You are correct that there are certainly multiple possibilities to explain the cycle. But, your examples, like your "few cpus and gpus at school" are likely too rare and diminutive to account for the roughly 6 or 7 Mhash difference between the "base" hashrate of ~20 Mhash and the peak of ~27 Mh.
Maths:
If a rather decent CPU or GPU can hash ~300 h/sec, let's call it 333.33 h/s for simplicity. So, 3 processors can produce 1 khash. So, 30 proc's produce 10kh, 300 proc's produce 100kh, and it requires roughly 3000 processors to produce 1 Mhash.
With my (optimistic) example of 333.33 h/s per processor, roughly 20,000 CPU's and GPU's would be required to account for the 6 to 7 Mhash difference. That is not going to come from a few students' educational rigs.
(multi-core server CPU's can hash more, but 333 h/s is roughly equal to a mid-range 8-core Xeon)
Please correct me if I am wrong.
Maths:
If a rather decent CPU or GPU can hash ~300 h/sec, let's call it 333.33 h/s for simplicity. So, 3 processors can produce 1 khash. So, 30 proc's produce 10kh, 300 proc's produce 100kh, and it requires roughly 3000 processors to produce 1 Mhash.
With my (optimistic) example of 333.33 h/s per processor, roughly 20,000 CPU's and GPU's would be required to account for the 6 to 7 Mhash difference. That is not going to come from a few students' educational rigs.
(multi-core server CPU's can hash more, but 333 h/s is roughly equal to a mid-range 8-core Xeon)
Please correct me if I am wrong.
I'm not saying you're wrong, and I can't explain the recent jump up to 27-28 Mh/s, but I would note that it's been as high as 30 Mh/s in the past. One explanation I've heard for the daily oscillations in the hashrate is that someone is mining with a rather large botnet concentrated in one geographic area, so the times that you see the network hashrate drop down every day correlate with the night time of whatever geographic location this botnet is in, so when botnetted people are turning their computers off for the night, that's when hashrate dips down.
