Tune OV for latency (#20160)

This commit is contained in:
Nicolas Mowen 2025-09-21 17:52:04 -06:00 committed by GitHub
parent 0e61d3f153
commit e4d5f1f94e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -195,6 +195,9 @@ class OpenVINOModelRunner(BaseModelRunner):
# Apply performance optimization # Apply performance optimization
self.ov_core.set_property(device, {"PERF_COUNT": "NO"}) self.ov_core.set_property(device, {"PERF_COUNT": "NO"})
if device in ["GPU", "AUTO"]:
self.ov_core.set_property(device, {"PERFORMANCE_HINT": "LATENCY"})
# Compile model # Compile model
self.compiled_model = self.ov_core.compile_model( self.compiled_model = self.ov_core.compile_model(
model=model_path, device_name=device model=model_path, device_name=device