dev: fixing oc

clore_hosting/main.py

@@ -537,8 +537,6 @@ class CloreClient:
                 cpu_usage = await get_specs.get_cpu_usage()
                 ram_usage = await get_specs.get_ram_usage()
                 gpu_list = current_specs["gpus"]["nvidia"]+current_specs["gpus"]["amd"]
-                print("realtime gpus")
-                print(gpu_list)
                 submit_document = {
                     "update_realtime_data":True,
                     "gpus": gpu_list,
@@ -559,12 +557,8 @@ class CloreClient:
                 print(current_specs)
                 if self.last_hw_specs_submit < (utils.unix_timestamp()-1800):
                     self.last_hw_specs_submit=utils.unix_timestamp()
-                    print("submit specs start")
                     await self.submit_specs(current_specs)
-                    print("submit specs end")
-                print("update realtime start")
                 await self.update_realtime_data(current_specs)
-                print("update realtime end")
                 try:
                     if self.xfs_state == "active" and len(current_specs["gpus"]["nvidia"]) > 0 and not self.runned_pull_selftest:
                         await clore_partner.check_to_pull_selftest(current_specs)

lib/constants.py

@@ -0,0 +1,4 @@
+GPU_ID_TO_NAME = {
+    "0x20C210DE": "NVIDIA CMP 170HX",
+    "0x208210DE": "NVIDIA CMP 170HX"
+}

lib/get_specs.py

@@ -4,6 +4,7 @@ import xml.etree.ElementTree as ET
 from lib import docker_interface
 from typing import Dict, List, Optional
 from lib import utils
+from lib import constants
 import subprocess
 import speedtest
 import platform
@@ -241,11 +242,6 @@ def get_bus_spec(bus_id):
         return PCIBusInfo()
 
 def get_gpu_info():
-    GPU_ID_TO_NAME = {
-        "0x20C210DE": "NVIDIA CMP 170HX",
-        "0x208210DE": "NVIDIA CMP 170HX"
-    }
-
     gpu_str = "0x Unknown"
     nvml_err = False
     gpu_mem = 0
@@ -273,8 +269,8 @@ def get_gpu_info():
                 parts = [s.strip() for s in line.split(',')]
                 if len(parts)>12 and index>0:
                     gpu_name_xl = parts[1]
-                    if gpu_name_xl == "NVIDIA Graphics Device" and parts[13] in GPU_ID_TO_NAME:
+                    if gpu_name_xl == "NVIDIA Graphics Device" and parts[13] in constants.GPU_ID_TO_NAME:
-                        gpu_name_xl = GPU_ID_TO_NAME[parts[13]]
+                        gpu_name_xl = constants.GPU_ID_TO_NAME[parts[13]]
 
                     xl_gpu_info={
                         "id":index-1,
@@ -309,8 +305,8 @@ def get_gpu_info():
 
                     gpu_name = parts[1].strip()
                     gpu_id = parts[5].strip();
-                    if gpu_name == "NVIDIA Graphics Device" and gpu_id in GPU_ID_TO_NAME:
+                    if gpu_name == "NVIDIA Graphics Device" and gpu_id in constants.GPU_ID_TO_NAME:
-                        gpu_name = GPU_ID_TO_NAME[gpu_id]
+                        gpu_name = constants.GPU_ID_TO_NAME[gpu_id]
 
                     gpu_str = f"{len(lines)-1}x {gpu_name}"
                     gpu_mem = round(int(filter_non_numeric(parts[4]).strip())/1024, 2)

lib/nvml.py

@@ -1,6 +1,7 @@
 from lib import config as config_module
 from lib import logging as logging_lib
 from lib import get_specs
+from lib import constants
 
 config = config_module.config
 log = logging_lib.log
@@ -97,6 +98,15 @@ def init(gpu_specs_file=None, allow_hive_binaries=True):
                 break
             gpu_handle = pynvml.nvmlDeviceGetHandleByIndex(i)
             gpu_uuid = pynvml.nvmlDeviceGetUUID(gpu_handle)
+
+            
+            print("name")
+            print(pynvml.nvmlDeviceGetName(gpu_handle))
+            print("device_id")
+            pci_info = pynvml.nvmlDeviceGetPciInfo(gpu_handle)
+            print(pci_info.pciDeviceId)
+            
+
             gpu_name_list.append(pynvml.nvmlDeviceGetName(gpu_handle))
             if not f"{i}-{gpu_uuid}" in parsed_specs_keys:
                 parsed_specs={}