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

[tnkflx]

could that be a driver issue

The last stable release is 2.7.5. Maybe try an upgrade first?

Don't be an asshat.  2.7.5 is the latest release.  It is not stable for everyone.  Just because it works for you, doesn't mean that everyone else should be using it.

Wow... rough night?

The last stable release in the 2.6 series was 2.6.6. The current last stable release is 2.7.5. Maybe help Kano and Con with debugging info if the versions you are using are stable and the latest ones don't work?

[1714065837]

1714065837

[1714065837]

1714065837

[mdude77]

Using 2.7.4.

The overheat function doesn't seem to be working right.  I have it set to 85.  Twice I've found it in the low 90s and not making any effort to shut down or cool down.

M

--temp-cutoff (default is 95)

This is in my cgminer.conf.  This isn't enough?

"temp-cutoff" : "90",
"temp-overheat" : "85",
"temp-target" : "75",

M

[kano]

could that be a driver issue

The last stable release is 2.7.5. Maybe try an upgrade first?

Don't be an asshat.  2.7.5 is the latest release.  It is not stable for everyone.  Just because it works for you, doesn't mean that everyone else should be using it.

Wow... rough night?

The last stable release in the 2.6 series was 2.6.6. The current last stable release is 2.7.5. Maybe help Kano and Con with debugging info if the versions you are using are stable and the latest ones don't work?

Or if anyone is interested in additions after 2.7.5

The master branch in my git has a rather large number of commits I've worked on in the last 24 hours (and a couple before that)
The list is: debug control in the API, generic HW error handling for all devices except ztex, Diff1 information, BFL identify via a led flash, BFL throttle counter in 'notify' and a few other bit and pieces.

The pull request is here: https://github.com/ckolivas/cgminer/pull/310 that shows them all.
That is simply my master branch - and each change I've done I keep adding to that pull request.

Anyone interested in any of those additions feel free to compile my master and let me know if you have any problems.
I haven't changed the ztex code coz I don't have any ztex hardware and the original code does some weird shit dealing with HW errors - if someone has a ztex and has time to visit IRC I could add some changes in another branch and get them to test it also.

Thanks for anyone who can test this.

Edit: of course this is what I am running on my 2 rigs - BFL, and ICA+GPU

[kano]

Using 2.7.4.

The overheat function doesn't seem to be working right.  I have it set to 85.  Twice I've found it in the low 90s and not making any effort to shut down or cool down.

M

--temp-cutoff (default is 95)

This is in my cgminer.conf.  This isn't enough?

"temp-cutoff" : "90",
"temp-overheat" : "85",
"temp-target" : "75",

M

The line of code related to that specifically says if the temp is above temp-cutoff it will change the device status to RECOVERING.
That's why I suggested that.
It will show the device status as 'REST' and give a message as below.
It also requires both --auto-gpu and --auto-fan
It also requires ADL working - which you will see via the GPU showing it's temperature.

Code:

        if (engine && ga->autoengine) {
                if (temp > cgpu->cutofftemp) {
                        applog(LOG_WARNING, "Hit thermal cutoff limit on GPU %d, disabling!", gpu);
                        *denable = DEV_RECOVER;
                        newengine = ga->minspeed;

                        cgpu->device_last_not_well = time(NULL);
                        cgpu->device_not_well_reason = REASON_DEV_THERMAL_CUTOFF;
                        cgpu->dev_thermal_cutoff_count++;

Edit: of course if there is a bug and it's not working as expected in the current 2.7.5 - yep we'll need to chase it down and fix it.

[Roland68]

it's working nicely in 2.7.4 ... don't know in 2.7.5

[mdude77]

Using 2.7.4.

The overheat function doesn't seem to be working right.  I have it set to 85.  Twice I've found it in the low 90s and not making any effort to shut down or cool down.

M

--temp-cutoff (default is 95)

This is in my cgminer.conf.  This isn't enough?

"temp-cutoff" : "90",
"temp-overheat" : "85",
"temp-target" : "75",

M

Is the fan working? How many mining threads do you have running?  What is the status and the temperature displayed in the main cgminer window?

Fan's working.

Two threads.

cgminer is happily mining away, listing it at 92C when I find it.  No indication it found anything wrong.  Fan no higher than the 60% I left it at.

M

[crazyates]

Fan no higher than the 60% I left it at.

M

Why only 60% fan? You might need to bump that a little higher.

[ralree]

My reject rates have gone to almost 2% according to cgminer after going from 2.7.0 -> 2.7.5.  The weird part is my pool (bitminter) has them at 0, so that's a plus.  How can I figure out what's happening here?

cgminer seems to be reporting lots of Duplicates (many more than usual).

[mdude77]

Fan no higher than the 60% I left it at.

M

Why only 60% fan? You might need to bump that a little higher.

I don't run any of my GPUs above 60%.  All but this one do just fine between 50 and 60%.  Most run in a very cooled environment (air conditioning blowing right on them), "free" from any chasis with a box fan blowing on them.  They usually run high 60s, low 70s.  This one is in my main workstation, which has cooling issues.  Usually 60% is fine, sometimes I can do 50-55% w/o a problem.  However, twice recently I found it in the low 90s, and cgminer seemed oblivious to it, even though it clearly showed the temp on the screen in cgminer.

M

[crazyates]

Fan no higher than the 60% I left it at.

M

Why only 60% fan? You might need to bump that a little higher.

I don't run any of my GPUs above 60%.  All but this one do just fine between 50 and 60%.  Most run in a very cooled environment (air conditioning blowing right on them), "free" from any chasis with a box fan blowing on them.  They usually run high 60s, low 70s.  This one is in my main workstation, which has cooling issues.  Usually 60% is fine, sometimes I can do 50-55% w/o a problem.  However, twice recently I found it in the low 90s, and cgminer seemed oblivious to it, even though it clearly showed the temp on the screen in cgminer.

M

Well there you go. You just answered your own question. If all your caseless, air conditioned GPUs do fine with 60% fan, but your workstation GPU (which has cooling issues) isn't fine with 60%, then I say again, bump up the fan speed! "--auto-fan --gpu-fan 25-75".

[mdude77]

I don't run any of my GPUs above 60%.  All but this one do just fine between 50 and 60%.  Most run in a very cooled environment (air conditioning blowing right on them), "free" from any chasis with a box fan blowing on them.  They usually run high 60s, low 70s.  This one is in my main workstation, which has cooling issues.  Usually 60% is fine, sometimes I can do 50-55% w/o a problem.  However, twice recently I found it in the low 90s, and cgminer seemed oblivious to it, even though it clearly showed the temp on the screen in cgminer.

M

Well there you go. You just answered your own question. If all your caseless, air conditioned GPUs do fine with 60% fan, but your workstation GPU (which has cooling issues) isn't fine with 60%, then I say again, bump up the fan speed! "--auto-fan --gpu-fan 25-75".

My question wasn't why it's overheating, my question was why didn't cgminer shut down when it exceeded my threshold?  Obviously I can speed the fan up, and obviously I don't want to otherwise I would!!

M

[TheHarbinger]

I don't run any of my GPUs above 60%.  All but this one do just fine between 50 and 60%.  Most run in a very cooled environment (air conditioning blowing right on them), "free" from any chasis with a box fan blowing on them.  They usually run high 60s, low 70s.  This one is in my main workstation, which has cooling issues.  Usually 60% is fine, sometimes I can do 50-55% w/o a problem.  However, twice recently I found it in the low 90s, and cgminer seemed oblivious to it, even though it clearly showed the temp on the screen in cgminer.

M

Well there you go. You just answered your own question. If all your caseless, air conditioned GPUs do fine with 60% fan, but your workstation GPU (which has cooling issues) isn't fine with 60%, then I say again, bump up the fan speed! "--auto-fan --gpu-fan 25-75".

My question wasn't why it's overheating, my question was why didn't cgminer shut down when it exceeded my threshold?  Obviously I can speed the fan up, and obviously I don't want to otherwise I would!!

M

It's overheating because you don't have auto-fan and auto-gpu turned on.  You have told cgminer to run your GPU at a set clock speed, with a set fan speed, and that's exactly what it's doing.  It will happily continue to do so until you GPU bursts into flame. The temp settings don't mean anything if you aren't allowing cgminer to control the temperature.

Just another case of operator error.

[The00Dustin]

It's overheating because you don't have auto-fan and auto-gpu turned on.  You have told cgminer to run your GPU at a set clock speed, with a set fan speed, and that's exactly what it's doing.  It will happily continue to do so until you GPU bursts into flame. The temp settings don't mean anything if you aren't allowing cgminer to control the temperature.

Just another case of operator error.

To clarify, you need something like this:
--auto-gpu --temp-cutoff <temp> --gpu-engine <MHz> --gpu-memclock <MHz> --gpu-fan <%>

Note that all of the values in <>'s are chosen by you and can be set or ranges, so in this case, --gpu-fan could be 60 or 50-60.  I don't remember exactly what --temp-cutoff does, so you might want to set it a few degrees cooler than you think you should, but it sounds like you have plenty of room to experiment without damaging your card.

[Roland68]

using the provided example.cfg file would help ...just cut/paste auto settings into your cfg file ...

[kano]

...
It's overheating because you don't have auto-fan and auto-gpu turned on.  You have told cgminer to run your GPU at a set clock speed, with a set fan speed, and that's exactly what it's doing.  It will happily continue to do so until you GPU bursts into flame. The temp settings don't mean anything if you aren't allowing cgminer to control the temperature.

Just another case of operator error.

Made me laugh
Yes that is correct of course

[mdude77]

It's overheating because you don't have auto-fan and auto-gpu turned on.  You have told cgminer to run your GPU at a set clock speed, with a set fan speed, and that's exactly what it's doing.  It will happily continue to do so until you GPU bursts into flame. The temp settings don't mean anything if you aren't allowing cgminer to control the temperature.

Just another case of operator error.

Okay, let's start over again.

My cgminer.cfg clearly states, as I said a few posts back:

Code:

"gpu-engine" : "0-1125",
"gpu-fan" : "0-55",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "0.870",
"temp-cutoff" : "90",
"temp-overheat" : "85",
"temp-target" : "75",

Where is the problem?  Is the config file missing something, or do I need to add more parms to the command line?  Note that I didn't put any of these lines in here, cgminer wrote them when I told it to create a .cfg.  All I did was change some parms.  It's a bit misleading for these parms to be here and do nothing.  Specifically the temp-cutoff part, I would think that means, well, when it reaches that temperature, it stops dead in the water.

And, I would hope the internal shutdown on the GPU would prevent it from bursting into flame.

BTW, thanks for the sarcastic help.  Much appreciated.

M

[crazyates]

It's overheating because you don't have auto-fan and auto-gpu turned on.  You have told cgminer to run your GPU at a set clock speed, with a set fan speed, and that's exactly what it's doing.  It will happily continue to do so until you GPU bursts into flame. The temp settings don't mean anything if you aren't allowing cgminer to control the temperature.

Just another case of operator error.

Okay, let's start over again.

My cgminer.cfg clearly states, as I said a few posts back:

Code:

"gpu-engine" : "0-1125",
"gpu-fan" : "0-55",
"gpu-memdiff" : "0",
"gpu-powertune" : "0",
"gpu-vddc" : "0.870",
"temp-cutoff" : "90",
"temp-overheat" : "85",
"temp-target" : "75",

Where is the problem?  Is the config file missing something, or do I need to add more parms to the command line?  Note that I didn't put any of these lines in here, cgminer wrote them when I told it to create a .cfg.  All I did was change some parms.  It's a bit misleading for these parms to be here and do nothing.  Specifically the temp-cutoff part, I would think that means, well, when it reaches that temperature, it stops dead in the water.

And, I would hope the internal shutdown on the GPU would prevent it from bursting into flame.

BTW, thanks for the sarcastic help.  Much appreciated.

M

We're not being sarcastic. You need the --auto-gpu and --auto-fan options, and you've been told this many times! In the config, it would look like this:

Code:

"auto-gpu" : true,
"auto-fan" : true,

My grandfather has a saying: When 3 people tell you you're drunk, you better sit down.

[kano]

Yeah I mentioned that at the start also
https://bitcointalk.org/index.php?topic=28402.msg1158935#msg1158935

Though I did add the edit back then saying if there was a bug it would need following up just in case I was wrong.

I guess that comment about --auto-gpu --auto-fan has been missed each time?

[mdude77]

We're not being sarcastic. You need the --auto-gpu and --auto-fan options, and you've been told this many times! In the config, it would look like this:

Code:

"auto-gpu" : true,
"auto-fan" : true,

Alright, twice someone has said to use parms, this is the first time someone said to add it to the config file.  I'll adjust it now.

What this means is the lines in the config file, written by cgminer, are misleading, as they don't do a thing unless you tweak it.  Which is fine..

My grandfather has a saying: When 3 people tell you you're drunk, you better sit down.

If this isn't sarcasm, and "another case of operator error", what is?

M

[mdude77]

Yeah I mentioned that at the start also
https://bitcointalk.org/index.php?topic=28402.msg1158935#msg1158935

Though I did add the edit back then saying if there was a bug it would need following up just in case I was wrong.

I guess that comment about --auto-gpu --auto-fan has been missed each time?

No, I saw it, and I already changed my command line to include it.  Someone else, might have been you, to add the cutoff, which was already in the config.

What was tripping me up was the lines in the config that I didn't put there, and someone telling me to add the temp cutoff line.

Thanks for the assistance.

M

EDIT: It occurred to me after I wrote this that the auto-gpu and auto-fan lines might have been in the config file to begin with, and I might have removed them.  That of course would lead to this situation.  I don't recall doing so, but that doesn't mean it didn't happen.