Bitcoin Forum

 I am pleased to present to you a modified version of AOCLBF 1.75 (http://forum.bitcoin.org/index.php?topic=10264.0) :)
which I give a number 1.76. Since the source code was attached by author I decide to modify it slightly. ;) I'd like to dynamically adjust the gpu clocks within max and min temperature instead of stopping the client. There's a max gpu clock threshold the user sets to provide video driver working  correctly.

Now it looks like this
                                                                                                    
http://i54.tinypic.com/6gisnl.png


Source attached, in case you don't trust the precompiled version!

Tested on my 5850 and 5770. Waiting for your test results.
 
Big thanks to lvlrdka22 for his job and open source!

Version 1.77 released

changes:
 1. Added fanspeed by user on startup.
 2. Added automatic fanspeed when set to 0
 3. Added OSD information about clockspeed and fanspeed

Version 1.8 released

changes:
 1. Full phoenix 1.6.x compatibility.
 2. Added combobox to choose the kernel
 3. Fixed the bug with adapter index in combobox after Overclock gpu string when more than one adapter used
 4. Device nickname moved after Client # in OSD to leave more space for temp, clocks, fanspeed displaying
 5. Minimizes on startup now
 6. Set to default all adapters on exit

Version 1.81 released

changes:
 1. Changed text on the form to right one (thanks to newunit16 for remark).
 2. Now if clocks change automatically there's no "Set to" message on OSD
 3. No matter if chekbox "Downclock GPU" is checked Temp, clocks and RPM will be displayed on OSD
 4. "OSD On/Off" traymenu item was added.

Version 1.83 released

changes:
 1. Completely AutoFanspeed now (at first it sets fanspeed to 100% by 5% in step and then only adjusts clockspeed if temp is above maxtemp)
 2. Many optimizations made to increase stability
 3. Native icon in exe :)

Version 1.8.4 released

changes:
 1. Added ability to fix fanspeed if autofanspeed isn't needed
 2. Added shares per minute indicator

This topic has moved from https://bitcointalk.org/index.php?topic=32125.0

Could it be optional? Sometimes I exit AOCLBF to start another miner, so I want to save all clocks & fans settings.

Xanderus
You couldn't exit aoclbf when switching to the other miner.Only stop all miners in aoclbf. So all clocks wouldn't set to default.

Everything working great! I really hope you could make the voltage mod work (go past the BIOS limit and still work). It still won't get past the BIOS Limit.

Just a thought, would you be able to create a Frontend to CGMiner (I mean if it is possible)? That would be perfect!

because of the on the fly menu options with cgminer your better of with a batch script and barelyclocked.exe(for setting clocks and fan speed)
if you need I can show how an example batch script using these would look like. ALso you can set affinity with batch script making a script the most efficient way to run cgminer on Windows.

Thanks for the offer. I recently learned how to set CPU affinity using the script for CGMiner. I guess I just value the All-in-one mining solution (i.e. OC, mining, monitoring, tweaks).

Getting an Autoit error "Line 8736 Error:Array variable subscript badly formatted." When this dialog pops up which is about once a day the OSD is no longer updating. Miners are still working but once you click OK button program shuts down.

Any ideas?

+1. It happens with 1.80 also.

Im still using the 1.75 Version with moded kernels, Will you upload a version where this is included?
Or will i have any increase in my Mhash using yours? I saw those OC options and all, but my cards are picky with their voltage, so i have to use Trixx(for my 5830) and Afterburner (for my 5850 MSI TFII).

I get this. Set up a script program to restart my box every 12 hours to "fix" the problem.

I also get the array problem, but then I need to install to non GPU systems and GPU systems of many different kinds.  Is there anything else out there as an alternative to this?

I am getting an array error as well, "Error: Subscript used with non-Array variable"

Windows 7 64bit Ultimate, fully patched
3x 5830's

I tried dropping (https://bitcointalk.org/index.php?topic=10264.msg293309#msg293309) a downloaded atiadlxx.dll into system32, no dice.  Of course, it could be malware as I did download it from what looked like a content farm (it did pass a Total Virus check).

Windows 7 64bit Ultimat
about 30-40 hours bug
http://pikucha.ru/i7bWg/thumbnail/AOCLBF.jpeg (http://pikucha.ru/i7bWg)

AOCLBF 1.81 by WinD
Phoenix 1.6.4

what could be the problem?

How many video adapters do you have working?
If OSD is off do you have this problem?

I get this as well usually after 2 or 3 days of mining. 

Win7 32 , 3 gpus

The mining does continue  although the program is "stuck"

It seems to me that it usually happens when the pool I'm on has connection problems
but it could be just a coincidence .

I got this error if when i was mining at less than 24 hours continuously.
I was mining with OSD on & this error came when i was mining more than 10+ hours.
Only was is to, click ok & restart miner.
So i even regularly restarting miner less than every 24 hours.

now i am 100% sure that memory leak is caused by disconnectes. I have very stable internet and no problem. ok ... will be watching the code....

It is also very tormented.

i also would like to ask about this bug? is it in aoclbf 1.75? someone tested it?

don't able to remember, coz it was more than 3 months ago.

try this one... http://www.mediafire.com/?1qa45cb1182bjve
what about stability?

it downclocks each gpu by 10 MHz every second

RedLine888
post your settings (screenshot preferred)

client settings or gpu settings?

Wind. Great to see you are still around. the GUI is running very well. I Get the same error as well. Even More so on my big rigs with a couple 6970s or 5870s. the more miners the more often i see the error. I had the reconnect set to 15 sec. Some times when the connection gets bad it was enough to crash my systems. Increased the page file to 4 to 8gb increased ram and increased the reconnect window to 60 sec and it occurs less but still every 3 days or so. Tried 1.82b too and it was same thing. Also if you could add a NO OSD on start up so its completely stealth that would rock. And an Option to increase the Power Tune to 20% in the overclock window would help on the 6 and 7 series coming out. Its a bother when you have a headless unit and they are aren't X-fired together. Thank you so Much! Happy Holidays!

Would love to see a solution to the array error.

Well... trying to fix this error...
Give you on testing a new one http://www.mediafire.com/?8kak78yeq11t8a0
Waiting for your test results....
forget to say you should test on phoenix only-poclbm is without changes yet

Thank you! Testing right now. 1.82b2 with Phoenix 1.72 phatk2 2x5870 930/300. Leaving Office...now...

Yes, with the same feature disabled downclock GPU power drops to 10 per second, working on 1.8, if a bug flies we are talking about a miner still continues.

I get a trojan win 32 generic alert when I want to open the new aoclbf1.82b file. Now I'm scared. Is that normal?

Source code is attached! You can yourself compile it.

tried this version to solve the script error that pops up.

The speeds of the hashing on gpu 2 and 3  were very low and not stable
and the whole system got stuck eventually ( not sure if it is related)
couldnt run it long enough to see if script error went away

running on win7 64

putting the old version back solved the problems ( script problem still there of course)

What vesrsion is that?

I also ran the the beta. still got the array error. fiddled around with some stuff and i think the memory leak and the array error are two different problems. I was thinking the leak was coming from auto restarting so i i went from 60 to 90 seconds timer on all my single card miners 60 to 3 min on my double cards and 5 min on my 3 card miners. still got the array error but much much less memory usage. I mine at bitcoinpool.com and the had a power outage and a few miners got hit with the array error then another time for no reason all of them got the error. Only time i can produce the error myself is when swapping cards and not starting with a fresh copy of the GUI. For an example trade a 5870 for a 6970 and boot the system up. When the GUI starts it gives the error. If it helps beta 2 did resist the error longer than beta 1 and 1.81 but after i got the first array error i reverted back to 1.81. Thank you for staying with this. Any chance of being able to save the hide OSD command? kinda annoying to hit hide osd every time it starts up. esp with array bug and having to restart every few days.

1.82b2

went back to 1.81

FRanz33
farfiman
thanx for testing. i'm now working on optimizing the code . it'll be faster when disabling OSD and may be array and memory problem will be solved..i hope..
in plans is smth kinda sheduler to change aggression-useful for machines a user works on
 
will be done

so how about those 7970 optimizations?  ;D

BigBobo
it seems to work with 7970. Unfortunately i am unable to test it.

2All
Catch aoclbf 1.82b3 by WinD (http://www.mediafire.com/?2pwcogym33egw39)
Another try to fix an array error

brilliant. the hide osd by default is enough to mkae me deploy this beta on on my units! thanx will post results soon. also swtiching bck to my default restart times of 60 sec.

Sorry to say still get an Array error with beta 3 however i did notice something that may help. The array error happened right after we solved a block and i think I caught it early on. The auto hide OSD helped alot tho!

FRanz33
pity...how about your internet connection? Are there frequent breaks? How stable pools you are on? Are you sure your system memory is ok?

I like it, but I wonder why the OSD shows an hash rate quite lower than what I get with Phoenix 1.7.x (with phatk2, same aggression).
On a 5870@970/300 I get 410-420 Mhs against 440 Mhs with Phoenix. Similar difference with a 5830 and a much smaller one with 2 5750.
Maybe it is just because hash rate does not seem to refresh frequently?

System Memory is fine..great in fact. Internet connection is ok to. I actually have multiple boxes on the same router in office spaces and at home and alot of times all of them on the same network are affected while other networks aren't. Sometimes only half on the same network while the other half are not. But i will say this. the more cards in the box the more likely the error. and i had another array error today and we solved to blocks early morning maybe there is a connection or something to do with the way long polling works idk. Im going to upgrade to Phoenix 1.7.4 from 1.7.2 to see if that helps. Im running half of my units with beta and the other half on release version. No difference.

conspirosphere.tk
hashrate refreshes once per 2 sec
FRanz33
i'm close to fixing this ... but have no much time..may you launch 1.75 version on one machine for comparison?

been using 1.75 on my 2x5870@930/300 for a couple days. NO array error...still testing. using Phoenix 1.7.5..will update in another 3 days

conducted research for aoclbf 1.75 too. Result is that memory leak occurs! For testing I changed pools to wrong ones so miner was unable to connect and was continuously restarting. I watched for Commit charge By procexplorer and it was growing up very quickly. Now I did some changes in code to minimize leak but can't completely remove it while..

Yea that was a problem with 1.75 but you can get around it i think with longer reconnect times. Phoenix is no where near as prone to idle as it was in the past so this feature isn't as important. 1.75 is lacking so many features compare to 1.82b3 but i think the array error lies in one of those features you added because i dont remember seeing it until before then. I will be trying a test soon tho. Im getting new cards and will see if the array error comes up when i switch the cards out like it does with 1.82. Keep at it. i think your close!

so many changes in algorithm were done that I decided to start 1.83 number :) After testing it is 3-4 times lower in memory leak when a lot of disconnects occur.
Welcome aoclbf_1.83b_by_WinD (http://www.mediafire.com/?f93no4mjfo8yb3u)

Thx :)

Dont know how to insert screen shots buuut.
Line 8708 (file "c:\Phoenix\aoclbf 1.83b.exe"):
Error: Array Variable subscript badly formatted.
 on 5870 plus 5770 x2
Aggression 12 9 9 phatk2 256 930/300
5 min restart

Only unit with 3 cards (for now) only unit to fire an error even when pool was down for 12hrs
good Job. Almost there

 FRanz33
Can you launch a script itself? I wanna know what line in script causes an error.
p.s. seems to be an old error https://bitcointalk.org/index.php?topic=21380.0

done some optimizations in getting data from drivers
aoclbf_1.83b2_by_WinD (http://www.mediafire.com/?w7qog463ogfgkva)

TESTING

I used to mine on 5870.
I get strange problem. The miner mines with correct clock speed, then clock speed gradually reduces & mining also reduces.
I set 930/300 & from 930 it went down to 360 mhz in less than 3 minutes.

look at this settings
https://bitcointalk.org/index.php?topic=39697.msg660594#msg660594
you should set MaxTemp, MinTemp and MaxClock
if not-try to delete aoclbf.ini and then reset all settings

ok, working

Is their any way to make OSD on & not minimize AOCLBF?
like command line parameters which can be given to show OSD on.
New versions not showing OSD which makes it useless, unless every time has to click & select OSD on/off

may be will do this as option if it is so important
how about stability in new beta?

Its important for me to see OSD, that gives me temperature of the cards.

Stability, i am not using continuously for 24 hours, but at least 12 hours it run with out any error or problem.
12 hours perfect.

Phoenix 2 is in beta.  will you support it ?

if phoenix 2 is more powerful then 1.x one I will think about it..

looks very good so far... fixes lots of problems

How to make OSD visible default?

nohow so far..will be realised as option later
is it really big troble for you? Do you often restart AOCLBF?

Power cut makes me restart 10-15 times a day

beta 2 is much better. just had two array errors today for fist time since release. One Box has 3 6970's 940/825 aggression 11 with 6 gb ram other box has 4 6870's 930/830 aggression 9 6gb ram. Im wondering if maybe a higher than normal aggression causes twohe error to occur more often. Maybe not but whatever your doing makes it more stable. As far as OSD goes i prefer it to be auto off as most my units mine in office areas and are used by other people. Perhaps you can make it a savable option to Auto hide or Auto show. An auto Click thru would to and a more robust option menu when you right click the tray icon. Just throwing out ideas. Thank you for all your hard work!

well thank you
but I wonder where error is. I would like to ask to launch AOCLBF from script itself 'cause compiled one doesn't give me info what line in script causes an error.
FRanz33 can you do that?

Im ashamed to say that i don't know what ur talking about. walk me thru and ill launch it as a script on my fastest rig to test. or some one else can try.

Also my rig with 4x6870 and 3x 6970 errored today. Im going to try something diff to see if i can help.
On one rig (4x6870) im going to disable the restart option. seeing as how the accepted shares reaches into 10s of thousands and phoenix is still going after the error the restart option isnt as important as it was. and the other rig im going to disable the down clock feature. Maybe one of these are causing the error as they are using resources to actively monitor something. what do you think?

Its easy to recreate the error.  Just put in wrong pool info  or disconnect the internet from the machine
and let it switch back and forth between main and backup pool... eventually it will crash. 

Just an update. Tried with the down clock disabled. Tried with the auto restart disabled. Still got the error. Going to try with both disabled.

Hi everyone. I really like this program. Ever since this poclbm issue recently and the entire SDK2.6 problem, I'm running many of my rigs with the guiminer and diaphatk kernel. This is more an emergency measure though. I haven't been able to get aoclbf to run with diaphatk so far. I do not want to switch to CG miner, since my results with it were dissatisfying.

I'd like to go back using aoclbf so I can have a backup pool option. Furthermore, is there a way to solo mine using aoclbf's backup function?

The reason is following: I mined 7 blocks in 8 months at slush's pool and one block in one month at btcguild yet only got about 250 BTC paid out alltogether. This makes me wonder if it wouldn't be great to have a solo mining option. I will never catch up and who knows with decreasing btc prices, the difficulty might go down again another time.

With diaphatk I got most of my performance back, but the 100%CPU bug is still there. It was kind of nice deciding whether to spare the cpu or go for full performance, but now with the additional pcoblm bug, I got confused. I do not want to spend another few weeks figuring out how to get things back up and running smoothly.

So, quick recap:

Which aoclbf is best to use with SDK2.6 (catalyst 11.2-12.1)?
Should I use the old version of guiminer that is not affected by the recent bug?
Which files do I copy exactly, and where?

Thank you for your help!

 ??? Use the newest beta of AOCLBF. to use a different kernel just copy and overwrite the files in the phak folder with the kernel of your choice. use 2.6 if your using 7xxx card 2.5 if using 6/5xxx...Gui miner is more stable but has less features in my opinion. I think you can solomine the same way u do in phoenix just put in the local address instead of the pool address. but umm anywhoo...
DUDE. it just dawned on me how to fix (temp.) the error. Is it possible to have AOCLBF exit and restart itself every 12 24 36 or 48 hrs like an extra setting that you could choose?  I know this program is prolly dude for an overhaul since phoenix 2 is out but still. I thik that would put an end to this problem so we can move on to other stuff  8)

I run mine with serverdoc, hence if AOCLBF dies it gets restarted with a batch of other important processes.(I double my miner as my server too)

Your miner supports 6 video card?

I think yes like on Win7 only. You can try that.

On Win7 tried in the hardware list, only 4 points.

http://pikucha.ru/i8FRE

You can select manual & then it will show number next to Download GPU.

By manuals are all set, but at the same time is only 5 cards, no matter in what order.

I am getting this once a day from a few days. When it happens, the prog crash.
http://img444.imageshack.us/img444/1893/errdv.jpg
 ???

Its the same error that always has been. It happens mostly after a lot of connect/disconnect to pools.
Happens to me once every few days now. The mining stills go on ( at least for me with phoenix).
I just push ok and run again.

http://img856.imageshack.us/img856/3451/autoiterror.th.png (http://imageshack.us/photo/my-images/856/autoiterror.png/)
Does it work at all?

have been using it since july 2011 , so.. yes

Welcome aoclbf_1.83_by_WinD (http://www.mediafire.com/?17vrr6s7dsz1s9v)
And some remarks about AutoFanSpeed logic: by dafault you can set to 0 fanspeed that means you have your fanspeed on auto. But if temp>maxtemp fanspeed grows up by 5% and so on while it reaches 100% and only then clockspeed will be decreased. If temp<mintemp clockspeed will be increased and only then fanspeed. So providing of maxhashrate is reached if temp of your videocard is changing in a wide range within days.

Thanks, but I want a constant fanspeed. I like to keep it fixed at 46% on my cardz, and downclock if temp>maxtemp . It is still possible?

may be i will add maxfanspeed as option in next release

How to ON OSD ?  ???

Right click on tray icon "OSD On/Off"

Is final 1.83 has any significant advantage over 1.83b or 1.83b2 versions except AutoSpeed? I just NOT want automatic speed, I think it got to be an option. I'm on 1.81 now, and I'm not using 1.83 exactly because of autospeed.

Yea 1.84 will come up. Autofan won't be if you fix fanspeed.
The difference is in optimization of the code but complete stability is not reached yet.
1.83b2 differs only in autofanspeed

Version 1.8.4 released (http://www.mediafire.com/?lvz5785wa3sa44a)

changes:
 1. Added ability to fix fanspeed if autofanspeed isn't needed
 2. Added shares per minute indicator

Emsisoft thinks that is a trojan, maybe because is embedded in some botnet:

Emsisoft Emergency Kit - Version 1.0
Last update: 5/12/2012 4:38:04 PM

Scan settings:
Scan type: Smart Scan
Objects: Memory, Traces, Cookies, E:\Windows\, E:\Program Files
Scan archives: Off
Heuristics: Off
ADS Scan: On

Scan start:   5/12/2012 4:39:45 PM

[1244] E:\Program Files\Bitcoin\aoclbf.exe    detected: Gen.Trojan.Heur!IK
E:\Program Files\Bitcoin\aoclbf.exe    detected: Gen.Trojan.Heur!IK

https://bitcointalk.org/index.php?topic=39697.msg690039#msg690039

Anyway to show OSD with out clicking OSD on/off?
Like command line parameters?
or any other way to show OSD automatically?

Is this nice thing supposed to work with XP x64?
I am getting this every time that I try to launch it:
http://i49.tinypic.com/1x107.jpg (http://i49.tinypic.com/1x107.jpg)

edit: No, now it works. It was because I needed .Net framework.
Now it just does not connects to the pools. :-/