I think, you did not understand what I meant. I wasn't referring to the overall hashrate of the whole clevermining-pool including all users, but the single hashrate of one user.
Why is it dropping, if you count each and every submitted share even of the previous mined coins, after the pool switched? If it was like this, the hashrate could not drop from let's say 80 MH/sec to 60 MH/sec. for some time (sometimes 5 minutes or more), right?
Then I think you don't understand what is hashrate reported by the pool. Let me explain.
The pool doesn't know your hashrate. It can only estimate it based on number of shares which your miner submits, so what the pool shows is only an estimation. You miner knows your exact hashrate because it knows how many hashes it calculated. But it only sends to the pool hashes that meet required difficulty (they're called shares) and the pool estimates your hashrate based on these submitted shares (based on knowledge than it takes X MH/s to find one Y-difficulty share per second). But number of shares which your miner finds varies wildly as mining is a kinda-random game. Sometimes you find 10 shares in a minute, sometimes you won't find even a single share in a minute. If you look at 5-minute values then your hashrate estimated based on shares submitted can jump between 60 and 80 MH/s and even more. The shorter the timespan the bigger the variance. Just like in throwing a dice. If you throw a dice three times and look at each result individually then you can get 1, 3 and 6. If you throw 30 times and take 10-throw averages then you can get 3, 3.6, 4.1. If you throw it 300 times and take 100-throw averages then you can get 3.4, 3.5, 3.6. It's the same with checking 5-minute, 1-hour and 24-hour mining averages. What you want to compare is 24-hour average, anything less is just too vulnerable to variance.
Nothing new here to me.
I'm talking about the reported hashrate over at least 24 hrs. On other pools 80 MH/sec. out of the miner means ~79 MH/sec. reported at the pool side. On your pool it means ~76-78 at best, caused by the massive break-ins (presumably when the coins switch?) and times when the hashrate drops seriously (sometimes even below 70 MH/sec. for a longer time). Wafflepool is the best example on how to use the hashrate the most effective - at least for switching pools (on P2Pools with only one coin the miners output is 82-84 MH/sec. and yields 80-82 net/pool-side hashrate, only to mention it). Aside from the problem, that they only have a reported profitability of ~72% for Scrypt-coins at the moment. But that's another story and not related to the difference between submitted and paid-for hashrate (I hope so, at least ...).
Edit: Just tested your US-server for several hrs, but the result seems to be worse (of course, presumably because of the longer latencies, but you never really know in advance and have to test it), so now switched back to your EU-one.
the hashrate could not drop from let's say 80 MH/sec to 60 MH/sec. for some time (sometimes 5 minutes or more), right
so I didn't have a chance to guess you're talking about 24h averages.
In that case you should read the following post and experiment with manual difficulty which might help:
In short: there are some ASICs with faulty chips which tend to go dead/frozen if new work is sent very frequently (and CM switches coins more aggressively than other pools). Miner's controller probably restarts them if they don't send a share within reasonable time but that reasonable time is probably based on difficulty and expected time to find a share by the chip. The pool's vardiff adjusts difficulty so that your miner find a share very 10 seconds, but if you have dozens of chips inside one miner then it means each of your chip finds a share once per 5-15 minutes. If the miner restarts dead chips after 3x this time of not producing a share, then your chips can be frozen for half an hour before being restarted. If your miner has lot of frequently faulting chips then this will result in lower hashrate (because some of your chips are not working) and hashrate going up and down randomly. This doesn't happen on single-coin pools and doesn't happen on coin-switching pools which are switching coins slower than CM. This also doesn't happen on CM with miners which don't have faulty/freezing chips. Read the above post and try manual difficulty according to what the posts recommends depending on number of chips and check if this will help. It helped most of people who had similar issue.