Geekbench ML has been the go - to programme for benchmarking the AI potentiality of central processor for most .
However , that is go to transfer with Geekbench AI , a young benchmarking dick let go of by Primate Labs to accurately fix how adept the mainframe is for AI work load .
standardized to the one-time Geekbench ML dick , Geekbench AI let substance abuser totest the CPU , GPU , and NPU on several AI task .
However , the adept matter is that Geekbench AI now patronise several novel fabric that were not plunk for sooner .
diving event into Geekbench AI
Geekbench ML has been the go - to programme for benchmarking the AI capableness of processor for most .
However , that is endure to exchange with Geekbench AI , a newfangled benchmarking instrument release by Primate Labs to accurately learn how unspoilt the CPU is for AI workload .
like to the old Geekbench ML pecker , Geekbench AI set aside substance abuser totest the CPU , GPU , and NPU on several AI labor .
However , the effective affair is that Geekbench AI now support several Modern framework that were not suffer in the first place .
The previous Geekbench AI app impart livelihood for Qualcomm QNN on Windows which intend you’ve got the option to now benchmark theSnapdragon X Elite’sNPU .
before , you could only try out its mainframe and GPU on AI workload .
aside from that , it corroborate the OpenVINO model ( extensively used by Intel ) , TensorFlow Lite , ONNX , CoreML , Samsung ENN , ArmNN , Qualcomm QNN on Android , and much more .
This was geekbench ai now also render three dissimilar type of score which let in single precision , half precision , and quantal mark .
The Single Precision scorehighlights accuracyat perform AI task ; Half Precision is probable to rack up more since it ’s run at a low preciseness , and the quantal musical score will probably show the mark using INT8 data point character which is a abject - preciseness data format .
diving event into Android
asunder from that , it back up the OpenVINO theoretical account ( extensively used by Intel ) , TensorFlow Lite , ONNX , CoreML , Samsung ENN , ArmNN , Qualcomm QNN on Android , and much more .
Geekbench AI now also show three unlike case of heaps which let in Single Precision , Half Precision , and quantal grievance .
The Single Precision scorehighlights accuracyat execute AI undertaking ; Half Precision is potential to mark more since it ’s run at a low preciseness , and the quantal mark will in all likelihood show the musical score using INT8 data point character which is a grim - preciseness formatting .
you might download and begin using the young Geekbench AI app decently forth .
This was it ’s uncommitted on windows , macos , linux , android , and ios .
