What happened? Difficulty is increasing rapidly but hashrate dropped. I expected that every block would have 30 seconds average to be found and to achieve such time, correct difficulty would be computed after each previous block. Now it seems to be one block per hour if not more...

11:28:15

{
"blocks" : 245394,
"currentblocksize" : 2715,
"currentblocktx" : 3,
"difficulty" : 22.27329694,
"errors" : "",
"generate" : false,
"genproclimit" : -1,
"hashespersec" : 0,
"networkhashps" : 10531715441,
"pooledtx" : 238,
"testnet" : false
}


13:22:39

getmininginfo


13:22:39

{
"blocks" : 245395,
"currentblocksize" : 2715,
"currentblocktx" : 3,
"difficulty" : 73.56347644,
"errors" : "",
"generate" : false,
"genproclimit" : -1,
"hashespersec" : 0,
"networkhashps" : 4824319513,
"pooledtx" : 1543,
"testnet" : false
}

# infinitecoind getmininginfo
{
    "blocks" : 244999,
    "currentblocksize" : 7365,
    "currentblocktx" : 13,
    "difficulty" : 1.08697816,
    "errors" : "",
    "generate" : false,
    "genproclimit" : -1,
    "hashespersec" : 0,
    "networkhashps" : 162242767,
    "pooledtx" : 19,
    "testnet" : false
}

are we forked or what

yep, we were running the latest version. though re-downloading the blockchain fixed the fork, but our blocks worth of ~25M IFC were orphaned

It's fantastic
From 72 to 1.7 within one hour
Good test.
Very impressive!

Every client Ive seen can be slow and/or very off when reporting the network's hashrate. Thus very unreliable.

There is only 21 mhash showing here. https://ifc.coin.lu/index.php

Sad to see the coin hoppers 'forkin' you guys some more.

Nice 'fix'.

i think each block diff change should have some limit, like maximum 50% jump per adjust

yeh, something wrong

The algorithm is not his, it's the same as before, using the last 120 blocks to determine the next diff. The only difference is that instead of doing it every 120 blocks, now it is doing it every block. But the base is still last 120 blocks, I'd expect fluctuations should be smoothed out, but apparently it is not the case.

Do a diff in the code, there's no change in the algorithm itself, except diff retarget at every block instead of every 120 blocks.

Thanks monocolor for the suggestions. I will look into them. Yes it looks like the linear retarget format not working fine, although we used a pretty long block interval (120 blocks) to determine the diff. It oscillates now with a few high diff blocks every 120 blocks or so. But at least the overall block time is averaged at the design time, and there are just a few high diff blocks that won't block for long. Before it can be stuck for a day in the high diff zone (for 120 blocks).

This is similar to the Galaxycoin, which also used a linear formula with fewer blocks used.

Let's observe the behavior for a while, if the community thinks we need to move to PPCoin retarget algorithm, which seems smoother, we could do it. But I don't want to release too many patches. So let's watch for a while first.

you are in an equal position and can do the same, so no complain.

The problem is when at lower diff it's like instamine so most people will get orphans.