Catalyst 13.4 screwed 7970, how to sort out?

[kjib]

For once I decided to give my 7970 a rest & play a game which informed me that my driver was out of date so without think I just clicked the upgrade.

It seems that Catalyst 13.4 simply blew my VTX HD 7970 bios away, disabled the over clocking and screwed up the custom fan settings. Why the hell don't AMD at least offer you the option to leave this stuff alone!

At this point I could still mine coins but SHA256 lost about 15% and scrypt about 50% with no ability to over clock. Even with the (now) hard wired fan setting the card ramped up the heat really quickly running 10+ degs hotter doing less.

Thinking I could get around this I re-blew the bios myself but I got the wrong rom. Doh! Realising this I blew it again with what I believe to be the correct one.

I also went back to Catalyst 13.1. MS After burner reports the ability to change settings on the card is now back so that's good.

Now as soon as I try to get CGminer to do some mining (scrypt or SHA256) Windows 7 informs me that CGminer has crashed.

One possibility is that I've semi-killed the card but I seem to be able to run Windows the same as usual, just I can't mine any more. I tried upgrading to the latest stable CGminer (3.1) and no joy.

Suggestions to find the solution anyone?

[LordTheron]

I had similar thing and in my case drivers reinstall or change wouldn't do anything but changed APP SDK from 2.8 to 2.6 and I could mine again , a bit slower than  usual so  reinstalled 2.8 again and its back to how it was before all problems. Give it a go and see if it works for you.

[kjib]

Thanks for the suggestion about the SDK but that made no difference. 

I did find this crash report, no idea what most of it is all about but it does refer to a module name:

Faulting Application Path:   C:\cgminer-3.1.0-windows\cgminer.exe

Problem signature
Problem Event Name:   APPCRASH
Application Name:   cgminer.exe
Application Version:   0.0.0.0
Application Timestamp:   517d4469
Fault Module Name:   StackHash_0a9e
Fault Module Version:   0.0.0.0
Fault Module Timestamp:   00000000
Exception Code:   c0000005
Exception Offset:   00000000
OS Version:   6.1.7601.2.1.0.256.1
Locale ID:   2057
Additional Information 1:   0a9e
Additional Information 2:   0a9e372d3b4ad19135b953a78882e789
Additional Information 3:   0a9e
Additional Information 4:   0a9e372d3b4ad19135b953a78882e789

Does that lot shed any light on the problem to anyone?

  Any suggestions welcome... 

[lbr]

give me 2 hours max, some reward and remote access(vnc/teamviewer) to ur pc and u will be hashing again at ur original rate or higher.

[Zanatos666]

System restore back to before you did the AMD update and see where that takes you.

[kjib]

give me 2 hours max, some reward and remote access(vnc/teamviewer) to ur pc and u will be hashing again at ur original rate or higher.

Well I have about 50 PPcoins but not in a position to mine anything new now! I can't offer access to the PC as it has work stuff on it & I'd get dragged over hot coals if something got trashed. If you have suggestions I'm happy to try them & if you provide the solution that works for me, give me a PPC address & they're yours.

[kjib]

System restore back to before you did the AMD update and see where that takes you.

Good suggestion but you guessed it, no restore to restore from. 

[kjib]

In case it helps, here is what Catalyst reports:

Code:

Driver Packaging Version       9.012-121219a-154961C-ATI
Catalyst Version	                        13.1
Provider	                                Advanced Micro Devices, Inc.
2D Driver Version	                        8.01.01.1278
2D Driver File Path                        /REGISTRY/MACHINE/SYSTEM/ControlSet001/Control/CLASS/{4D36E968-E325-11CE-BFC1-08002BE10318}/0001
Direct3D Version	                        9.14.10.0945
OpenGL Version                           6.14.10.12002
Catalyst Control Center Version      2012.1219.1521.27485
AMD Audio Driver Version	        7.12.0.7712

[lbr]

give me 2 hours max, some reward and remote access(vnc/teamviewer) to ur pc and u will be hashing again at ur original rate or higher.

Well I have about 50 PPcoins but not in a position to mine anything new now! I can't offer access to the PC as it has work stuff on it & I'd get dragged over hot coals if something got trashed. If you have suggestions I'm happy to try them & if you provide the solution that works for me, give me a PPC address & they're yours.

something won't get trashed cause u'll be sitting and starring at the monitor watching what I'm doing ; )
though video of s1 getting dragged over hot coals sounds like fun > : )

[Zanatos666]

Thanks for the suggestion about the SDK but that made no difference. 

I did find this crash report, no idea what most of it is all about but it does refer to a module name:

Faulting Application Path:   C:\cgminer-3.1.0-windows\cgminer.exe

Problem signature
Problem Event Name:   APPCRASH
Application Name:   cgminer.exe
Application Version:   0.0.0.0
Application Timestamp:   517d4469
Fault Module Name:   StackHash_0a9e
Fault Module Version:   0.0.0.0
Fault Module Timestamp:   00000000
Exception Code:   c0000005
Exception Offset:   00000000
OS Version:   6.1.7601.2.1.0.256.1
Locale ID:   2057
Additional Information 1:   0a9e
Additional Information 2:   0a9e372d3b4ad19135b953a78882e789
Additional Information 3:   0a9e
Additional Information 4:   0a9e372d3b4ad19135b953a78882e789

Does that lot shed any light on the problem to anyone?

  Any suggestions welcome... 

Update to the latest version of cgminer, 3.2.2, see if that doesnt change something.

[kjib]

I tried running CGMiner with verbose output and even without doing anything (I was doing this):

Code:

cgminer --benchmark -D --verbose 2>logfile.txt 

this is the entire extent of the logfile:

Code:

 [2013-06-18 21:13:10] Started cgminer 3.1.0

Clearly this is not giving much away but I guess it does tell us that it's quite early on in what cgminer does to set things up...

[kjib]

Update to the latest version of cgminer, 3.2.2, see if that doesnt change something.

Same problem with 3.2.2 but thanks for the suggestion.

[zyk0]

Try
a) deleting the .bin files from cgminer's directory (cached compiled opencl kernels)

b) making sure you have matching openCL drivers for version.

1) uninstalling Catalyst 13.1
2) manually deleting amdocl*.* from C:\Windows\System32 and if x64 C:\Windows\SYSWOW64\
3) reinstalling Catalyst 13.1

AMD likes to leave those files behind and not replace newer versions.

[kjib]

I tried following the CGminer debug instructions which work but did not generate anything that means a lot to me:

Code:

cgminer_debug.exe caused an Access Violation at location 7564c9f1 Reading from location 00000000.

Registers:
eax=0028f198 ebx=02af200c ecx=0028f1a4 edx=02aa8a28 esi=02af1fe8 edi=00000000
eip=00000000 esp=0028f184 ebp=0028f19c iopl=0         nv up ei pl zr na po nc
cs=0023  ss=002b  ds=002b  es=002b  fs=0053  gs=002b             efl=00010246

Call stack:
00000000
5209F938  amdocl.dll:5209F938
521CC76F  amdocl.dll:521CC76F  clGetSamplerInfo
521CCBFA  amdocl.dll:521CCBFA  clGetSamplerInfo
521CE190  amdocl.dll:521CE190  clGetSamplerInfo
521D614A  amdocl.dll:521D614A  clGetSamplerInfo
521D621A  amdocl.dll:521D621A  clGetSamplerInfo
521C5356  amdocl.dll:521C5356  clGetSamplerInfo
52198321  amdocl.dll:52198321  clIcdGetPlatformIDsKHR
654A1081  OpenCL.dll:654A1081
654A3F37  OpenCL.dll:654A3F37  clGetExtensionFunctionAddress
654A1219  OpenCL.dll:654A1219  clGetPlatformIDs
0043B6A0  cgminer_debug.exe:0043B6A0
0043A180  cgminer_debug.exe:0043A180
0041EE96  cgminer_debug.exe:0041EE96
004010B9  cgminer_debug.exe:004010B9  __mingw_CRTStartup  crt1.c:244

00401284  cgminer_debug.exe:00401284  WinMainCRTStartup  crt1.c:274

763B33AA  kernel32.dll:763B33AA  BaseThreadInitThunk
77CD9EF2  ntdll.dll:77CD9EF2  RtlInitializeExceptionChain
77CD9EC5  ntdll.dll:77CD9EC5  RtlInitializeExceptionChain

If it means something to you, let us know...

[kjib]

Try
a) deleting the .bin files from cgminer's directory (cached compiled opencl kernels)

b) making sure you have matching openCL drivers for version.

1) uninstalling Catalyst 13.1
2) manually deleting amdocl*.* from C:\Windows\System32 and if x64 C:\Windows\SYSWOW64\
3) reinstalling Catalyst 13.1

AMD likes to leave those files behind and not replace newer versions.

I give that a go, thank you.

[LordTheron]

Do you have any errors or warnings  in system or application logs?

[kjib]

Do you have any errors or warnings  in system or application logs?

No warnings etc, no.

In case anyone is getting excited about the offer of PP coins, sorry, it is Terra coins (TRC) that I have. I don't think there's a massive difference in value anyway....

[kjib]

Try
a) deleting the .bin files from cgminer's directory (cached compiled opencl kernels)

b) making sure you have matching openCL drivers for version.

1) uninstalling Catalyst 13.1
2) manually deleting amdocl*.* from C:\Windows\System32 and if x64 C:\Windows\SYSWOW64\
3) reinstalling Catalyst 13.1

AMD likes to leave those files behind and not replace newer versions.

zyk0, you are the winner. Let me have a TRC address if you'd like some coins.

On the downside, I am still not back up to full speed with scrypt mining (only about 400K when 6-700 sounds more like the norm). Any further suggestions on that one, (it's slower than it was before I installed 13.4 and now that's gone along with my KH  ).

[zyk0]

zyk0, you are the winner. Let me have a TRC address if you'd like some coins.

On the downside, I am still not back up to full speed with scrypt mining (only about 400K when 6-700 sounds more like the norm). Any further suggestions on that one, (it's slower than it was before I installed 13.4 and now that's gone along with my KH  ).

Just glad I could get you 1/2 way there.
What's your worksize, gpu-thread-concurrency, engine&memory clocks, gpu-threads & intensity set to?
and is the GPU_MAX_ALLOC_PERCENT envrionment variable set to 100?

[kjib]

zyk0, you are the winner. Let me have a TRC address if you'd like some coins.

On the downside, I am still not back up to full speed with scrypt mining (only about 400K when 6-700 sounds more like the norm). Any further suggestions on that one, (it's slower than it was before I installed 13.4 and now that's gone along with my KH  ).

Just glad I could get you 1/2 way there.
What's your worksize, gpu-thread-concurrency, engine&memory clocks, gpu-threads & intensity set to?
and is the GPU_MAX_ALLOC_PERCENT envrionment variable set to 100?

GPU_MAX_ALLOC_PERCENT is at 100.
For SHA256 I have Core clock = 1160MHz, Memory clock = 1050 and everything else on default.
For scrypt I typically have Core clock = 1130MHz, Memory clock = 1550 and  -I 13 --shaders 2048 --thread-concurrency 8192 --lookup-gap 2 -g 4 -w 256