OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 606.

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: BCMan on April 11, 2012, 07:29:37 AM

Quote from: bulanula on April 11, 2012, 05:41:53 AM

Why not leave autofan on ?

That way it maintains a constant temperature for the GPUs which should be good for longer-term usage.

Default ati algorithm is better imo and save fan & gpu for longer-term usage as well. With cgminer's autofan on they're less efficient. For example, when 50% is enough for one card for keeping good temp, cgminer switch it to 60%. Or when more speed is needed to keep temp below 69C ati algo spinning it at 48%, and keeping great temp., cgminer switch it to 40% and gpu starting to slowly overheating. These are real examples from my rig. So no, thanks.
And yes, as DeathAndTaxes said, auto-fan don't fix the bug.

That's funny. Cgminer only goes to the temperature you choose and if you don't choose a temperature, it will use 75 degrees, so of course it will go over 69C. Unless of course you set 69 C.

* ckolivas shrugs

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: bitlane on April 08, 2012, 03:39:08 PM

I'm a bit out of the loop, but, did something drastic change in 2.3.2 ?

I have rigs reporting as high as 170% Efficiency right now, with 125-150% being the average range for the others.

Win7 x64, CGMiner 2.3.2
5 rigs each with either SDK 2.1 or 2.4 and CAT 11.12 on all.
Mixed 5xxx/6xxx series card machines, exclusive 5xxx series card machines and exclusive 6xxx series card machines.

The performance is not specific to any card/config. The ONLY thing they all have in common is the 11.12 driver and CGminer version 2.3.2.

WTF ?

I double checked and didn't rely on the Efficiency% readout only. The ratio of GetWork's to Accepted's (even with Rejected included/excluded) does not lie.

Colour me impressed Wink

I'm guessing your pool operator just changed software or settings at their end. The efficiency shouldn't have changed much from 2.3.1 to 2.3.2

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: BCMan on April 10, 2012, 02:40:14 AM

Huge issue with cgminer cant shut down gpu if its overheats. Cgminer trying it to disable the gpu over and over again, but its continuing to mine!

I must have broken it when I instituted the REST followed by restart if it detected overheat. Unless of course it overheated, cooled enough and then restarted over and over again in short bursts? Was it submitting shares at the same rate?

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Eveofwar on April 09, 2012, 11:42:09 AM

Woke up to a hung cgminer Sad

0% activity and no hashing

The log below hasn't changed in the last 5 hours, so I've restarted cgminer (had to X out of the window, as "q" wasn't working)

GPU 0/1 = 5970
GPU 2 = 5830
GPU 3 = 5830
GPU 4 = 5830

[2012-04-09 04:07:28] Accepted 00000000.398219cb.0ccf2195 GPU 4 thread 9
[2012-04-09 04:07:30] Accepted 00000000.5bb7eedb.2e813ef8 GPU 3 thread 7
[2012-04-09 04:07:30] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:30] Will attempt to re-initialise ADL
[2012-04-09 04:07:30] ADL re-initialisation complete
[2012-04-09 04:07:32] Accepted 00000000.031c03e9.d02f5836 GPU 3 thread 6
[2012-04-09 04:07:32] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:32] Will attempt to re-initialise ADL
[2012-04-09 04:07:32] ADL re-initialisation complete

Well that settles it, I cannot successfully re-initialise ADL. I haven't said it for a while since I've been away for a week, but thanks AMD Roll Eyes

I guess the other solution is for cgminer to completely restart with all its original settings. Would people like cgminer to attempt to do this? The problem with doing this unconditionally is that if a GPU has hung, usually the other GPUs can keep mining, but if you try to stop cgminer, they all stop mining. So I would need to make it try to restart itself from scratch only if it hasn't got a dead GPU. Comments?

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Doff on April 11, 2012, 10:41:34 PM

I have a segfault issue id like to see if anyone can help me fix. The segfault only happens when I try to use the 2.4, or 2.5 SDK Drivers with the 12.3 drivers, on Debian/unstable. What is odd is even when I recompile with the 2.6 SDK it still doesn't work I have to reinstall the drivers to get it to stop segfaulting. Witch leads me to believe I'm doing something wrong with the SDKs, Although I switch between 2.4, and 2.5 without an issue using a stable build of Debian with older ATI drivers, so I'm basically at a loss.

Im using a single 5850 on this machine., and cgminer compiles fine, and even shows the correct SDK loaded when I compile for 2.4. The 2.5 just flat out segfaults even with -n.

Id really like to get the 2.4/2.5 SDK working with the 12.3 drivers if possible.

Thanks

Doff

12.3 is your problem. It's a stinker. Drop down to 12.1 or 12.2 if you need 79x0 support.

Doff

sr. member

Activity: 327

Merit: 250

I have a segfault issue id like to see if anyone can help me fix. The segfault only happens when I try to use the 2.4, or 2.5 SDK Drivers with the 12.3 drivers, on Debian/unstable. What is odd is even when I recompile with the 2.6 SDK it still doesn't work I have to reinstall the drivers to get it to stop segfaulting. Witch leads me to believe I'm doing something wrong with the SDKs, Although I switch between 2.4, and 2.5 without an issue using a stable build of Debian with older ATI drivers, so I'm basically at a loss.

Im using a single 5850 on this machine., and cgminer compiles fine, and even shows the correct SDK loaded when I compile for 2.4. The 2.5 just flat out segfaults even with -n.

Id really like to get the 2.4/2.5 SDK working with the 12.3 drivers if possible.

Thanks

Doff

BCMan

hero member

Activity: 535

Merit: 500

Quote from: bulanula on April 11, 2012, 05:41:53 AM

Why not leave autofan on ?

That way it maintains a constant temperature for the GPUs which should be good for longer-term usage.

Default ati algorithm is better imo and save fan & gpu for longer-term usage as well. With cgminer's autofan on they're less efficient. For example, when 50% is enough for one card for keeping good temp, cgminer switch it to 60%. Or when more speed is needed to keep temp below 69C ati algo spinning it at 48%, and keeping great temp., cgminer switch it to 40% and gpu starting to slowly overheating. These are real examples from my rig. So no, thanks.
And yes, as DeathAndTaxes said, auto-fan don't fix the bug.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: ShadesOfMarble on April 11, 2012, 01:11:21 AM

After starting cgminer, is it safe to delete the .bin-files or will that break functionality?
Reason is I want to run cgminer from one folder on different machines (dropbox folder) and they don't have the same SDK installed.

Yes. bin file is simply a cached copy of the compiled kernel. Once a copy of the kernel has been loaded by the GPUs the bin file isn't used until the next start of cgminer.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: bulanula on April 11, 2012, 05:41:53 AM

Why not leave autofan on ?

That way it maintains a constant temperature for the GPUs which should be good for longer-term usage.

auto-gpu also maintains constant temperature and keeps fan at a constant speed.

Also I think everyone is missing the point: auto-gpu vs auto-fan isn't the issue.

cgminer is showing REST. Which indicates it has idled a GPU but the GPU continues to burn away at 100% load. If you had a fan failure for example and that bug occurred it would indicate REST but the GPU would continue at 100% load until it destroyed the core.

bulanula

hero member

Activity: 518

Merit: 500

Why not leave autofan on ?

That way it maintains a constant temperature for the GPUs which should be good for longer-term usage.

BCMan

hero member

Activity: 535

Merit: 500

Quote from: DeathAndTaxes on April 10, 2012, 08:05:01 AM

only auto-gpu should be required. that is how I have my 5970 farm configured.

The fact that cgminer is showing REST indicates a deeper problem. cgminer IS trying to shutdown the card. If it was a config issue it simply would never even attempt a shutdown.

I have never seen this bug before and it is somewhat worrisome. I just had a 5970 idle due to overheat yesterday without issue.

Yeah, don't want to mess with version with such critical bug. Back to 2.2.1 and its working as intended.

ShadesOfMarble

donator

Activity: 543

Merit: 500

After starting cgminer, is it safe to delete the .bin-files or will that break functionality?
Reason is I want to run cgminer from one folder on different machines (dropbox folder) and they don't have the same SDK installed.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

only auto-gpu should be required. that is how I have my 5970 farm configured.

The fact that cgminer is showing REST indicates a deeper problem. cgminer IS trying to shutdown the card. If it was a config issue it simply would never even attempt a shutdown.

I have never seen this bug before and it is somewhat worrisome. I just had a 5970 idle due to overheat yesterday without issue.

jjiimm_64

legendary

Activity: 1876

Merit: 1000

Quote from: BCMan on April 10, 2012, 03:04:04 AM

I have only autogpu enabled (dont want cgminer to control fans), and it worked before, on older version.

isn't the fan more important? I actually do the opposite. I set the gpu fixed, with auto-fan. when things start to warm up I use the api to lower clocks and volts

BCMan

hero member

Activity: 535

Merit: 500

I have only autogpu enabled (dont want cgminer to control fans), and it worked before, on older version.

bulanula

hero member

Activity: 518

Merit: 500

Quote from: BCMan on April 10, 2012, 02:40:14 AM

Huge issue with cgminer cant shut down gpu if its overheats. Cgminer trying it to disable the gpu over and over again, but its continuing to mine!

Do you have auto fan and auto gpu both enabled ?

I think that is the condition for the safety to kick in.

Read a few pages back, it is all there !

BCMan

hero member

Activity: 535

Merit: 500

Huge issue with cgminer cant shut down gpu if its overheats. Cgminer trying it to disable the gpu over and over again, but its continuing to mine!

Eveofwar

sr. member

Activity: 406

Merit: 250

Quote from: kano on April 09, 2012, 04:26:39 PM

Quote from: Eveofwar on April 09, 2012, 11:42:09 AM

Woke up to a hung cgminer Sad

0% activity and no hashing

The log below hasn't changed in the last 5 hours, so I've restarted cgminer (had to X out of the window, as "q" wasn't working)

GPU 0/1 = 5970
GPU 2 = 5830
GPU 3 = 5830
GPU 4 = 5830

[2012-04-09 04:07:28] Accepted 00000000.398219cb.0ccf2195 GPU 4 thread 9
[2012-04-09 04:07:30] Accepted 00000000.5bb7eedb.2e813ef8 GPU 3 thread 7
[2012-04-09 04:07:30] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:30] Will attempt to re-initialise ADL
[2012-04-09 04:07:30] ADL re-initialisation complete
[2012-04-09 04:07:32] Accepted 00000000.031c03e9.d02f5836 GPU 3 thread 6
[2012-04-09 04:07:32] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:32] Will attempt to re-initialise ADL
[2012-04-09 04:07:32] ADL re-initialisation complete

That would probably be the new code that attempts to restart ADL when it stops reporting GPU info.
Someone else mentioned a problem with it in IRC.
Looks like it doesn't work ... has anyone had it work yet?

It's this commit:
https://github.com/ckolivas/cgminer/commit/d4c513030f6d6da4cb54c0d1499d332a3987c376

If you remove the lines added at 686 to 690 (the whole 'if') it should stop it from attempting to do that for the time being until ckolivas gets back.

Good to know I'm not alone then. It just started in 2.3.2 and was fine in 2.3.1-2 Windows

I'll just wait for a fix in the meantime.

GenTarkin

legendary

Activity: 2450

Merit: 1002

Can confirm, I have this same error on my 6950. The fan stops "reporting" or w/e and cgminer goes poop. I just restart CGMINER for now, but this happens maybe once a day or 2.

kano

legendary

Activity: 4634

Merit: 1851

Linux since 1997 RedHat 4

Quote from: Eveofwar on April 09, 2012, 11:42:09 AM

Woke up to a hung cgminer Sad

0% activity and no hashing

The log below hasn't changed in the last 5 hours, so I've restarted cgminer (had to X out of the window, as "q" wasn't working)

GPU 0/1 = 5970
GPU 2 = 5830
GPU 3 = 5830
GPU 4 = 5830

[2012-04-09 04:07:28] Accepted 00000000.398219cb.0ccf2195 GPU 4 thread 9
[2012-04-09 04:07:30] Accepted 00000000.5bb7eedb.2e813ef8 GPU 3 thread 7
[2012-04-09 04:07:30] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:30] Will attempt to re-initialise ADL
[2012-04-09 04:07:30] ADL re-initialisation complete
[2012-04-09 04:07:32] Accepted 00000000.031c03e9.d02f5836 GPU 3 thread 6
[2012-04-09 04:07:32] GPU 3 stopped reporting fanspeed
[2012-04-09 04:07:32] Will attempt to re-initialise ADL
[2012-04-09 04:07:32] ADL re-initialisation complete

That would probably be the new code that attempts to restart ADL when it stops reporting GPU info.
Someone else mentioned a problem with it in IRC.
Looks like it doesn't work ... has anyone had it work yet?

It's this commit:
https://github.com/ckolivas/cgminer/commit/d4c513030f6d6da4cb54c0d1499d332a3987c376

If you remove the lines added at 686 to 690 (the whole 'if') it should stop it from attempting to do that for the time being until ckolivas gets back.

Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 606. (Read 5805746 times)