acc_vr_thrm = metric_info_b.vr_thm_residency_acc;
|
||||
violation_status->acc_hbm_thrm = metric_info_b.hbm_thm_residency_acc;
|
||||
|
||||
ss << __PRETTY_FUNCTION__ << " | "
|
||||
<< "[gpu_metrics A] metric_info_a.accumulation_counter: " << std::dec
|
||||
<< metric_info_a.accumulation_counter
|
||||
@@ -818,7 +834,7 @@ amdsmi_status_t amdsmi_get_violation_status(amdsmi_processor_handle processor_ha
|
||||
}
|
||||
if ( (metric_info_b.hbm_thm_residency_acc != std::numeric_limits<uint64_t>::max()
|
||||
|| metric_info_a.hbm_thm_residency_acc != std::numeric_limits<uint64_t>::max())
|
||||
&& (metric_info_b.hbm_thm_residency_acc >= metric_info_a.vr_thm_residency_acc)
|
||||
&& (metric_info_b.hbm_thm_residency_acc >= metric_info_a.hbm_thm_residency_acc)
|
||||
&& ((metric_info_b.accumulation_counter - metric_info_a.accumulation_counter) > 0) ) {
|
||||
violation_status->per_hbm_thrm =
|
||||
(((metric_info_b.hbm_thm_residency_acc -
|
||||
|
||||
@@ -864,7 +864,8 @@ class TestAmdSmiPythonInterface(unittest.TestCase):
|
||||
print()
|
||||
self.tearDown()
|
||||
|
||||
# Only supported on MI300+ ASICs
|
||||
# amdsmi_get_violation_status is only supported on MI300+ ASICs
|
||||
# We should expect a not supported status for Navi / MI100 / MI2x ASICs
|
||||
@handle_exceptions
|
||||
def test_get_violation_status(self):
|
||||
self.setUp()
|
||||
@@ -882,6 +883,17 @@ class TestAmdSmiPythonInterface(unittest.TestCase):
|
||||
print(" Violation Timestamp: {}".format(
|
||||
violation_status['violation_timestamp']))
|
||||
|
||||
print(" Current Prochot Thrm Accumulated (Count): {}".format(
|
||||
violation_status['acc_prochot_thrm']))
|
||||
print(" Current PVIOL (acc_ppt_pwr) Accumulated (Count): {}".format(
|
||||
violation_status['acc_ppt_pwr']))
|
||||
print(" Current TVIOL (acc_socket_thrm) Accumulated (Count): {}".format(
|
||||
violation_status['acc_socket_thrm']))
|
||||
print(" Current VR_THRM Accumulated (Count): {}".format(
|
||||
violation_status['acc_vr_thrm']))
|
||||
print(" Current HBM Thrm Accumulated (Count): {}".format(
|
||||
violation_status['acc_hbm_thrm']))
|
||||
|
||||
print(" Prochot Thrm Violation (%): {}".format(
|
||||
violation_status['per_prochot_thrm']))
|
||||
print(" PVIOL (per_ppt_pwr) (%): {}".format(
|
||||
|
||||
在新工单中引用
屏蔽一个用户