You can't really compare GPU cores between between Nvidia, AMD and Intel.
Each vendor defines a "core" in a different way and uses them differently.
Because Intel has them closely integrated in the CPU and makes them also subject to CPU technologies like dynamic performance scaling (aka "SpeedStep") and thermal throttling (of the CPU) it is really hard to compare to Nvidia and AMD where the GPU cores are far more independent.
Additionally, to further boost performance, Intel chipset graphics can off-load some operations to greater or lesser extend to the CPU cores when those have unused capacity. So software may believe it is running 100% on the GPU but could actually only be doing 80% on the GPU and 20% on the CPU. (Or whatever other distribution that is most efficient at that particular moment.)
Intel is also quite vague on specifications: They call the GPU cores "Execution Units" or EU's for short and usually Intel doesn't specify how many a given CPU has.
When you download the datasheets from the Intel ARK website (ark.intel.com) you really need to dig for this kind on info. E.g. For the 6th Gen Core-I CPU's Intel just says "there is a maximum of 72 EU's".
There is probably a way to query this via the Intel graphics driver, but, as I said above, question is whether or not those cores are actually usable in full.
How exactly an Intel EU compares to a Nvidia SM is in my opinion impossible to tell.
SM and EU aren't exactly the same thing to start with and I'm not even sure the notion of "GPU utilization" makes sense for Intel chipset graphics anyway, because of the dynamic way Intel CPU/GPU combo's can redistribute tasks between the components of the CPU/GPU.