OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.1

[-ck]

I hope I didn't miss anything, but isn't it impossible to see the worksize if you let cgminer choose it?

--verbose -D

[Diapolo]

Con, may I suggest a change to CGMINER code, which I don't want to do by myself, as I don't know how you would prefer it to be implemented. It would be a good thing to have one place in ocl.c, where optimal defaults for kernels could be placed.

Something like this (only example values here), which is used, if the user does not specify any settings. It could also include some detections for SDK versions, as this has impact on optimal parameters, too. Should be a rather small change, but removes the need for special solutions, like the one where you forced -w 64 for poclbm on Tahiti (which seems to be the default for all kernels on Tahiti, which could be counterproductive).

Code:

switch (clState->chosen_kernel) {
	case KL_POCLBM:
		if (strstr(name, "Tahiti") && strstr(vbuff, "844.4")) {
			clState->preferred_vwidth = 1;
			clState->work_size = 64;
		}
		else if (strstr(name, "Tahiti") && strstr(vbuff, "851.4")) {
			clState->preferred_vwidth = 1;
			clState->work_size = 128;
		}
	case KL_DIAKGCN:
		if (strstr(name, "Tahiti")) {
			clState->preferred_vwidth = 2;
			clState->work_size = 256;
		}
		else if (strstr(name, "Cypress")) {
			clState->preferred_vwidth = 8;
			clState->work_size = 256;
		}
}

Dia

[-ck]

That overrides any user specified values. The defaults are fine for now. New SDKs and if any faster kernels come up, I will change them.

[Diapolo]

That overrides any user specified values. The defaults are fine for now. New SDKs and if any faster kernels come up, I will change them.

As I said it should not override user-defaults (and I did not include any check for this, as my code was just an example ^^), but offer one place for kernel-specific settings ... well if you don't like the idea okay.

Code:

switch (clState->chosen_kernel) {
	case KL_POCLBM:
		if (strstr(name, "Tahiti") && strstr(vbuff, "844.4")) {
			if (!opt_vectors) clState->preferred_vwidth = 1;
			if (!opt_worksize) clState->work_size = 64;
		}
	case KL_DIAKGCN:
		if (strstr(name, "Tahiti")) {
			if (!opt_vectors) clState->preferred_vwidth = 2;
			if (!opt_worksize) clState->work_size = 256;
		}
}

Dia

[lituar]

Guys i have a problem with cgminer 2.2.6/2.2.7 and my 6950. I am running catalyst 11.6 and the amd sdk 2.4. But when i try to ./configure, cgminer returns "OpenCL not found". And the binary dont initialize.

Someone knows how do i fix this?

[cablepair]

ckvolias

2.2.7 has some serious problems in windows if pool 0 is not live

takes a very long time just to start and then it starts to show gpus as sick when they are not
does not happen if the pool 0 is live

this may have been the case in 2.2.6 and not noticed it , have a feeling it has something to do with your new pool probing setup

[DeathAndTaxes]

What is the best/preferred way to do log rotation with cgminer? I'm finding the logs pretty big and would like to manage them automatically.

Since all other functionality of cgminer can be controlled via config file could conman consider adding some log configs.

Code:

"log" : "off(default)|static|dynamic"  dynamic appends timestamp to filename
"logname" : "./cgminer.log" (default) | C:\logs\blahblahblah | /etc/logs/cgminer/cg-log"  
"logrotate" : "100"  max size of file (bytes, MB, GB, lines, etc)

If there is enough interest from both developers and users maybe a bounty is in order?

[DeathAndTaxes]

ckvolias

2.2.7 has some serious problems in windows if pool 0 is not live

takes a very long time just to start and then it starts to show gpus as sick when they are not
does not happen if the pool 0 is live

this may have been the case in 2.2.6 and not noticed it , have a feeling it has something to do with your new pool probing setup

+1 I had a similar issue:

Not sure if it occurred on other versions but I had a network cable become unplugged (I swear it wasn't me) it seemed to hang forever without reporting connectivity issues.

[matthewh3]

Sorry if this has already been asked or shown but how do you install CGMINER on Ubuntu 

[_Vince_]

Could you guys help me with cron and cgminer? I need a sample script that can use to start cgminer from cron

I tried many different methods but could not make it; cgminer just did not run and did not create any log output

The same script was used to start phoenix without any problem! What a headache for me

[tnkflx]

Could you guys help me with cron and cgminer? I need a sample script that can use to start cgminer from cron

I tried many different methods but could not make it; cgminer just did not run and did not create any log output

The same script was used to start phoenix without any problem! What a headache for me

Why do you need a cron script to start cgminer? Just add it to your startup scripts. Which distribution are you using?

[DeathAndTaxes]

kano would it be possible to include a secret key for RPC in the config.

something like:

Code:

"api-key" : "this-is-my-secret-it-must-be-included-in-all-API-calls-when-active".

I always hate when people use user/password when there is only one possible username so I think key works fine.

I would like to publicly host my monitoring webpages along with database for historical performance, etc.  For privacy reasons I would prefer to put this on a webserver unrelated to my physical location/IP.  

Also ANUBIS and "cgminer web monitor" seem to be popular but can be tough for newbies to setup.  With breaking version changes staying up to date can add to workload.  The authors could provide public hosting (as a paid service) but some level of security would be a good idea before punching holes in firewalls. IP Address limiting works but is harder to manage.  

I don't image this being very difficult to implement:
a) api-key is off by default.
b) if api-key is set then any RPC call needs to include the key as a JSON parameter or the call fails.

Your thoughts?

[_Vince_]

Could you guys help me with cron and cgminer? I need a sample script that can use to start cgminer from cron

I tried many different methods but could not make it; cgminer just did not run and did not create any log output

The same script was used to start phoenix without any problem! What a headache for me

Why do you need a cron script to start cgminer? Just add it to your startup scripts. Which distribution are you using?

This machine I also need to use it for something else, not a dedicated miner.

I am using a minimal version of Ubuntu Natty; without xdm,gdm or gnome. I installed xfce4 and startx directly from rc.local

The script was running fine if I run it from terminal; but not from cron

What kind of env does cgminer need?

I set DISPLAY already (although not needed if I don't want monitoring functions); execute cgminer from full path. What else?

[kano]

kano would it be possible to include a secret key for RPC in the config.

something like:

Code:

"api-key" : "this-is-my-secret-it-must-be-included-in-all-API-calls-when-active".

I always hate when people use user/password when there is only one possible username so I think key works fine.

I would like to publicly host my monitoring webpages along with database for historical performance, etc.  For privacy reasons I would prefer to put this on a webserver unrelated to my physical location/IP.  

Also ANUBIS and "cgminer web monitor" seem to be popular but can be tough for newbies to setup.  With breaking version changes staying up to date can add to workload.  The authors could provide public hosting (as a paid service) but some level of security would be a good idea before punching holes in firewalls. IP Address limiting works but is harder to manage.  

I don't image this being very difficult to implement:
a) api-key is off by default.
b) if api-key is set then any RPC call needs to include the key as a JSON parameter or the call fails.

Your thoughts?

I have already put up a pull request a while ago (4 hours) which is somewhat related to this

Basically it restricts all RPC API access to only be allowed to run commands that return data.
If you use the --api-allow command you can put W: in front of any IP/subnets that you want to have privileged access (Write access)

This means that you can put something like W:MYIP,0/0 on --api-allow and then you can let anyone read it (that you tell) but only MYIP is allowed to send commands like 'switchpool', 'gpuintensity', 'save', 'quit' (i.e. all the commands that change something)
Of course  you can have as many W:IP and W:IP/sub as you like  - then put 0/0 on the end.
The first match decides the access.

To have true password access, I would implement a secure protocol in the API.
Passing clear text passwords/keys is rather pointless.

Yes I have been thinking about this for a few weeks and decided to implement the simplest of the two options already today
We'll see if ckolivas thinks it's OK or not tomorrow when he wakes up.

[vapourminer]

ck, youre a machine

I miss the donation bit as it was a "set-and-forget" option but I understand the reasons for dropping it.

of course now i have to actually pay attention and donate manually again. *grumble*workworkwork* *sigh*

kidding about the grumbling of course.

sent 1.221122 BTC your way.

[urlord]

Hi, I'm using the linux-usb-cgminer instructions and just tried 2.2.7 , seems to be working fine,  Should I know upgrade from 2.4 SDK?

[P_Shep]

Kano: API issue...

Setting intensity to 'D' doesn't seem to work.
Can set other values fine 0, 1, 2 etc., but when D is sent it reverts to 0.

[urlord]

try lower case d?

[P_Shep]

Naa, I got it... Looked at the code. API only checks the first character so doesn't like a space between ',' and the 'D'. This doesn't effect numbers.

[Raize]

ckvolias

2.2.7 has some serious problems in windows if pool 0 is not live

takes a very long time just to start and then it starts to show gpus as sick when they are not
does not happen if the pool 0 is live

this may have been the case in 2.2.6 and not noticed it , have a feeling it has something to do with your new pool probing setup

+1 I had a similar issue:

Not sure if it occurred on other versions but I had a network cable become unplugged (I swear it wasn't me) it seemed to hang forever without reporting connectivity issues.

Ditto. I have experienced this as well. Most of us that have p2pool set as our primary pool will not actually have it up and running when we are troubleshooting issues with cgminer. This problem explains the lock ups I was experiencing now... I believe I'm using 2.2.6, though, so this must have been a recent-developed bug.