I'm late to the party here, don't come around here as often. But Aeon fork upcoming, I'd love to help with CPU miner settings.
Your i7 should do at least 1.1 kH/s (8MB L3)
And the i5 (6MB L3) can do 600-700 H/s probably. Also these have intel integrated graphics that are OpenCL capable and can add in a few extra 100 H/s if you set it up to do so (this isn't normally done or talked about, I figured out how to do this on my own in Debian based Linux.)
For Intel CPUs, xmrig I find to be faster (1-2%) over xmr-stak. This is highly dependant on OS and C++ compiler used to make the binaries. I use Ubuntu server (16.4 LTS) with gcc-7.3 and sometimes gcc-8 is better. Configuration is generally pretty simple, you want one thread per MB of L3. Now the i5 has 4 cores (no HTs) and thus for best performance I've found running 3 cores each at "low power mode" (double threaded) to be best. No affinity usually works best on these Intel CPUs over the long run, but that's dependent on whatever else you might be running on the machine. If it is a dedicated miner you can run all 4 cores and set the miner to run at nice -20 (highest priority.) If you use all four cores, it's best to set affinity and not to use a DE (desktop environment) stick to server terminal and SSH in for control.
I'm willing to help anybody with configurations (mostly in linux, but config/affinities apply to Windows as well, just have to do things differently and IMO it's more difficult from Windows GUI.)
You can contact me on Telegram (at)LoginRoot (t.me/loginroot) or I am often active in a CPU mining group (t.me/CPUtalk) but please don't join the group (it's not my group) just to ask me questions directly. I'm always happy to help out fellow CPU miners! Can't wait for the hard fork June 3rd (or whenever it's supposed to happen now.)
Also I wanted to add that for AMD CPUs, sometimes xmr-stak is better than xmrig... but Intel CPUs almost always better with xmrig.
I also have xmrig based ARM miner binary for android phones, dev boards, TV sticks, etc. These get low hashrate, but are the most power efficient miners I know of other than ASICs and some FPGAs. They generally get between 24-120 H/s at 3-10 watts. Aarch64/ARMv8 (64 bit ARM CPUs) are far better than armhf (32 bit ARMv7) due to CPUs having AES instructions. I get 12 H/s on monero and 24 H/s on Aeon with my Samsung Galaxy S5 (ARMv7, not the best CPU for mining, but it uses ~3 watts and only gets slightly warm.)
Intel i7-26xxM (4MB L3 laptop CPU) gets 425 H/s on Aeon and 120 H/s on Monero (v7.)
Intel Xeon E3-1270v2 (8MB L3 server CPU) gets ~1.2 kH/s on Aeon and 300 H/s on Monero(v7) @ 3.7 GHz
Intel i5-6400U (6MB L3 desktop, running Windows 10 and many other programs) got ~500 H/s on Aeon and ~140 H/s on Monero(v7) This could do way better, but it's a heavily used PC.
AMD Athlon II X2 (No AES, no cache, old desktop) got ~60 H/s on Aeon and ~28 H/s on Monero.
AMD Opteron dual socket CPU server gets ~2.2 kH/s on Aeon and currently 714 H/s on Monero(v7.)
I'll have more ARM benchmarks as I work with people who own better ARMv8 CPUs; I don't have any to test myself.
GPUs I'm starting to play with RX550 gets ~320 H/s on Monero(v7) without BIOS mod and older AMD PCIe GPUs (2GB vRAM) get around ~315 H/s on Monero(v7) no BIOS mod. Double those numbers at least for Aeon, maybe triple as I play with xmrig-amd's codebase.
Intel iGPUs (integrated HD/Iris graphics) get ~100 H/s on Monero(v7) and easily double for Aeon... I haven't been mining Aeon since ASICs took over, but I'll be back soon