Ran the cards one at a time adding another card until I got them all going.
First card by itself got about 7,800 kh/s. Added card 2 about 5 minutes later and it ran at 7,700 kh/s and no problems. Added third card and it got 7,700 kh/s and the cpu usage still under 35%. Added fourth card and it ran at 7,700 kh/s each and cpu usage under 45%. So everything running fine up till now. Added last card and it starts off at 6,800 kh/s and cpu uasge jumps to about 70-75%. Now the first four cards kh/s is dropping. Right now they are all about 6,500 kh/s and still falling.
So time for a new cpu?
Yes, sure sounds like it's a CPU problem. The only other test I might consider doing/trying is to take the 5th card and switch it with the 1st card and try again. I'd only do this just to prove there isn't a problem with the 5th card (doubtful).
Out of curiosity do you have another computer around you could just move one of the GPUs to until you get a new CPU? This could also allow you hash at full speed.
I'd be interested in opinions on auto profit switching.
If x11 is deemed the most profitable coin, how long should we mine it before checking which coin is best again?
I've set it to check every 60 seconds and switch if necessary. It's been running for 1 hour 10 minutes, and has switched algo 16 times. Too much? Or, maybe it's good to mix the coins so much?
I'm not sure if I should leave a larger interval between scanning for most profitable algo?
I'll share a few of the basic things I built into my profit switching algo. These are just basics but can get you thinking.
You should work up an algo to do this for you. Example:
Only check prices ever 5 minutes.
If top price is different then track but do not switch yet.
Check again in 5 minutes. If still top price then switch.
Many pools penalize you for switching so you don't want to switch unless you need to.
From what I've seen many of the API services often times return the wrong/incorrect data at times but if you run it again within a minute or two then it comes back correct.
So as part of your switching algo you could track old price and new price and make sure they are roughly close to each other or you could consider them invalid results. As an example if you pull every 5 minutes then you could track diff and exchange rate to make sure they didn't jump (within 5 minutes) to something unreasonable/unlikely.
You can also track the amount of time since your last overall switch. Since you know how profitable the new coin is going to be you could base the switch on time. For example lets say you calculated 0.15 as the btc you earn on coin X (coin currently mining) per rig for the day. Coin B (best new price) would be 0.15004 per day. Since this is only 0.00004 btc per day difference you don't want to switch right away as you'll probably loose more by jumping coins.
Little changes like the above can have you changing coins every minute or two if you don't take time into the equasion. But if you take into account the time since your last coin switch you could do something like this:
If A > B then do no switch (current coin is most profitable) DUH
if B > (A+0.01) and 15 minutes passed then switch
If B > A and 60 minutes have passed then switch
So basically use some type of time element plus rate of change into consideration to figure how often you can/should switch or hold firm.
Not to complicate things but as you develop this you also want to take into consideration how long it takes a coin to mature (how long for you to get it to an exchange) . What the trading volume is and if they exchange rate is going up, down or staying roughly even.
Nothing is worse then switching coins you are mining but can't get the coins earned onto an exchange for a hour, only to have the price tank.
I want buy a new rig,
what is the best nvidia card for nist5 and new x11,x12...
Thanks
x12
I can add it if you want
While your adding x12, throw in CryptoHash
