I was able to run OpenVino inference on my older CPU Intel Core i7 870 to be exact (Lynnfield, older than 5th gen). If you run into a compilation such as _mm_blendv_ps build error, you can check my blog on it.
Thanks for the input. Since I knew from you that the inference engine could run, I decided to try the "minimum work" solution. I moved a copy of the "object detect sample ssd" executable, all of its dependencies and a model file set I had created from my main development machine (a 6th gen i7) to the target machine (the 5th gen i7 that serves as the main computer in autonomous robot prototype). It ran with out any problems.
For anyone interested, the dependencies are:
MKLDNNPlugin.dll (for CPU)
Thanks Terry for sharing!
By default from version R4 builds the project inside ~\Documents\Intel folder (this is to prevent accidentally overriding the C:\Intel folder contents)
So if anyone would like to run the sample application inside the ~Documents\Intel folder you'll find that some dependencies are missing as mentioned above. You can either copy these files manually from the following directories:
C:\Intel\computer_vision_sdk_2018.4.420\inference_engine\bin\intel64\Release (Note: Either Debug or Release depending on your build)
Or set it on your PATH environment so on your CLI:
set PATH=%PATH%;C:\Intel\computer_vision_sdk_2018.4.420\inference_engine\bin\intel64\Release (Note: Either Debug or Release depending on your build)
It might also be necessary to run the batch file below to setup the initial environment:
Btw, these settings applies for Windows environment.