Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 249. (Read 3426936 times)

Many thanks for all the advice Carlo, and to others for the earlier suggestions.

I definitely need to experiment more.

not in the first shot for sure...

S_tring,

Just thought of two other basic things you want to take into consideration:
1) You may want/need a "fudge" factor for some pools or algos.  As an Example, you will hardly ever earn the same amount (long term) as a pure math calculator will tell you due to orphons, coin or pool issues.  As an example you might earn 5% less on Jackpot then a calc will tell you but you will probably earn 15% less on CryptoNight.  So when doing your calculations you would multiple Jackpot * .95 and any CryptoNight coin * .85, etc  This type of thing would take into affect pool fees or your could track them separately.  You will have to assemble this info as you go but it's worth tracking and adding to your profit switching algo as you go.

2) Make sure you have very, very accurate rates for each hash algo you are going to run.  This is the reason I added the average hash rate using the last 50 submitted shares to nvMiner.  I found after adding and using this many of my hash rates were not accurate. Some high and some low but it made a difference.  If your hash rates are off then everything else will be off of course.  Garbage in = garbage out.

Hope this wasn't too much info but I'd consider these the basics of what to track and how to switch to get started.  Then you can add additional things and optimize it.

Carlo

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'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.


While your adding x12, throw in CryptoHash  

Mining the most profitable coin and how long we should mine it is a good question. As "most profitable" is only "most profitable" if said coin is exchanged at that same moment or at least at same price after it matures(if it applies) .

i wrote a "most profitable" proposal about 2 weeks ago regarding mining pools and the request to display the price when it is "most profitable" and then the price when exchanged to reflect transparency of the value of said coins as "most profitable".
The post can be viewed and commented on here:
https://bitcointalksearch.org/topic/m.7799592

I hope this helps a little bit. Everyone has a different view of "most profitable" and what they want to achieve as a goal with mining, so trying to get the timing set for everyone to like would be a difficult task.

What model?

Gigabyte and Asus are the best price

there is only one choice for now, 750ti, more nvidia vga are coming at the end of this year...

hahaha sorry  
It was a small mistake

Here's what I used when I wanted to restart miners. It's not the best but it does work:

First you need two batch files, one is your mining .bat in a loop which looks something like this:

jpc_solo.bat: