Topic: [ANN][Pool][Profit-Switch][Optional Auto-Exchange per Coin][Vardiff] ~ Hashcows - page 149. (Read 347329 times)

That sounds extraordinary. Which model would that be?

> Nice,  I've got Sapphire 7950s, sku 11196-19-20g,  supposedly the stock bios is voltage unlocked, I still have to verify if cgminer is actually doing adjustments for me though,  or if I have to go to other means to adjust.  

> Not that I'm a big fan of Windows,  but I'm still resistant to trying to deal with linux at the moment,  since I really don't know anything about it.

> Can't say that I agree with you on anyone being stuck at I:13 because of windows.  From what I understand, that's usually due to running an sha-256 kernel.   Perhaps I am mistaken.  I'm running a scrypt kernel since I am mining scrypt,  which allows me to go to I:20 with stock versions of cgminer.  
Changing from I:13 to anything higher will definitely show an increase in hashrate on my system. I can crank them up to I:20 and get 650 a card,  but temps, stales and rejects seem to go up and my WU-to-hashrate percentage is lower as compared to I:18.

> I had one instance tonight where cgwatcher shut down cgminer due to connectivity issues, and despite me having it set to restart cgminer when the accepted shares fell off after a few minutes,  it did not restart it for some reason so I had to make one change in the settings for it to re-start cgminer unless it is stopped within cgwatcher. Then it restarted cgminer as soon as I checked that box, but I'm interested to see if it will fail again and now restart on its own without any further help from me. I'm going to get cgwatcher set so that I know that it runs cgminer smoothly,  and then I'm going to experiment with that earlier version of cgminer without cgwatcher to see if adding load balancing and/or fail overs can maintain smooth performance on its own.

> Anyone have any insight on those tagcoins?  Seems like they get mined a lot and never mature before they fall off of the list....do we get credit for those before they mature,  or are they piling up until they mature and get traded?  Would like to see a record of immature coins after they fall off of the list if we are not credited for them prior to that happening.  I might just start collecting them if we are not getting credit for them and can't see a record of them either.

Wanted to do a little test but you can be sure you probably won't see it again.

Why is hashcows mining csc?

ANC was never removed.  The ones we are mining are shooting up similarly though  

Pretty crazy, almost every single alt is at least x2.  Those who kept the coins are probably pretty happy about now!

Damn all the ALT Coins just shot up like 500-5000%.  I really hope we have been mining them...
Please start adding htem back in... ANC up to almost 0.003, GLD up to 0.001, and a shit load more... just check the exchanges now.

Yes, I do... You are running Winblowz.  Therefore, you are stuck with an max intensity of 13 and not 14-20.

Try BAMT and see your results.

As an example, I get speeds from 640-730kh/sec on my 7950 and one 7970. 

BTW: Go Sapphire all the way.  Solid as a rock at 0.962v and getting 730kh/sec out of it.

Actually the Failover 5 minute delay before switchback "Feature" was introduced in 3.5.1. So you can safely use 3.5.0 and earlier... Though my preference is 3.2.1.   It seems to have the best scrypt hashrate and stability for me.

Thank you for the update 

To the pool operator.  MEC is now at 25 coins per block.. Not sure if your coin calculations are taking that into effect.. Just a heads up..

Anything past 3.4.3 you run into the issue of the cgminer throwing a 5 minute cooldown on disconnected pools..  This doesn't affect me because I have a load balance specified.. One pool goes down, the other picks up the slack.  Failover means when one pool crashes it swaps to the other until your original pool comes back up.  All cgminer versions have these options.(there are actually 5)  I did notice your thread concurency was higher than mine, but the fact you're still hashing means you're not above what the cards can handle.. If you're not running a failover pool I highly encourage you to do so..  Downtime is lost money... Even if you're mining something a bit less desireable.

I don't use the config files.. I prefer the hands on Dos 5.0 batch files, but whatever you can do in the config you can do in the batch file so no worries there.

-Yep,   I went with powered risers.

-I don't think that my config is an issue,  short of not having any load-balancing/failovers set...

-I had tried to run Afterburner and could not get an active voltage slider with it.


When the problem does occur,  it does seem to be when the pool is changing coins.

Funny you mention downgrading down to cgminer to 3.4.3,  because I downloaded the newest cgminer that will still run scrypt with the intent of trying fail over and load balancing settings because I thought that the newer versions were better for avoiding that delay between working. Are you saying that only 3.4.3 does this?  (Avoids the delay?)

I also seem to remember reading about someone having issues with the either the failover or the load balancing quota not catching some errors and resetting since the hash rates seem to freeze,  idk. I'm sort of in information overload due to searching too much.  

I'm definitely going to try everything and experiment though.  For now,  I'm going to see how cgwatcher handles things,  since I've already set it up and it's running currently.  If it seems to work,  then I know that I can use it if necessary.  However, I would prefer to keep things simple and it would be nice if I could get cgminer to handle the issue on it's own,  so I will go back and experiment with the load balancing settings as well.  

To set up the load balancing, do I have to add anything else to the config in addition to the "quota" : "#; ..." in the pools list or is that it?  

Would it make sense / is it possible to load balance between the same pool,  to trick it into reconnecting and staying on the same pool,  and then having a different pool set as a fail over if the first pool actually goes down?

I'll try to figure this out on my own as well,   but feel free to answer if you know. =)

I'm guessing that it's bad to heat-cycle the gpus and psu too rapidly or too often? Is that why you mention protecting them from stop-start cycles?  Are you worried about heat cycles or something else like power surges or something?

forcefedvr6, I'd bet your issues have to do with a connection reset by hashco.ws when it switches coins. This interferes with cgminer, for whatever reason

The way to work around this, is to load balance to other pool, with a quota. For example:

{
"pools" :
[
   {
      "quota" : "1;stratum+tcp://pool1.us.multipool.us:3342",
      "user" : "Wipeout2097.1",
      "pass" : "x"
   },
   {
      "quota" : "20;stratum+tcp://stratum01.hashco.ws:8888",
      "user" : "Wipeout2097.1",
      "pass" : "x"
   }
]


Above, I'm giving ~96% hash rate to the cows and the rest to some coin. The 1 and 20 after "quota" are arbitrary, but the secondary pool must still have "enough" quota. When hashco.ws disconnects briefly, cgminer handles it gracefully and requests work from the other pool. When it detects the connection becoming alive, it resumes asking most of the work from hashco.ws

The setup above also protects your video cards and psu from stop/start cycles

Then, you should downgrade cgminer. Newer versions resume work from interrupted pools after 5 minutes, while for example cgminer 3.4.3 does so almost immediately

I'll post my bat file for you.. This is stable with 3 7950 msi twin frozr 3's.  I use msi afterburner to set my voltages and overclocks as well as in cgminer.  I hope this helps you out.  If it does, mine for me for 15 minutes


setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
C:\cgmine372\cgminer.exe --scrypt -I 19 --temp-target 84 --no-submit-stale --auto-fan -g 1 -w

256 --thread-concurrency 21712 --gpu-engine 1030 --gpu-memclock 1250 --gpu-vddc 1.050 -o

stratum+tcp://stratum01.hashco.ws:8888 -u merlinjr.1 -p x --load-balance -o stratum

+tcp://pool1.us.multipool.us:7777 -u spiffy_1.1 -p x

since you have six cards I'm assuming you have adequate power to the motherboard or powered risers as well.
AMD catalyst 12.8 and cgminer 3.72 in use.

Highly recommend cgwatcher version 1.3.1.1 as the new version gives me grief. lol

 I don't think that it is causing my issue though,  mainly because I'm not entirely convinced that cgminer is effectively undervolting my cards despite what the one guy that I found who posted about it claims to have seen.  I'm going to pick up a kill-a-watt and do some experimenting with changing the voltages through cgminer.  If it does not seem to make a difference,  I'll try running trixx again,  although I tried trixx when I first started running my rig and seemed to be having trouble with it maintaining my profile with lower voltages.

My issue seems to have to do with connectivity and cgminer getting stuck when there is a problem with that,  just based on what I have seen so far. 

That IS true.

That may be true,  but in my case I think that this is an issue with cgminer disliking connectivity issues, possibly due to the distance between me and the server.  You can see that I have issues with GF and RF errors when I'm hooked up to Hashcows.  I will note that I did not see these GF and RF errors,  nor did I have these connectivity issues when hooked up to Middlecoin.  

I did some searching based on the "stratum connection to pool 0 interrupted" message and found this thread:

https://bitcointalksearch.org/topic/cgminer-stratum-connection-to-pool-0-interrupted-issue-188533

and decided that I should start running cgwatcher and set it to restart the miner if the accepted shares fall flat after a couple of minutes.  

I had to slightly modify my config as compared to what I posted above as cgwatcher did not like the
"--scrypt" that I had typed in the URL.  

I also set cgwatcher to run automatically, and for it to start the miner automatically on powerup, so I took the cgminer batch file that I had in my startup folder out of there.

So we will see if the cgwatcher helps now.

I guess that I lost three or four hours of mining dicking around with this tonight,  so today will probably be somewhat of a wash for my comparison,  but I do have two full days recorded so far,  after washing the first day out due to starting a bit late and not having any immatures built up.   I want to go for a week before posting any results.  As long I can get this thing running smoothly here on hashcows,  and as long as the guy who I am watching over there on middlecoin,  or someone with identical power to him (and me) keeps running over there,  I'll be able to continue to run my comparison with some degree of accuracy,  and I will report back with some info.

Undervolting too far can cause stability issues as well.

I hope that my most recent adjustment of lowering my thread concurrency,  on top of lowering my "I' settings, will allow my rig to run 24/7 without issue on hashcows.  

Interestingly enough,  I did see a very slight bump in hash rate after dropping my thread concurrency.

Thinking back, when I was running at I:20, that not only increased stales and temps,  but IIRC it caused a full lockup after a few hours, at some point along the way.

My cards will undervolt,  and I have them set to run at .900 through cgminer.  Cgminer does not report the voltages accurately,  but I did some internet research and found a post somewhere out there stating that despite the reporting inaccuracy,  it will actually adjust the voltages.  I intend to buy a kill-a-watt for testing so I can go through each card and adjust and see for myself in case that info is not correct for my version,  since I upgraded to the newest that will run scrypt (3.7.2 for windows).

I have played around with the engine and memory clocks (using a .7 to 1 ratio) until I found the "best" hash-rates for my cards,  although I may go through them again and see if I can fine-tune that some more, since I can't remember what intensity or thread concurrency I was at when I set them last,  and also I was making changes in 50 Hz whacks.

My temps are pretty good,  right now I'm seeing 68 to 70,  and even when I had the "I" cranked up,  I was seeing 75 to 77.  All of my cards are on risers in a milk crate and each has dual fans built-in,  and I also have three more big 5" a.c. computer style fans controlled by a rheostat blowing down into the cards.

Noob Question to the pool runners about immatures...do we get paid or not get paid on those until they mature?  If we wait until they mature then I would definitely want to see a record of all immatures,  beyond the 30 round ledger.  I always see those tag coins fall off of the list and have yet to see one mature before doing so. Seeing a full record of coins like that would be nice, since as is stands,  we have no simple way of knowing what happens to them once they disappear off of that list. I guess that I could collect them and exchange them myself,  but I am trying to keep things simple,  at least for now.


EDIT: Well the problem occurred again.  I brought up the task manager and then tried to open a browser and locked up the rig so I took a couple of pics with my camera.  Not sure if this will help out,  but maybe someone can tell me what is going on somehow.





I noticed that the CPU was actually pegged at 99% while cgminer was still producing messages,  which I did not notice before.   The last message before cgminer stopped producing messages was "stratum connection to pool 0 interrupted",  although I still had an internet connection to my netbook, and I had not locked up the computer yet.
It's possible that the wired internet connection that is connected to the miner was messed up somehow,  so I recycled my router before rebooting and connecting to the pool.



Hmm,  this is frustrating me.  If anyone has any insight on what could be going on,  that would be great.

EDIT:  Figured that I would post up my most recent config:
(I know gpudyninterval does not do anything for me,  I just haven't deleted it )

{
"pools" : [
   {
      "url" : "--scrypt stratum+tcp://stratum01.hashco.ws:8888",
      "user" : "xxx",
      "pass" : "xxx"
   }
]
,
"intensity" : "18,18,18,18,18,18",
"worksize" : "256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "22400",
"shaders" : "0",
"gpu-engine" : "1050",
"gpu-fan" : "0-85",
"gpu-memclock" : "1500",
"gpu-memdiff" : "0",
"gpu-powertune" : "20",
"gpu-vddc" : ".900,.900,.900,.900,.900,.900",
"temp-cutoff" : "95",
"temp-overheat" : "85",
"temp-target" : "70",
"api-mcast-port" : "4028",
"api-port" : "4028",
"auto-fan" : true,
"gpu-dyninterval" : "10",
"gpu-platform" : "1",
"gpu-threads" : "1",
"scrypt" : true,
"hotplug" : "5",
"log" : "5",
"no-pool-disable" : true,
"queue" : "0",
"scan-time" : "2",
"expiry" : "2",
"temp-hysteresis" : "2",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}

Hi All,

I seem to be getting a huge number of rejects / stales. Are there any backup stratum servers available?Could I be doing something wrong?