1movmedia/pyworker
2
0
Fork 0
Code Pull Requests Activity

removed 5 sec sleep and warmup request on load

Browse Source
This commit is contained in:
Colter Downing
2025-10-28 15:28:30 -07:00
parent 5d5bc197d7
commit 8d9ffb3a6c
1 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/backend.py
+5 -5
View File
@@ -325,10 +325,10 @@ class Backend:
with open(BENCHMARK_INDICATOR_FILE, "r") as f: with open(BENCHMARK_INDICATOR_FILE, "r") as f:
log.debug("already ran benchmark") log.debug("already ran benchmark")
# trigger model load # trigger model load
payload = self.benchmark_handler.make_benchmark_payload() # payload = self.benchmark_handler.make_benchmark_payload()
_ = await self.__call_api( # _ = await self.__call_api(
handler=self.benchmark_handler, payload=payload # handler=self.benchmark_handler, payload=payload
) # )
return float(f.readline()) return float(f.readline())
except FileNotFoundError: except FileNotFoundError:
pass pass
@@ -404,7 +404,7 @@ class Backend:
log.debug(f"ModelLoaded observed after {elapsed:.2f}s: {log_line}") log.debug(f"ModelLoaded observed after {elapsed:.2f}s: {log_line}")
# some backends need a few seconds after logging successful startup before # some backends need a few seconds after logging successful startup before
# they can begin accepting requests # they can begin accepting requests
await sleep(5) # await sleep(5)
try: try:
t_bench0 = time.time() t_bench0 = time.time()
max_throughput = await run_benchmark() max_throughput = await run_benchmark()