mattxia

serving deepfm performance issue

4

Machine is 72 core CPU, 384 GB memory, Linux OS 1.Openblas_threads set to 36 2.deepfm input one instance with 61 fields , 5 threads , we get only 4ms latency...

serving deepfm error under high performance (fc_layer matrix null)

4

1.using 400 concurrent 2.by gRPC 3.default deepfm model and also some production model(61 field, 9xxx dim) 4.some error occured 5.and we stepped in and find some times fc_layer is null...

serving one model multiple versions error

1

servableHandle = {ServableHandle@9593} "ServableHandle(UntypedServableHandle({name: lr, version: 6},com.tencent.angel.serving.core.SimpleLoader@7849b420))" untypedHandle = {UntypedServableHandle@9611} "UntypedServableHandle({name: lr, version: 6},com.tencent.angel.serving.core.SimpleLoader@7849b420)" id = {ServableId@9613} "{name: lr, version: 6}" **name = "lr" version = 6** loader = {SimpleLoader@9614}...