[ANN] profit switching auto-exchanging pool - www.middlecoin.com

[ButterBread]

Also, this analysis is fundamentally flawed because mining is time-dependent. Run the miners on 6 new addresses for several days (STARTING AT THE SAME TIME) and compare those graphs. That would tell you a lot more.

The graphs are from six different miners taken at the same time. I can understand that not knowing who was mining where is a flaw in the analysis (although, why should WE be choosing where to mine @ middlecoin in the first place).

But I don't understand your reference to their timing? If the graphs are showing the same time period and you can see the begin & end situation of each miner..

[chilin_dude]

Butter read my explanation, your 0.32 balance paid out, plus 0.3 unexchanged gained through the day - meaning it willl be paid out at a later date. Add them together, 0.32+0.3 = 0.62, like the top 3 you mentioned got paid.

[hodger]

Also, this analysis is fundamentally flawed because mining is time-dependent. Run the miners on 6 new addresses for several days (STARTING AT THE SAME TIME) and compare those graphs. That would tell you a lot more.

The graphs are from six different miners taken at the same time. I can understand that not knowing who was mining where is a flaw in the analysis (although, why should WE be choosing where to mine @ middlecoin in the first place).

But I don't understand your reference to their timing? If the graphs are showing the same time period and you can see the begin & end situation of each miner..

You're right, under normal circumstances, *we* shouldn't have to choose where to mine, but my interpretation of what I've read in the last 20 or so pages is that h20 and co have had some growing pains (5000MHash increase last 7 days by the looks of it, for example, and steeper in previous couple of weeks I think) that have resulted in some stop-gap solutions on some servers (EU server was constantly on one coin for a while I think)...

Be interesting to see the results of another selection of miners once all the new servers are out of "beta" etc.

[neofatalist]

Temp cut off 99?? I was reading on the 290x (we have the same card) and it can run 94 but thats about safe levels.

Anyway, how do you run your TC so high? Mine won't let me. I set it to 12000 once and it said I needed a higher LG, which basically kills hash rate.

https://i.imgur.com/mOAPW5D.png

Soooo I noticed it said "SICK" (R9 290) a few mins ago and thought it was weird that i had never noticed it.
Then a few mins later I saw the temp drop so i checked again and now it says "DEAD".
A full restart of the rig seems to have it back running but I don't know what would have caused it? Anyone seen this before?

Either:

1) Too high overclock.
2) Too high heat.
3) Too high intensity.
4) GPU is dying.

https://i.imgur.com/mOAPW5D.png

Soooo I noticed it said "SICK" (R9 290) a few mins ago and thought it was weird that i had never noticed it.
Then a few mins later I saw the temp drop so i checked again and now it says "DEAD".
A full restart of the rig seems to have it back running but I don't know what would have caused it? Anyone seen this before?

Did you overclocked your GPU ?
This happen to me when the card voltage is too low compared to my overclocking.

It also happens around 2-3 times a month for no reason.

Well this is the config for that particular card:

Code:

 
 -I 20 \
 -w 512 \
 --gpu-threads 1 \
 --thread-concurrency 32765 \
 --gpu-engine 990 \
 --gpu-memclock 1500 \
 --gpu-powertune 11 \
 --gpu-vddc 1.025 \
 --device 3 \
 --remove-disabled \
 --temp-cutoff 99 \
 --temp-overheat 95 \
 --temp-target 85 \
 --auto-fan \
 --gpu-fan 50-85

It doesnt give any HW errors like this.
It does run hot but all 290's run hot (85c - 90c, although I have seen it go over 95c once before).
It's got it's own psu to it's self.

Hope it's not on the way out, not had it long lol.

[ButterBread]

Butter bread, look at where the unexchanged STARTED the day for miner 2,5,6 that got the good payouts. For those 3 the unexchanged started and finished the same or lower than the day.

Now look at your miner, miner 0, you started with near 0 and ended up with near 0.3. So you need to add taht 0.3 to your 0.32 you got paid out (since it will get paid at a later date). That makes 0.62
0.62 is pretty similar to miners 2,5,6 you refer to.

Thanks. The thought had occurred to me to take that into account (which is why I included a slither of the day before and the day after), but there's still a rather significant difference I think even after correcting for this.

Even if I gained 0.03 in unexchanged (didn't measure the pixels but it looks like a good guess), that gets me to 0.062 vs. 0.073 for 'comparable' miners 2, 5 and 6. Doesn't sound like much, but 0.01 btc difference is 6 eur and 16% of the net result for that day.

I can imagine a lot of settings influence the net result for the day : who's mining where, when did they switch, what settings are they using for dealing with disconnects, etc. just like an earlier poster remarked.

It's just surprising how much difference such 'small' influences make on the day's net profit

[sykal]

So I had a power failure overnight and one of my rigs had zero power when i turned it back on...It's a good 860 seasonic platinum.
I didn't have a surge protector on it so i figured it fried...

I tested both outlets with a box fan and both plugs in the outlet worked...


get home from work tonight, put all my rigs on surge protectors, and the machine powers up fine...lol

anyone ever seen anything like that before?

[kalus]

Very interesting analysis would be good to hear some expert opinions

would be informative if we had the numbers at the beginning and the end of the day for immature, exchanged, and exchanged (earning) for each of the 6 miners. then we could at least run a student's t-test with a post-hoc analysis (e.g. tukey's) to determine if there is statistical significance between the 6 experimental cohorts.  if there is statistical significance in earning differences, take the derivative and redo the t-test for significance for the slope (rate of earning).  we could at least determine if there is statistically significant deviation from the norm, or if it falls within random chance.  

[ButterBread]

get home from work tonight, put all my rigs on surge protectors, and the machine powers up fine...lol

anyone ever seen anything like that before?

Yes, after a PSU has turned itself off (due to overloading, overheating, shorting, ...), it has often come back to life after I unplugged it (and plugged it back in again of course )

Don't know why that is though..

[cdesjard09]

I think considering the definiton of the unexchanged balance may also clarify some of this. The unexchanged balance is an estimate of the value of a portfolio of altcoins you've previously mined which have not yet been exchanged.

Now say I'm Miner 2 (I'm not) and I mined on the US server all day Friday and we were mining a coin that was not autotrading (so my unexchanged was shooting up). Now imagine we changed the coin we were mining at some point on Saturday and you started mining on the US server at some point after. So now my portfolio has all these coins I mined on Friday that have not been exchanged yet, and your portfolio has none of these coins because you were mining on the EU server on Friday (for example). Now we mine all day and our balances grow at similar rates (as your graphs suggest they do, the slopes being fairly similar all things considered), but at the close of Saturday our portfolios are drastically different because I still have all these coins I mined on Friday that you don't have. Now if H2O decides its time to sell those coins (the ones mined on Friday), well my unexchanged plummets and my balance shoots up, but yours does not because he didn't sell the coins that were in your unexchanged portfolio.

To address your point about how Miners 2,5, and 6 returned to essentially the start of day unexchanged values, well thats likely a product of the how much unexchanged they accrued over the day that wasn't whatever coin was mined on Friday, and hence these coins acted exactly the same as yours, give or take some variability for differences in the altcoin composition of everyone's portfolio based on which days you've mined what servers in the past.

Now I don't know if this is how H2O approaches everyone's share of the unexchanged balance, and this assumes no daytrading of altcoins that squeeze out every last mBTC, but it seems like a potential explanation.

[Archer13]

Well this is the config for that particular card:

Code:

 
...
 --gpu-vddc 1.025 \
...

Seems that your voltage is too low. Reference voltage for R9 290 is 1.140-1.180. Try a higher value in gpu-vddc parameter.

[ButterBread]

would be informative if we had the numbers at the beginning and the end of the day for immature, exchanged, and exchanged (earning) for each of the 6 miners.

Unfortunately I don't have the exact numbers. I do have the wallet addresses for each miner, as well as the last week's graph and payouts.

Perhaps coingeek keeps some kind of history of such information?

[Kluge]

So I had a power failure overnight and one of my rigs had zero power when i turned it back on...It's a good 860 seasonic platinum.
I didn't have a surge protector on it so i figured it fried...

I tested both outlets with a box fan and both plugs in the outlet worked...


get home from work tonight, put all my rigs on surge protectors, and the machine powers up fine...lol

anyone ever seen anything like that before?

This can happen if power cycled off and on rapidly. I experience it occasionally when I'm too lazy to hold down the designated power button on a case and instead just flip the power supply off and on. I'm not fully aware of the actual cause, but I'm guessing some capacitors were able to hold a charge through the cycle and certain components relying on those capacitors failing as indication that power shut off thus don't tell the graphics cards, mobo, or something else that the power shut off, so it's refusing your command to turn it back on since it thinks it's already on.

I've also occasionally run into an issue where simply turning the power off and on with the PSU doesn't work to "cycle" it and I have to physically unplug it or switch the voltage from 120V to 220V, press the power button, then switch it back to 120V for it to power on. As they say... shit's fuckin' goofy.

[jam3zs]

Temp cut off 99?? I was reading on the 290x (we have the same card) and it can run 94 but thats about safe levels.

Anyway, how do you run your TC so high? Mine won't let me. I set it to 12000 once and it said I needed a higher LG, which basically kills hash rate.

haha! yes 99 is too high, i keep meaning to change that.

Dunno about the TC, but my card is a 290, not a 290x so maybe that makes a difference.

So I had a power failure overnight and one of my rigs had zero power when i turned it back on...It's a good 860 seasonic platinum.
I didn't have a surge protector on it so i figured it fried...

I tested both outlets with a box fan and both plugs in the outlet worked...


get home from work tonight, put all my rigs on surge protectors, and the machine powers up fine...lol

anyone ever seen anything like that before?

Yes, similar and strangely enough also last night.

Note: I did have 3 PSU's and 4 GPU's.
1 decent PSU powering the rig + GPU's 1&2.
1 shitty PSU powering GPU 3
another shitty PSU powering GPU 4.

So I woke up at 5am freezing my tits off (my rig is in my room keeping me warm and i have the window open)
The rig was on but doing fuck all. One of the shitty PSU's (only powering one card) had stopped working and this somehow caused the whole rig to fuck up. I couldn't get it to start again in any way. In the end it turned out that the USB stick with bamt on it had somehow become a bit corrupted... So i re-installed bamt and it all seems to be working again (well sort of... see above lol) with a 1 card out of action due to the shitty PSU that's not working.

Hope that made sense?

[jam3zs]

Well this is the config for that particular card:

Code:

 
...
 --gpu-vddc 1.025 \
...

Seems that your voltage is too low. Reference voltage for R9 290 is 1.140-1.180. Try a higher value in gpu-vddc parameter.

TBH I don't think that line is doing anything at all.

As I understand it, you can't adjust the voltage in Linux unless you modify the bios of the card.
This is purely a Windows parameter [citation needed]

[Phantomf]

Does this pool have any servers with diff lower then 512?  That is really way to high for lower hashing cards.

[zSprawl]

Does this pool have any servers with diff lower then 512?  That is really way to high for lower hashing cards.

It averages out over time.

[Archer13]

As I understand it, you can't adjust the voltage in Linux unless you modify the bios of the card.
This is purely a Windows parameter [citation needed]

I had exactly the same behavior, firstly SICK then DEAD on my 280X (Windows), when voltage was not in pair with engine clock (voltage too low or engine too high).
Did you tried to run it without engine and memclock parameters, just on reference (bios) clocks?

[Phantomf]

Does this pool have any servers with diff lower then 512?  That is really way to high for lower hashing cards.

It averages out over time.

I get twice as many rejects on pools with more then 256 diff, even with my 7950 that hashes at 650KH/s

[ButterBread]

I think considering the definiton of the unexchanged balance may also clarify some of this.

Thank you, that was very enlightening!

[Shaban]

Redline is hidden behind the moon and isn't going up anymore