undervolting "cheap" cards - how to? ( 7850 powercolor )

[drakoin]

Afterburner doesn't allow to change the voltage, cgminer doesn't do it.

I have:
1 PowerColor HD 7850 (AX7850 2GBD5-DH) with 85.3% ASIC quality
1 XFX R7970 DD (FX-797A-TDFC) with 68.6% ASIC quality

How to unlock the voltage control? Undervolt the cards? Overwrite the vga-bios?
Any hint welcome. Manuals to read?

I checked techpowerup.com but I already have the newest BIOS.

Thanks a million!
Enjoy your new years celebration!

[1714617002]

1714617002

[1714617002]

1714617002

[1714617002]

1714617002

[1714617002]

1714617002

[Wipeout2097]

I changed the voltage of my 7970 and 7870 with this tool:
http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/

1.05V for the 7970 allows 1100Mhz and 1V for the 7870 allows 950+Mhz. YMMV, of course

[drakoin]

http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/

SUPER, thanks.

And which brand were your 7?70 - or doesn't that matter?




How can I find out, which voltage regulator is on my cards?
 CHL822x, CHL8214, UP1637, UP1801, ST6788A, VT1556M


Thx :-)

[Wipeout2097]

Mine are VTX3D (crap brand, I guess). I was lucky though: cheap, hash well, voltage regs supported and trouble free so far.

To figure out the voltage regulators, perhaps you have to remove the cooler, I dunno.  You can register at that forum and ask...

[drakoin]

You can register at that forum and ask...

I have done that, thanks a lot.

Have a look here, perhaps anyone has any ideas:
http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/page-23#post-3042091

Happy New Year!

[drakoin]

I just want to announce SUCCESS :-) I did it!
Undervolting XFX 7970 DD by tweaking the original BIOS;
amazing results, saving 1/3 energy!

Instructions here:
https://bitcointalk.org/index.php?topic=395004.msg4282261#msg4282261

 

[Wipeout2097]

Very nice results! Wish I knew about that other thread before, could have been of assistance.

[novymivo]

excellent
Do you know how to undervolt r9 280x cards ??

[drakoin]

I am in the process of finding the optimal voltage for my PowerColor HD 7850 (AX7850 2GBD5-DH) with 85.3% ASIC quality - and learning a lot of things on the way :-)

I have two cards (7850 & 7970), both undervolted, so the cause for a crash is never 100% sure (or do you have a criterion?), and none of the tools are perfect; even though e.g "GPU-Z" is free and amazing, and it beautifully logs all hardware values into a file, once per second - that logfile is always only for one card at a time, how stupid.

Please help me with my reasoning now. If you have knowledge that can help us to understand this better, please share.


The cgwatcher logfile says:

Code:

... after 5 hours of happy mining ...
[05/01/2014 05:30:53]  CGMiner (4676): Network difficulty is now 320.
[05/01/2014 05:31:23]  CGMiner (4676): Network difficulty is now 253.
[05/01/2014 05:39:50]  CGMiner (4676): Network difficulty is now 3,195.
[05/01/2014 05:43:51]  CGMiner (4676): Network difficulty is now 253.
[05/01/2014 05:49:33]  CGMiner (4676): Network difficulty is now 47.
[05/01/2014 05:49:54]  CGMiner (4676): Network difficulty is now 49.
[05/01/2014 05:50:14]  CGMiner (4676): Network difficulty is now 51.
[05/01/2014 05:51:04]  CGMiner (4676): Network difficulty is now 54.
[05/01/2014 05:51:14]  CGMiner (4676): Network difficulty is now 3,195.
[05/01/2014 05:54:25]  CGMiner (4676): Network difficulty is now 253.
[05/01/2014 05:55:25]  CGMiner (4676): GPU0 (AMD Radeon HD 7900 Series) status is INACTIVE
[05/01/2014 05:55:25]  CGMiner (4676): GPU0 (AMD Radeon HD 7900 Series) status is UNKNOWN
[05/01/2014 05:55:28]  CGMiner (4676): Process closed.

So here, the 7850 is not even mentioned.
But at 5:55:25 something happens to the 7970 card, for sure - and within 3 seconds cgminer is dead.

Does a Catalyst driver crash look like that?


I don't suspect the 7970 card to be the cause, because I underclocked it first, and it was running stable for more than a day now.
But still, you never know. They sit tightly together, one overheating might trigger the other one?


I have more insight about the other card, a 7850 PowerColor. Fortunately, GPU-Z logged that one.

The columns are:
* Date           
* GPU Core Clock [MHz]    
* GPU Memory Clock [MHz]    
* GPU Temperature [°C]    
* Fan Speed (%) [%]    
* Fan Speed (RPM) [RPM]    
* GPU Load [%]    
* Memory Usage (Dedicated) [MB]    
* Memory Usage (Dynamic) [MB]    
* VDDC [V]

The 20 minutes before were without any change:

2014-01-05 05:33:00    968   1289   62   84   3839   94   1123   41   0.939
                           
2014-01-05 05:34:00    968   1289   62   84   3845   93   1123   41   0.939
                           
2014-01-05 05:40:00    968   1289   61   84   3858   97   1123   41   0.939
                           
2014-01-05 05:53:00    968   1289   61   84   3844   95   1123   41   0.939

Then something happened to the memory usage.

All over the logfiles, I do see seldom jumps in the memory usage,
and they don't (?) seem to sync with pool / difficulty change
(which is logged in the cgwatcher.log above).

Anyways, that is the first real change of values here, 2 minutes before the crash:

2014-01-05 05:53:14    968   1289   62   84   3834   94   1123   41   0.939
2014-01-05 05:53:15    968   1289   61   84   3834   95   1123   41   0.939
2014-01-05 05:53:16    968   1289   62   84   3838   94   1123   41   0.939
2014-01-05 05:53:21    968   1289   62   84   3838   63   1123   41   0.939
2014-01-05 05:53:21    968   1289   59   84   3864   58   1123   41   0.939
2014-01-05 05:53:23    968   1289   60   82   3833   51   1079   48   0.939
2014-01-05 05:53:26    968   1289   60   83   3838   51   1057   38   0.939
2014-01-05 05:53:27    968   1289   60   82   3821   79   1097   40   0.939
2014-01-05 05:53:27    968   1289   60   82   3829   79   1104   40   0.939
2014-01-05 05:53:28    968   1289   61   83   3833   84   1111   41   0.939
2014-01-05 05:53:29    968   1289   60   83   3847   86   1107   41   0.939
2014-01-05 05:53:30    968   1289   61   83   3846   84   1107   41   0.939

apart from that nothing special happening.

2014-01-05 05:54:00    968   1289   60   80   3808   95   1107   41   0.939


then first (!) and strangely, the fanspeed is slowing,
then the GPU temperature falls a bit:
                           
2014-01-05 05:54:01    968   1289   60   80   3809   93   1107   41   0.939
2014-01-05 05:54:02    968   1289   60   80   3801   94   1107   41   0.939
2014-01-05 05:54:03    968   1289   60   80   3801   95   1107   41   0.939
2014-01-05 05:54:04    968   1289   60   79   3783   94   1107   41   0.939
2014-01-05 05:54:05    968   1289   60   79   3780   94   1107   41   0.939
2014-01-05 05:54:06    968   1289   60   79   3794   95   1107   41   0.939
2014-01-05 05:54:07    968   1289   60   79   3782   94   1107   41   0.939
2014-01-05 05:54:08    968   1289   60   79   3781   93   1107   41   0.939
2014-01-05 05:54:09    968   1289   60   79   3786   93   1107   41   0.939
2014-01-05 05:54:10    968   1289   60   78   3748   93   1107   41   0.939
2014-01-05 05:54:11    968   1289   60   78   3750   94   1107   41   0.939
2014-01-05 05:54:12    968   1289   60   77   3740   93   1107   41   0.939
2014-01-05 05:54:13    968   1289   59   77   3745   92   1107   41   0.939
2014-01-05 05:54:14    968   1289   59   77   3741   92   1107   41   0.939
2014-01-05 05:54:15    968   1289   59   77   3739   93   1107   41   0.939
2014-01-05 05:54:16    968   1289   59   76   3724   93   1107   41   0.939
2014-01-05 05:54:17    968   1289   59   76   3711   93   1107   41   0.939
2014-01-05 05:54:18    968   1289   59   76   3718   93   1107   41   0.939
2014-01-05 05:54:19    968   1289   59   76   3727   94   1107   41   0.939
2014-01-05 05:54:20    968   1289   59   75   3700   93   1107   41   0.939
2014-01-05 05:54:21    968   1289   59   75   3690   94   1107   41   0.939
2014-01-05 05:54:22    968   1289   59   74   3674   93   1107   41   0.939
2014-01-05 05:54:23    968   1289   59   74   3670   93   1107   41   0.939
2014-01-05 05:54:24    968   1289   59   74   3678   94   1107   41   0.939
2014-01-05 05:54:25    968   1289   59   74   3677   93   1107   41   0.939
2014-01-05 05:54:26    968   1289   59   73   3657   93   1107   41   0.939
2014-01-05 05:54:27    968   1289   59   73   3659   93   1107   41   0.939
2014-01-05 05:54:28    968   1289   59   72   3647   93   1107   41   0.939
2014-01-05 05:54:29    968   1289   59   72   3633   93   1107   41   0.939
2014-01-05 05:54:30    968   1289   59   72   3623   92   1107   41   0.939
2014-01-05 05:54:31    968   1289   58   72   3623   92   1107   41   0.939
2014-01-05 05:54:32    968   1289   58   71   3609   93   1107   41   0.939
2014-01-05 05:54:33    968   1289   59   71   3606   93   1107   41   0.939
2014-01-05 05:54:34    968   1289   58   70   3596   93   1107   41   0.939
2014-01-05 05:54:35    968   1289   59   70   3585   94   1107   41   0.939
2014-01-05 05:54:36    968   1289   58   69   3552   93   1107   41   0.939
2014-01-05 05:54:37    968   1289   58   69   3536   93   1107   41   0.939
2014-01-05 05:54:38    968   1289   58   68   3524   93   1107   41   0.939
2014-01-05 05:54:39    968   1289   58   68   3514   93   1107   41   0.939
2014-01-05 05:54:40    968   1289   58   67   3488   93   1107   41   0.939
2014-01-05 05:54:41    968   1289   58   67   3460   93   1107   41   0.939
2014-01-05 05:54:42    968   1289   58   66   3446   93   1107   41   0.939
2014-01-05 05:54:43    968   1289   58   66   3433   94   1107   41   0.939
2014-01-05 05:54:44    968   1289   58   65   3414   94   1107   41   0.939
2014-01-05 05:54:45    968   1289   58   65   3401   94   1107   41   0.939
2014-01-05 05:54:46    968   1289   58   64   3378   94   1107   41   0.939
2014-01-05 05:54:47    968   1289   58   64   3350   94   1107   41   0.939
2014-01-05 05:54:48    968   1289   58   63   3336   94   1107   41   0.939
2014-01-05 05:54:49    968   1289   58   63   3320   94   1107   41   0.939
2014-01-05 05:54:50    968   1289   58   62   3301   94   1107   41   0.939
2014-01-05 05:54:51    968   1289   58   62   3286   94   1107   41   0.939
2014-01-05 05:54:52    968   1289   58   61   3259   94   1107   41   0.939
2014-01-05 05:54:53    968   1289   58   61   3227   94   1107   41   0.939
2014-01-05 05:54:54    968   1289   58   61   3223   94   1107   41   0.939
2014-01-05 05:54:55    968   1289   58   61   3225   94   1107   41   0.939
2014-01-05 05:54:57    968   1289   58   60   3208   94   1107   41   0.939
2014-01-05 05:54:58    968   1289   58   60   3196   95   1107   41   0.939
2014-01-05 05:54:59    968   1289   58   59   3182   93   1107   41   0.939
2014-01-05 05:55:00    968   1289   58   59   3165   93   1107   41   0.939
2014-01-05 05:55:01    968   1289   58   59   3161   93   1107   41   0.939
2014-01-05 05:55:02    968   1289   58   59   3163   93   1107   41   0.939
2014-01-05 05:55:03    968   1289   58   58   3135   92   1107   41   0.939
2014-01-05 05:55:04    968   1289   58   58   3103   92   1107   41   0.939
2014-01-05 05:55:05    968   1289   58   57   3082   94   1107   41   0.939
2014-01-05 05:55:06    968   1289   58   57   3065   94   1107   41   0.939
2014-01-05 05:55:07    968   1289   58   57   3058   93   1107   41   0.939
2014-01-05 05:55:08    968   1289   58   57   3061   94   1107   41   0.939
2014-01-05 05:55:09    968   1289   58   56   3047   94   1107   41   0.939
2014-01-05 05:55:10    968   1289   58   56   3030   94   1107   41   0.939
2014-01-05 05:55:11    968   1289   58   55   3002   93   1107   41   0.939
2014-01-05 05:55:12    968   1289   58   55   2969   93   1107   41   0.939
2014-01-05 05:55:13    968   1289   58   54   2946   92   1107   41   0.939
2014-01-05 05:55:14    968   1289   58   54   2925   92   1107   41   0.939
2014-01-05 05:55:15    968   1289   58   54   2922   92   1107   41   0.939

Then the GPU load collapses within 5 seconds,
almost all memory is suddenly freed,
GPU rises again to 64% for five seconds ...
... then the GPU load drops to zero, and the clockspeeds go idle:

2014-01-05 05:55:16    968   1289   58   54   2921   92   1107   41   0.939
2014-01-05 05:55:17    968   1289   58   53   2907   92   1107   41   0.939
2014-01-05 05:55:18    968   1289   58   53   2885   92   1107   41   0.939
2014-01-05 05:55:19    968   1289   58   53   2877   94   1107   41   0.939
2014-01-05 05:55:20    968   1289   55   53   2887   80   1107   41   0.939
2014-01-05 05:55:21    968   1289   54   53   2891   60   1107   41   0.939
2014-01-05 05:55:22    968   1289   54   53   2890   47   1107   41   0.939
2014-01-05 05:55:23    968   1289   54   53   2889   29   1107   41   0.939
2014-01-05 05:55:24    968   1289   53   53   2889   9   1107   41   0.939
2014-01-05 05:55:25    968   1289   53   53   2887   64   79   23   0.939
2014-01-05 05:55:26    968   1289   53   53   2887   64   80   23   0.939
2014-01-05 05:55:27    968   1289   53   53   2889   64   80   23   0.939
2014-01-05 05:55:28    968   1289   52   53   2890   64   76   23   0.939
2014-01-05 05:55:29    968   1289   52   53   2892   64   76   23   0.939
2014-01-05 05:55:30    300    150   52   53   2890   0   76   23   0.825
2014-01-05 05:55:32    300    150   50   53   2891   0   76   23   0.825
2014-01-05 05:55:33    300    150   49   53   2891   0   76   23   0.825
2014-01-05 05:55:36    300    150   48   53   2891   0   76   23   0.825
2014-01-05 05:58:17    300    150   31   53   2847   0   76   23   0.825
2014-01-05 06:00:27    300    150   30   53   2828   0   76   23   0.825
2014-01-05 06:00:50    300    150   29   53   2827   0   76   23   0.825


Where it stays, cooling down ...

2014-01-05 05:58:17    300    150   31   53   2847   0   76      23   0.825

2014-01-05 06:00:27    300    150   30   53   2828   0   76      23   0.825

2014-01-05 06:00:50    300    150   29   53   2827   0   76      23   0.825

2014-01-05 06:30:00    300   150   28   53   2805   0   86   23   0.825

Until I find it half an hour later - because my little port4028 watcher-tool looks suspicious.



Hmmm ... that was all 7850 data. which I am currently testing to undervolt.
Unfortunately, I do not have parallel 7970 data :-(
Perhaps the 7970 malfunctioned first?


For now, I have raised the voltage of the 7850 a little bit.
Let's see if that is enough for stability.


What do you think, which card was the cause of the above crash?

[drakoin]

The 7970 might be readily tuned, measurements & results here:
https://bitcointalk.org/index.php?topic=395004.msg4320578#msg4320578

 

[kalus]

anybody know how to undervolt the R9 270x?  there is no bios switch, and i haven't been able to find hacked bios's either. 

[drakoin]

Do you know how to undervolt r9 280x cards ??

anybody know how to undervolt the R9 270x?

i haven't been able to find hacked bios's either.  

I don't think you need to find someone else's BIOS.
With my description you should be able to make your own tweaked BIOS:
https://bitcointalk.org/index.php?topic=395004.msg4320578 (read to end)

there is no bios switch

Yes, my cards don't have a BIOS switch either.
So ... good preparation & full concentration when I do this :-)

AFAIK a bricked card can sometimes be "unbricked" with a second card, so be extra careful if you only have one.

It's all risky, be sure you know what you do.

[kalus]

Do you know how to undervolt r9 280x cards ??

anybody know how to undervolt the R9 270x?

i haven't been able to find hacked bios's either.  

I don't think you need to find someone else's BIOS.
With my description you should be able to make your own tweaked BIOS:
https://bitcointalk.org/index.php?topic=395004.msg4320578 (read to end)

ya, i have seen flashes for 7xxx cards, but not for r9 2xxx cards.  oh well, i'll keep looking.

there is no bios switch

Yes, my cards don't have a BIOS switch either.
So ... good preparation & full concentration when I do this :-)

AFAIK a bricked card can sometimes be "unbricked" with a second card, so be extra careful if you only have one.

It's all risky, be sure you know what you do.

How do you 'unbrick' a card with another card?  

and i have 22 of them, so i got backups

[drakoin]

ya, i have seen flashes for 7xxx cards, but not for r9 2xxx cards.  oh well, i'll keep looking.

Ah, now I get you, sorry I was slow :-)

So what you say is the VBE7 editor doesn't work for the R9-2xxx cards?
If you risk bricking (see below) - why not simply try to use it for one of your R9s ?

And: If you find an R9-tool, please PM me, I might buy R9 cards soon.

How do you 'unbrick' a card with another card?  

Oh, I really don't know. Has'nt happened to me, and that makes me happy :-)
But I am sure that I read somewhere " ... then you need another card ... ".

and i have 22 of them, so i got backups

Oh wow. That is a pretty farm :-)
Still, don't brick a single one of them, I suggest  

Have a look here, there is a guy in this forum who is really really nice, and helpful:
http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/page-22#post-3034516
Perhaps he has ideas that help you with your R9s?

[drakoin]

anybody know how to undervolt the R9 270x?
i haven't been able to find hacked bios's either.  

ya, i have seen flashes for 7xxx cards, but not for r9 2xxx cards.  oh well, i'll keep looking.

Hi Kalus, I guess this might make you happy:
https://bitcointalk.org/index.php?topic=395004.msg4363696#msg4363696

Looks as if he did use VBE7 for editing an R9 VGABIOS.

 

[drakoin]

anybody know how to undervolt the R9 270x?
i haven't been able to find hacked bios's either.  
ya, i have seen flashes for 7xxx cards, but not for r9 2xxx cards.  oh well, i'll keep looking.

Hi Kalus, I guess this might make you happy:
https://bitcointalk.org/index.php?topic=395004.msg4363696#msg4363696
Looks as if he did use VBE7 for editing an R9 VGABIOS.
 

And?