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

[JWU42]

Is it possible to lower the memory clocks on the 7970.  Mine keep going up every time I try and reset them lower.

7970 in linux is stuck at engine -150 -- cannot go lower.

[1714782841]

1714782841

[1714782841]

1714782841

[1714782841]

1714782841

[The00Dustin]

The lowest memory clock AB will let you set is 685.  You can lower the voltage in AB, but cgminer will run at 1.17V, no matter
what you set it using AB or other tools.

You still don't get it, do you?  cgminer doesn't do anything unless you tell it to, AB doesn't do anything to the voltage, it runs at 1.17 and other tools tell you what it's set at while cgminer tells you what it's actually at.  A new version of AB might change it in the future, and then cgminer will show that value, but cgminer still won't be doing anything to it.

[phorensic]

I found a new version of OpenCL (AMD APP, Stream SDK, whatever they want to call it) in a leaked version of a beta driver.  This driver supposedly gives Windows 8 support.  Anyway, I don't give a bleep about the driver, just the OpenCL versions.

On Catalyst 12.3 with OpenCL version "2.6" my DLL versions are:
amdocl(64).dll       10.0.831.4
opencl.dll             1.2.0.0

From this leaked driver the DLL versions are:
amdocl(64).dll       10.0.898.1
opencl.dll             1.2.1.0

Uninstalled version "2.6", deleted the DLL's manually in System32 and SysWOW64 (necessary step going between versions!), then installed this new leaked version.  Deleted my kernel bins in cgminer dir as usual, then launched...

Can't really see any change in performance.  Hopefully this gives ckolivas and the kernel devs something to chew on and progress forward, though.

Driver link:  http://jumbofiles.com/h2l4qczhjkb4
I will host just the 32 and 64bit versions of the OpenCL.msi myself for those who don't care about the whole driver:  http://www.delta9IT.com/OpenCL32.rar  http://www.delta9IT.com/OpenCL64.rar

[kano]

Um ... to state the obvious, unless it's from the AMD website - no one in their right mind should download it.

Edit: however that 898 number isn't new if I remember correctly you can get 898 from AMD?

[boozer]

You sure your system didn't just run out of memory?

Lol... well... no   But it seemed cgminer was the only thing affected.  Other things I checked on the server were still running (web server/sshd).
Its a dedicated rig, and pheonix never had any issues... I was just trying out cgminer and it lasted for about 4 hours before it died to the above message.  Its 2 gig ram with small web server...  after reboot, it appears cgminer is taking about 800 meg of virtual RAM and 400 Meg of physical RAM, is this normal for 6 gpu's?

 3804 root      20   0  748m 356m  50m S  1.7 20.1   1:29.67 cgminer

Happened again after about 5 hours.  I have a couple 2 gig RAM rigs running cgminer just fine with 3 gpu's each... its the rigs that have 6 gpus each that it occurs on after 5ish hours..  No sign of trouble right now... after 2 hours running, I have 738Meg free out of 2 Gigs.  Full logs are located at:

http://pastebin.com/8MDeRMkD

Anyone have any ideas to help me troubleshoot whats happening?

[-ck]

thanks for your work on this tool.  it is just what i was looking for as i've developed in the bitcoin arena.  having total control in one app is great.

my experience did get off to a bad start though and was about to give up.  the system i decided to try it on first would not work.  i tried the compiled windows version, then tried to compile myself - oh that was fun. 

i just could not get it to initialize the GPU so i moved it over to another system and bam worked right off.  it turns out that my first attempt was on a pc that i was messing with the Intel OpenCL SDK earlier.  i removed that SDK then CGMINER worked.  this may not be news or unexpected but figured i would share in case someone else is in the same boat.

thanks again ckolivas, and all contributing members. now off to research centralized monitoring........

You're most welcome. Enjoy.

[phorensic]

Um ... to state the obvious, unless it's from the AMD website - no one in their right mind should download it.

Edit: however that 898 number isn't new if I remember correctly you can get 898 from AMD?

Comes from Guru3d forums, which has been around for a long time. Sorry, forgot that link to close the loop on where this came from, here:  http://forums.guru3d.com/showthread.php?t=360108

[-ck]

Since I've had a few cgminer updates waiting for quite a while, and I've also added another update to the miner.php that some might consider useful, I'll post about it here:

The latest version in my git waiting to be pulled:
https://github.com/kanoi/cgminer/blob/master/miner.h

This adds 3 changes to ckolivas git:
1) Split the devs output tables if there is more than 1 device type
2) A PHP deprecated update for 1)
3) A new option at line 10 "$readonly" that when set to true will force the script to not show any buttons to change cgminer
If it is set to false then it will use the cgminer API command 'privileged' to determine if it should show buttons

Change 3) is useful to force it to read only with true if you want to give public access to the web page but ensure it always is readonly no matter what API access level you set in cgminer

Will merge all this stuff whenever my mining rig is back online and I can actually test any code I commit.

[-ck]

Is there a way to run the -na command when using diablo kernel?

No. That's a workaround for the diablo kernel when you're using it on a 2.6 SDK that performs terribly (like 25mhash) by default. The poclbm kernel is actually faster than diablo with -na so there is no point.

[-ck]

con (or anybody with 7970),

Can you undervolt your 7970 with --gpu-vddc setting?

Thanks

Nope, they never respond.

[-ck]

Con, Kano

Do you know if anything can be done in cgminer to prevent this crash in aticaldd.dll?  I keep getting these from time to time, but it is always the same offset.

-------------------------------------------------------------------------------
Faulting module name: aticaldd.dll, version: 6.14.10.1703, time stamp: 0x4f3b188e

Exception code: 0xc0000005

Fault offset: 0x001671fa

Faulting process id: 0xb04

Faulting application start time: 0x01cd02bc779abfc0

Faulting application path: C:\Users\admin\Desktop\cgminer\cgminer.exe

Faulting module path: C:\Windows\system32\aticaldd.dll

Report Id: dea2867e-6eaf-11e1-b01a-8c89a5566fbd
-------------------------------------------------------------------------------

It's a crash in the driver or sdk code, therefore I can't do a thing about it apart from recommend a different driver version.

[-ck]

Is it possible to lower the memory clocks on the 7970.  Mine keep going up every time I try and reset them lower.

You can only set the 7970 memory 150 lower than the engine clock speed. That's precisely why the memdiff feature exists in cgminer, so add this to your commands:
--gpu-memdiff -150

[-ck]

cgminer just stopped for me sometime last night... This was the start of a long set of logs at this time:

Code:

<snip>
Mar 16 04:10:40 skynet kernel: [13242.684124] cgminer invoked oom-killer: gfp_mask=0x280da, order=0, oom_adj=0
Mar 16 04:10:40 skynet kernel: [13242.684129] cgminer cpuset=/ mems_allowed=0
Mar 16 04:10:40 skynet kernel: [13242.684133] Pid: 4022, comm: cgminer Tainted: P         C 2.6.32-5-686 #1
Mar 16 04:10:40 skynet kernel: [13242.684135] Call Trace:
Mar 16 04:10:40 skynet kernel: [13242.684143]  [<c1089a74>] ? oom_kill_process+0x60/0x201
Mar 16 04:10:40 skynet kernel: [13242.684148]  [<c1089ff1>] ? __out_of_memory+0xf4/0x107
<snip>

I can post all the rest if needed, but figured I would start with the above.

Yep you really have run out of ram. The only time I've seen cgminer itself waste lots of ram is when people have the submit-stale feature enabled and it caches lots of old work and keeps trying to submit them for extensive outages, creating more and more network threads since there is no upper limit to how long it will hold onto this work and try repeatedly to send them. I guess I should make them all time out and give up in a future version. I don't ever suggest using the submit-stale feature and I also recommend enabling the --net-delay feature. Why does the submit stale feature exist then? Because people asked for it. Nowadays it's redundant in my opinion.

[-ck]

I found a new version of OpenCL (AMD APP, Stream SDK, whatever they want to call it) in a leaked version of a beta driver.  This driver supposedly gives Windows 8 support.  Anyway, I don't give a bleep about the driver, just the OpenCL versions.
amdocl(64).dll       10.0.898.1
opencl.dll             1.2.1.0

898.1 is the sdk that's distributed from driver version 12.2 onwards, but it depends on which OS/combo etc. Having a different opencl.dll I'm not sure what it means.

[boozer]

Yep you really have run out of ram. The only time I've seen cgminer itself waste lots of ram is when people have the submit-stale feature enabled and it caches lots of old work and keeps trying to submit them for extensive outages, creating more and more network threads since there is no upper limit to how long it will hold onto this work and try repeatedly to send them. I guess I should make them all time out and give up in a future version. I don't ever suggest using the submit-stale feature and I also recommend enabling the --net-delay feature. Why does the submit stale feature exist then? Because people asked for it. Nowadays it's redundant in my opinion.

Thanks... I don't use the submit-stale option...  When i start it uses about 174 meg of mem, then a couple hours later, its using closer to 400 meg

this is my start script

/usr/bin/screen -dmS cgminer /opt/miners/cgminer/cgminer -Q 2 --api-listen --gpu-engine 700-780,700-780,700-810,700-805,700-820,700-815 --gpu-memclock 240 --auto-gpu --auto-fan --temp-target 75 -I 10 -o http://x:8332 -u x-p x -o http://x:80 -u x -p z

I'm not sure whats going on...

[kano]

Since I've had a few cgminer updates waiting for quite a while, and I've also added another update to the miner.php that some might consider useful, I'll post about it here:

The latest version in my git waiting to be pulled:
https://github.com/kanoi/cgminer/blob/master/miner.h

This adds 3 changes to ckolivas git:
1) Split the devs output tables if there is more than 1 device type
2) A PHP deprecated update for 1)
3) A new option at line 10 "$readonly" that when set to true will force the script to not show any buttons to change cgminer
If it is set to false then it will use the cgminer API command 'privileged' to determine if it should show buttons

Change 3) is useful to force it to read only with true if you want to give public access to the web page but ensure it always is readonly no matter what API access level you set in cgminer

Will merge all this stuff whenever my mining rig is back online and I can actually test any code I commit.

I just noticed that link was wrong - it points to miner.h not miner.php
Oh well I'm sure anyone (other than Xiangfu who is also using it other than me) who did want it, they would have worked that out.
The correct link was:
https://github.com/kanoi/cgminer/blob/master/miner.php

[-ck]

Thanks... I don't use the submit-stale option...  When i start it uses about 174 meg of mem, then a couple hours later, its using closer to 400 meg

this is my start script

/usr/bin/screen -dmS cgminer /opt/miners/cgminer/cgminer -Q 2 --api-listen --gpu-engine 700-780,700-780,700-810,700-805,700-820,700-815 --gpu-memclock 240 --auto-gpu --auto-fan --temp-target 75 -I 10 -o http://x:8332 -u x-p x -o http://x:80 -u x -p z

I'm not sure whats going on...

Neither am I. But I can recommend dropping intensity since 10 does not help anything less than a 7970 and consumes more CPU and makes hashrate more unstable. 7-9 depending on card is optimal.  There is a value, beyond which, the hashrate does not go up. 8 for most cards, 9 for 69x0s and 11 for 79x0s (only with export GPU_USE_SYNC_OBJECTS=1 on linux, otherwise again 9).

[Global BTC]

Uninstalled version "2.6", deleted the DLL's manually in System32 and SysWOW64 (necessary step going between versions!), then installed this new leaked version.  Deleted my kernel bins in cgminer dir as usual, then launched...

Can't really see any change in performance.  Hopefully this gives ckolivas and the kernel devs something to chew on and progress forward, though.

I'm very confused by the overall driver version thing. Is there a post (on this forum or somewhere else) that clarifies driver versions, OpenCL versions, which versions to use, how to change them and so on? I'm using Windows 7.

[jake262144]

...
http://pastebin.com/8MDeRMkD
Anyone have any ideas to help me troubleshoot whats happening?

Your machines run out of memory.

It is the job of the linux 'oom killer' to sacrifice one or more processes in order to free up memory for the system when all else fails.

You mentioned those are not dedicated mining rigs running web servers as well.
Just add a $12 memory module. As a stop gap, create a swap file.
Quest complete, you gain 200 experience points.

[boozer]

You mentioned those are not dedicated mining rigs running web servers as well.
Just add a $12 memory module. As a stop gap, create a swap file.
Quest complete, you gain 200 experience points.

The web server part is only for rig mining stats (BAMT).  It's cgminer that keeps growing in memory size.  I've been watching and I think every-time a thread is idle for 60 seconds and is restarted, it increases overall RAM usage (like the previous thread isn't getting removed from RAM).  I use gpumax, which can have high idle times occasionally....  cgminer starts at 174 meg of RAM for me and currently its using 800 meg, so I think adding another stick of RAM would help, but only by delaying the time it takes for the system to run out of RAM.  If I restart cgminer, then its back to normal mem usage and slowly increases again.

I've been reviewing the command switches and am not sure which one increases the time a thread can be idle before cgminer restarts it... can someone assist me with what switch does that?  I would like to set the thread idle limit to 120 or 180 seconds to see if that reduces my thread restarts (and therefore RAM usage).  I'm on cgminer v 2.3.1f.