attempt to fix memory monitor with multiple CUDA devices

This commit is contained in:
AUTOMATIC 2023-03-12 21:04:17 +03:00
parent 6033de18bf
commit a00cd8b9c1

View file

@ -23,12 +23,16 @@ class MemUsageMonitor(threading.Thread):
self.data = defaultdict(int)
try:
torch.cuda.mem_get_info()
self.cuda_mem_get_info()
torch.cuda.memory_stats(self.device)
except Exception as e: # AMD or whatever
print(f"Warning: caught exception '{e}', memory monitor disabled")
self.disabled = True
def cuda_mem_get_info(self):
index = self.device.index if self.device.index is not None else torch.cuda.current_device()
return torch.cuda.mem_get_info(index)
def run(self):
if self.disabled:
return
@ -43,10 +47,10 @@ class MemUsageMonitor(threading.Thread):
self.run_flag.clear()
continue
self.data["min_free"] = torch.cuda.mem_get_info()[0]
self.data["min_free"] = self.cuda_mem_get_info()[0]
while self.run_flag.is_set():
free, total = torch.cuda.mem_get_info() # calling with self.device errors, torch bug?
free, total = self.cuda_mem_get_info()
self.data["min_free"] = min(self.data["min_free"], free)
time.sleep(1 / self.opts.memmon_poll_rate)
@ -70,7 +74,7 @@ class MemUsageMonitor(threading.Thread):
def read(self):
if not self.disabled:
free, total = torch.cuda.mem_get_info()
free, total = self.cuda_mem_get_info()
self.data["free"] = free
self.data["total"] = total