CGWatcher 1.4.0, a GUI/monitor for CGMiner & BFGMiner to help minimize downtime

[milone]

If I get this correctly it switches to most profitable with restart of cgminer right? Would it be possible to do by just changing pool(same tipe of coin)? My cgminer sometimes dosen't start right but when it is running it is stable... I would like to avoid restarts

Correct. Right now you can only link coins to profiles. In order for it to switch pools based on profitability, you would have to assign a coin to each pool... something I haven't gotten to yet.

How to enable auto intensity?
It does not work on my 6970s and 5970s
What it does if I use auto-gpu then cgminer turns off overheating gpu for good.
if I don't use auto-gpu then gpu overheats and burns down
CGWatcher on one of my rig did auto adjust intensity and it was miracle, on rest of the rigs it ignores overheating problem
can't rely on it


--scrypt  -o stratum+tcp://stratum.give-me-ltc.com:3333 -I 16  --auto-fan --auto-gpu

If you enable auto-gpu, CGWatcher does not adjust intensity because cgminer/bfgminer will protect against overheat by adjusting clock and fan speeds, and if necessary disabling the device if temps exceed temp-cutoff.

If you don't enable auto-gpu, CGWatcher will automatically adjust intensity as long as the 'Overheat protection' option in the Monitor tab is enabled. When adjusting intensity, the following is performed during each monitor check (default every 10 seconds):

For each GPU that is mining...

    If the GPU's temperature has been above temp-cutoff for at least 10 seconds, disable it and mark it as disabled.

    Otherwise if the GPU's temperature has been above temp-overheat for at least 10 seconds, lower intensity by 1.

    Otherwise if the GPU's temperature has been less than temp-overheat for at least 10 seconds...

        If CGWatcher had disabled the device, re-enable it.

        Otherwise if CGWatcher had lowered intensity, raise it by 1 (unless it is already back to the original intensity)

        Otherwise do nothing.

If you haven't set temp-overheat and temp-cutoff, the default values are 85C and 95C respectively.

So all of this only occurs if auto-gpu is not enabled. I did not add this feature to replace cgminer's gpu management... I added it because CGWatcher is able to get Nvidia temperatures where cgminer isn't... so this was to give Nvidia miners at least some sort of protection, though it can be used for AMD GPUs as well. I have not tested to see which method gives better hashrates.

So if you don't have auto-gpu enabled and CGWatcher is not adjusting intensity as expected, let me know and I'll look into it... but I'll need more info. The best info would be right after CGWatcher fails to adjust intensity, going to the Tests tab and click 'Create Debug Report'. Then email me the report. This would help me see what exactly is happening, but since the report only includes the last 25 log entries, the sooner you can do it after the problem occurs, the better. You can also go to the Log tab and check the 'Show debug entries' checkbox and see if any info is given there indicating a problem.

[Lucko]

I know I could just test this...

Dustcoin has a delay in showing new difficulty(FTC for example sometimes switches beck up before it even shows decreased difficulty)... Don't know about Coinchoose. And if Coinchoose doesn't have it how do you know the correct one? If I would use Dustcoin alone I would make a lot of mistakes...

EDIT: I see that Coinchoose doesn't work for TRC for sure. Too slow... TRC is on 150% and it is still showing 96%...

[RandomQ]

Bug Report

"Restart Miner if accepted don't change for 10 mines" did not work recently for me

I think this is a bug in cgminer (3.1.1 - 3.3.1) because I have had it happen a few times on a certain pool.

What will happen is the stratum protocol will say disconnected, and will not try to reconnect and just hang

I was hoping that cgwatcher would detect my shares wasn't going up,and restart cgminer but no luck

I was able to manually within cgwatcher, stop cgminer and restart it.

I wasn't using a backup pool with this profile, so there wasn't anything to backup too



 

[milone]

Lucko:

There is not a whole lot I can do about this at the moment. I chose coinchoose because they supply data in json format. If there is a better alternative, I'm open to switching or offering a choice on where the data comes from. But only for sites that provide either json or xml or plain text, etc. If I have to get into page scraping that will further down the road. And I have no plans or time to try to capture and maintain this data myself. I'd rather use a site like coinchoose where that is the developer's  primary focus and I would imagine he does what he can to keep the data as correct and current as possible.

I know cgminer and bfgminer provide the current difficulty via API but I'm not sure how accurate it is for altcoins. Assuming it is accurate, CGWatcher could switch to the profile based on coinchoose data -> start the miner -> get the difficulty from the miner -> update its coin data -> reassess the coins using the updated difficulty. But this wouldn't be until after CGRemote is released.



RandomQ:

Without more information it is difficult to say what the problem is. The best way to resolve this is to email me a debug report right after it was supposed to restart the miner but didn't. So if you've set it to restart if accepted shares don't increase for 10 minutes and it has been 11 minutes, go to Tests tab and click Create Debug Report and email me the report.

If you go to the Report tab and scroll down to the [Debug information] section, it lists when and how long ago the last share count change occurred for accepted, rejected, stale, and total. So if you have it set to restart the miner if accepted shares don't change for 10 minutes, and it says the accepted share count last changed 11 minutes ago and it didn't restart the miner, there is a bug... but I'd still likely need the debug report to pinpoint where. I just woke up so I'll go over the code again shortly and if I find anything that needs corrected, I'll release a fix and post back here later.

[Lucko]

I complicity understand... This is more like a observation... So everyone knows that this is not 100% correct... I use pool data to calculate that but I'm working on dummy cgminer that will jump from pool to pool and get those data... Well it is more of an idea then work at the moment...

[SpAcEDeViL]

HY

The CGWatcher is the best Tool under windows! Great!

I have 2 Raspberrys, this Berrys are Mining, can you build in a "Multi"-API Watch? So i can lookup all Pi's under my windows

And a edit options, or a "switch pool" for the Pi's?   

[eurgbp2011]

CGWatcher is auto adjusting intensity incorrectly.
Intensity always goes up and down as GPU temp returns to below 85C and GPU overheats again up to 90 C
There should be some more stable auto intensity method...
Software should find particular Intensity where GPU temp is below and stick to it for some time.

Update:
testing GPUs, definitely auto intensity is not intelligent with this software
I am having ups and downs from 78C to 92C and Intensity from 10 to 18 on one GPU, it will die quickly this way, please make auto intensity much smarter than what we have now

I love auto intensity it is a wonder tool, but bring more stability to it, please!

I think this method is called "Hysteresis"

[milone]

SpAcEDeViL:

You'll have to wait for CGRemote to do this.


eurgbp2011:

Yes, the auto intensity adjustment is not "smart". I don't believe I've ever claimed it was. All it does is check the current temperature and change intensity or disable/enable the device depending on the temperature. Assuming the GPU is below temp-cutoff, it should eventually get to where it is switching back and forth between two intensity values. This is because it tries to use the highest intensity it can. So if temp-overheat is set to 85C and the original intensity is 14, it will do the following:

Code:

Each time a check is performed...

If temp has been above 85C for at least 30 seconds, lower intensity by 1

Otherwise if temp has been below 85C for at least 30 seconds, raise intensity by 1 (unless it's back at original intensity)

So it might eventually keep going between 10 and 11 for example because 11 causes the temp to exceed temp-overheat, but 10 causes it to go below temp-overheat. Since using a combination of 10 and 11 is better for your hashrate than just using 10, it will continue to go back and forth between these two.

If this is not what is happening, then there very well may be a bug that needs fixed. What are your current temp-target, temp-overheat, and temp-cutoff settings? And emailing me a debug report (go to Tests tab and click Create Debug Report) and your cgwatcher.log file would be helpful in finding any problems.

As I mentioned a few replies up, the overheat protection in CGWatcher can be used for AMD GPUs, but was added to give some protection for Nvidia GPUs, which cgminer doesn't get the temperatures for so it cannot protect. For AMD GPUs, the auto-gpu option will let cgminer manage your GPU clock and fan speeds to maintain target temperatures, and is probably a little more intelligent. I never meant for overheat protection in CGWatcher to replace this, which is why the overheat protection only works if auto-gpu is disabled. I'm not sure which program's method results in higher hashrates while keeping temps under control... I haven't done testing. But regardless of what GPUs you have, there may be some AMD users who prefer CGWatcher's method to cgminer's, so I want it to work properly for all GPUs. So if you can provide the info I suggested to help find problems I'd really appreciate it.

[eurgbp2011]

auto-gpu in cgminer does not work at all, GPU goes to 95C and cuts off.
your software is the best but auto intensity is not smart, please fix it

this is my setting

--scrypt  -o stratum+tcp://stratum.give-me-ltc.com:3333 -I 18   --auto-fan

maybe something like this?

previous_temp=0;


cycle 10 sec
{
 // hot summer day ~15:00, overheating outside 35C inside 41C, bloody host summer in London

 if (current_temp >= 85C) // current temp is getting hotter, lets say 86C
{
 
   
      if( current_temp > previous_temp ) // currrent temp is still going up, have to lower intensity again
      {
         I=I-1
         previous_temp = current_temp
      }

}

 if current_temp < 85 then exit ; // don't auto adjust to higher intensity, let it be there

// cool night 03:00 outside 18C inside 32C, current temp is lets say 75C, have to think how to increase intensity back to max possible at this moment



}


any ideas?

[timk225]

Please someone help me, this cgminer / cgwatcher program has me so pissed off right now I could just scream.  No matter how carefully I follow the directions, no matter what I do, something just isn't right.

Mining PC has a fresh install of 7 64 bit, a Radeon 6950 (just one to get this thing going, I have more) and all the proper drivers.  Driver version 12.3 as specified in the setup on the phenixcoin mining website, plus SDK 2.8.1 .

I made the cgminers' path as simple as possible, right on the C drive.  Path is: C:\cgminer-3.3.0-windows\CGminer.exe, and I have the C:\cgminer-3.3.0-windows\CGminer.bat path set for the config file.  I have even tried running just the .exe, just the .bat, both in the top line, with no change.

My CGwatcher 1.2.3.1 folder is at C:\Users\Computer\Downloads\CGWatcher-1.2.3 , and it seems to work, it just can't get the CGminer going.  

With the above path lines loaded in the miner path and config file text boxes, when I go to save those settings it'll pop up a warning box saying "An exception occurred while loading C:\cgminer-3.3.0-windows\CGminer.bat.  The config file was not loaded.  Check the config file for invalid data or corrupt Json format.  (Exception:  unexpected character encountered while parsing value s.Path", line 0 position 0.)

When I do save all this and hit the Start Mining button, the black CMD window will open for a couple seconds and then close and I'll get a box saying "I was unable to start and establish communication with CGminer.  Check that the path, arguments, and config file (if any) are correct and try again."

Well, I have tried and tried and tried, and I truly cannot imagine what is wrong, I tried to keep the paths as simple as possible, my config file text is copied right off the phenixcoin website setup article, and I just don't know what to do.  

I have had many problems with this damn cgminer POS for several coins, I have only gotten it to work once with lots of help from the pool operator, and the next day he took the pool down before I even got my main mining PC's up and running.  FML!

My config.bat file is as follows, except for my user name and password, on a Radeon 6950:

 setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS
C:\cgminer-3.3.0-windows\cgminer.exe --scrypt -o http://phenixmining.com:9500
-u MYNAMEHERE.1 -p MYPASSWORDHERE --shaders 1408 --gpu-engine 800
--gpu-memclock 1250 --intensity 9 --worksize 256 -g 2 --thread-concurrency 4096
--gpu-threads 1 --vectors 1 --expiry 1 --queue 0 --api-listen --api-allow W:127.0.0.1

Please help, I have spent over $1000 in buying mining PC's and GPUs, and it's only this goddamn CGminer POS that is messing me up now!  Sory for my swearing, I an just so extremely frustrated and pissed off at the moment.  And it only makes it worse when people say how easy it was to set up.  IT'S NOT EASY AT ALL!!!!!  BitMinter is easy!  But Bitcoin difficulty is sky high now.

[milone]

timk225:

You don't use a batch file and an exe file. You use one or the other.

First, make sure all of the miner options in your batch file are on the same line as the executable. I'm not sure if you just put them on separate lines here for readability, but your batch file should be:

Code:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS
C:\cgminer-3.3.0-windows\cgminer.exe --scrypt -o http://phenixmining.com:9500 -u MYNAMEHERE.1 -p MYPASSWORDHERE - shaders 1408 --gpu-engine 800 --gpu-memclock 1250 --intensity 9 --worksize 256 -g 2 --thread-concurrency 4096 --gpu-threads 1 --vectors 1 --expiry 1 --queue 0 --api-listen --api-allow W:127.0.0.1

In the CGWatcher Profile Manager window, clear the config file path textbox and arguments textbox. Then set the miner path textbox to your batch file. (If you browse to find your batch file, you may have to change the filter in the bottom right corner of the dialog window to "Batch and Command files".

Once you have set the batch file as the miner path, you should see the arguments from the batch file automatically loaded in the Arguments textbox. If you change any arguments here, they will be changed in the batch file when you save the profile.

So your profile manager window should say:

Code:

 Miner Path:     C:\cgminer-3.3.0-windows\CGminer.bat
 Config File:  
 Coin:              (optional)
 Arguments:      --scrypt -o http://phenixmining.com:9500 -u MYNAMEHERE.1 -p MYPASSWORDHERE - shaders 1408 --gpu-engine 800 --gpu-memclock 1250 --intensity 9 --worksize 256 -g 2 --thread-concurrency 4096 --gpu-threads 1 --vectors 1 --expiry 1 --queue 0 --api-listen --api-allow W:127.0.0.1

Click the Save button and once saved make sure the profile you just saved is set as the active profile (at the top of the Settings tab). If it isn't, select it from the drop-down list and click Save Settings. Then click the Start button.



eurgbp2011:

I will not be able to work on this until CGRemote is released. I'm not sure what the Flash file you sent me is but I'm out of town and borrowing someone's computer (I forgot my laptop charger) so I won't be opening any Flash files. I'll PM you when I am able to work on the auto adjust intensity code and will consider any suggestions on making it smarter.

[timk225]

Thanks for the advice, but it's still not entirely fixed.  Although I feel I'm making progress.  My notepad .bat file had the arguments broken into several lines just as I showed it in the post above.  I put them all back on one line and went with either .bat or .exe files in the Profile Manager window.  

When I'd save that, and have the section here as shown,

Miner Path:     C:\cgminer-3.3.0-windows\CGminer.bat
 Config File:  
 Coin:              (optional)
 Arguments:      --scrypt -o http://phenixmining.com:9500 -u MYNAMEHERE.1 -p MYPASSWORDHERE - shaders 1408 --gpu-engine 800 --gpu-memclock 1250 --intensity 9 --worksize 256 -g 2 --thread-concurrency 4096 --gpu-threads 1 --vectors 1 --expiry 1 --queue 0 --api-listen --api-allow W:127.0.0.1

It'd seem like it was going to start to work but then I'd get  an error of "I was unable to start and establish communication with CGMiner".

After the first few tries, I noticed that the list of arguments was getting cut off somehow.  In the Miner Arguments box of the Settings window, it'd only show " : W:127.0.0.1,api-listen : true,: : W:127.0.0.1,api-listen : true}}", and not the entire list of arguments.  At first it'd show what seemed to be the whole list of arguments as shown in the .bat notepad file, but all of a sudden it only shows the last part.  WTF?

EDIT - I went back into the cgminer folder and it seems like my .bat notepad file is getting changed, but by what and how?  When I freshly re-copy the .bat file into the CGminer folder and then go to the Manage Profiles window, it shows all the correct arguments.  Then upon starting CGWatcher, it'll flash the CMD black box for less than a second, and then it'll count about 10 seconds to "Waiting for bat file to launch CG Miner", then it'll change to "Waiting for Miner to initialize API", before giving me the "I was unable to start and establish communication with CGMiner" error report.

On the Monitor tab of CGWatcher, only the top box is checked.  On the Devices tab it correctly shows the single 6900 series GPU.  The Pools tab has nothing entered in the boxes.  But I am in a pool.  That's a problem, even though the pool info is in my list of .bat file arguments?

Also, I'm running Radeon 12.3 drivers as was suggested on the phenixmining page, and trying to run ATI SDK 2.8.1, but I could never get the damn SDK to properly install, it'd start to go then give "install package failure" errors over and over again.  I also have .NET4, Java 7.25 and Adobe Flash installed, so it should have everything it needs to run with.

[Tigggger]

timk225

You would be better off putting all the settings in a config file, eg make a new file in your cgminer directory called scrypt.conf, select it then use the inbuilt editor to fill all the settings in like gpu-engine etc

Then you would have
Miner Path:- C:\cgminer-3.3.0-windows\cgminer.exe
Config File:- C:\cgminer-3.3.0-windows\scrypt.conf
Coin:-
Arguments:- -o pool -u username -p password

I find this the easiest way as when you add a new coin you only need to change the last two items.

[milone]

Timk225

Are you able to launch cgminer using the batch file outside of CGWatcher?

If you copied that text right from your batch file, you need to add a - for the shaders option so it says

Code:

--shaders

instead of

Code:

- shaders

This is why the miner did not start and may be why the arguments got messed up.


Edit: Also, the intensity range for scrypt is 0 to 20, where it is -10 to 14 for sha256. So intensity 9 for scrypt mining will result in a very low hashrate. You may be aware of this, but I had created a profile using your settings to test just now and was only getting ~80Kh/s each on 7950s where I get ~550Kh/s on high intensities. Just thought I'd mention that.

[timk225]

Milone - the CGMiner would not start with or without CGwatcher.  When I get home I will try adding that "-" to the shaders command and see how it goes.  I hope it works, I figured it was probably one small thing somewhere screwing it up.  I did Copy and Paste the code right into this forum post on the mining pc.

I put the intensity down to 9 on purpose, I thought it'd be better to just get the thing running first, then I can go back and tune it and work the intensity up in steps.

I may be getting ahead of myself, but can these config files handle different GPUs in one computer?  This mining PC was mining BTC with 2 6950's and a 6870 while using BitMinter.  I couldn't run 3 6950's because the center one would run very hot (95 C), becasuse its cooling fan was drawing hot air right off the back of the 3rd 6950.  The 6870 has its fan at the rear of the GPU so it wasn't pulling in hot air as much.

In the case of a 6950/6870/6950 layout, my shaders would be 1408/1120/1408, and I think the GPU clock will be 800/850/800.  Got to check the mem clock speeds for the 6870 but I know the 6950's are 1250 Mhz.

My second mining pc, also with 3 GPUs, will be running a 7970 dual fan cooler, a 6950, and a 6770.

I hope this one little "-" is all it needs!

Tigggger - that sounds a little advanced for me, I just need to see this thing NOT error out on me first, but I'll see what I can do.

[timk225]

My mistake, the actual config batch file did have the -- in front of the shaders command.  Dammit dammit dammit, I KNEW I couldn't be lucky enough to have that fix it all!

Here's my config batch file as it is now, I changed the intensity from 9 to 11.:

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS
cgminer.exe --scrypt -o http://phenixmining.com:9500 -u MYUSERNAMEHERE.1 -p MYPASSWORDHERE --shaders 1408 --gpu-engine 800 --gpu-memclock 1250 --intensity 11 --worksize 256 -g 2 --thread-concurrency 4096 --gpu-threads 1 --vectors 1 --expiry 1 --queue 0 --api-listen --api-allow W:127.0.0.1

This 2 long lines of code you see here is one long line in the batch file.

If I try to click on the CGminer executable file that came in CGminer, it immediately gives an error that says "CGminer has stopped working".  A problem has caused the program to stop working correctly.  (No shit Dick Tracy, where'd you park the squad car?)  Windows will close the program and notify you if a solution is available."

Windows hasn't gotten back to me yet on this.

On the one test computer that DID run CGminer, I had to use Windows XP and 12.1 drivers, which have Open CL in them.  I can try that on this pc and see how it goes.

[milone]

If the miner doesn't start outside of CGWatcher, then there is not much I can do as far as CGWatcher goes.

I'd suggest opening the command window/command prompt and navigating to the cgminer folder. Then launch cgminer from the command prompt with the -D argument. This should give you more info on why it isn't working. When launching at the command prompt, you can still launch it using a batch file (just add the -D argument to the end of the other arguments).

The reason for launching from inside an already open command window is that the window won't disappear once cgminer closes, allowing you to see what it outputs.

[timk225]

Still having problems on the mining pc, but one of my other pc's is now mining PhenixCoin.

It is an older Socket 775 Asus board running a Pentium D 3.40 dual core with 4 GB of ram.  Windows XP Home, 12.1 Radeon drivers, AMD SDK 2.3 for xp 32 bit.  With a Radeon 6770 in it.  

I had it running in a digitalcoin mining pool until the bonehead pool operator took the pool down for a dumb hardware reason.  He's an idiot.

I input those 2 setx commands, closing and reopening the CMD window after each setx command.  Then in the CMD window, I type in cd \cg* to start up cgminer (it's right on the C:/, not in some sub folder), then I'll copy in the long line of code to start it running and tune it.

And it works!  But only in XP, no way in hell could I get it going on Windows 7 like this.

With the 6770 running at default clocks of 850 cpu and 1200 memory and Intensity 14, I'm in the 142-146 KHash/sec rate range.  But on the Phenixminer pool user page for me it shows anything from 20-60 Khash as my rate, I think it averages it over some amount of time.  If I raise the Intensity to 15 or more it starts lagging out badly.  Intensity 11 is still over 130 Khash.

But now back to the main mining pc.  Even when running it on a hard drive loaded with XP and all the attendant drivers and same setup as the 6770 machine, its latest trick is to tell me it doesn't see a GPU anywhere even though I know it is installed and has mined BitCoins for weeks, and I tried it in 2 of the 3 GPU slots too.

Any ideas why it suddenly doesn't see the GPU?

This is enough to piss off the pope!

I just put ads in my local Craigslist, offering to pay someone to come set these damn things up.  Anyone live within driving distance of Pittsburgh?

[Rishodi]

I've noticed that CGWatcher 1.2.4 is not distinguishing between the "Current Total Hashrate" and "Average Total Hashrate" on the status tab. Both fields always have the same value. This negates the functionality of monitoring and restarting the miner if the current hashrate falls below a specified value.

I tested this by starting CGWatcher and allowing the miner to run for a few minutes. I then disabled one of my GPUs, causing the 5-second hashrate as reported in CGMiner to drop significantly. However, CGWatcher continued to report identical values for both the current and average total hashrate. They fell slowly as the average hashrate decreased, and the miner was not restarted until the average hashrate fell below the specified value for 3 monitoring periods, even though the actual current hashrate had been below the specified value for much longer.

Edit: I don't know if it's related, but on the Devices tab, CGWatcher does not correctly report stats for any of the GPUs on the system while CGMiner is running. They are all reported as inactive, with a current and average total hashrate of 0, and accepted share count of 0.

I am using CGMiner 3.3.1.

[milone]

It's related. CGMiner doesn't report a total current hashrate via API, so in CGWatcher that value is the sum of all enabled device hashrates. If for some reason CGMiner reports a total average hashrate > 0 but all devices hashrates are 0, it instead uses the total average hashrate for this value. I originally did this for non-GPU miners back before CGWatcher had support for FPGAs and ASICs, so there would be a reasonably accurate value there (at least more accurate than 0). So the problem is in figuring out why CGWatcher thinks the GPUs are inactive.

Had you used previous versions of CGWatcher and if so, did it work correctly in them? Also, if you can email me a debug report created while the miner is running (go to Tests tab, click Create Debug Report) and possibly attach your cgwatcher.log file, that would be a good start in finding out why. It's possible it is a bug, or that the GPUs need to be manually mapped inside CGWatcher.