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

[sharky112065]

Is there a way to set the OpenCL platform per card? Cgminer keeps crashing with a mixed 5xxx and 6xxx rig.

I'm not sure as to why it is crashing on you. I have a rig that has 3 5870's and 1 6970, and it runs just fine. Cgminer also automatically generates .bin's for each type of card (Caymen and Cypress).

Anyway, right out of the freaking manual (README)...

Q: I have multiple SDKs installed, can I choose which one it uses?
A: Run cgminer with the -n option and it will list all the platforms currently
installed. Then you can tell cgminer which platform to use with --gpu-platform.

--kernel|-k <arg>   Override kernel to use (diablo, poclbm, phatk or diakgcn) - one value or comma separated

I'm trying to use the 2.1 SDK for the 5xxx and 2.6 for the 6xxx. Everything works fine using the 2.6 sdk, but it obviously isnt efficient for the 5xxx series.
Phoenix 2.0 lets me use the setup I want, but it lacks lots of features.

I also tried making cgminer use bins compiled for 2.1 while using the 2.6 sdk but it just recompiles them again. The ideal would be making --gpu-platform a per device option, something like --gpu-platform 0,0,1

I am using 2.5 SDK for 3 5870's and 1 6970 (960/170 for all cards). I am getting 428 Mh/s on the 6970, and 442 Mh/s on the 5870's. Now if your 5xxx cards are not cypress then I guess you would want a lower version of the SDK. I let Cgminer create it's own .bin's (phatk).

You could use the -d option to force Cgminer to create a .bin for each type of card and then maybe use the -k option using comma's to specify more than one. I have never had to do any of that funky stuff so I really cannot help you with mixed SDK's.

[kano]

I'm trying to compile cgminer for windows with CPU mining enabled, but I can't get it to link when I specify -msse2.
It seems that sha256_xmm.o isn't being included.
I'm getting the following error:

Code:

  CCLD   cgminer.exe
cgminer-sha256_sse2_i386.o:sha256_sse2_i386.c:(.text+0x155): undefined reference
 to `@CalcSha256_x86@12'
cgminer-sha256_sse2_i386.o:sha256_sse2_i386.c:(.text+0x16a): undefined reference
 to `@CalcSha256_x86@12'
collect2: ld returned 1 exit status

without -msse2 it compiles fine, but of course, without the SSE2 optimised code.

Any help to kick this into action???

You have to edit the code to fix that
CPU mining isn't supported.

[P_Shep]

The code looks OK, and digging into the Makefile, it's OK too. I can't work out why it doesn't work :/

[kano]

The code looks OK, and digging into the Makefile, it's OK too. I can't work out why it doesn't work :/

If you remove the --enable-cpumining you added it will work fine in windows.
Yes even without the ".exe" showing in that line I'd know that is a windows CPU mining link error.

[P_Shep]

I know it'll compile fine without CPU mining, but I want to do some testing with it enabled... I want to see how Anubis handles it. If it's more trouble to get it working than it's worth, I'll forget about supporting CPUs in Anubis.

[Vbs]

Anyone can now build Cgminer on Windows.

http://pastebin.com/3pzivj32

I spent a lot of time documenting and testing the documentation step by step on both Windows 7 and Windows XP. Shouldn't take more than 30 min to set up the build environment.

Thanks for taking the time to write this!

[DeathAndTaxes]

I'm trying to use the 2.1 SDK for the 5xxx and 2.6 for the 6xxx. Everything works fine using the 2.6 sdk, but it obviously isnt efficient for the 5xxx series.
Phoenix 2.0 lets me use the setup I want, but it lacks lots of features.

I also tried making cgminer use bins compiled for 2.1 while using the 2.6 sdk but it just recompiles them again. The ideal would be making --gpu-platform a per device option, something like --gpu-platform 0,0,1

You can't. 

Why do you want to use 2.6?  It is junk for 5x and 6x series.  Only reason to use 2.6 is if you have 7x series card.

2.1 is slightly better w/ 5x series but only with low memclock (and it is small).  Both 5x & 6x work well with 2.4 & 2.5 so that is the simplest answer.

[os2sam]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?
Thanks,
Sam

[DeathAndTaxes]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

[Vbs]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

Why not use 12.1? As long as you don't install the SDK runtime that comes with it, it's working 100% with my 5850's.

[os2sam]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.

Catalyst 11.6 was before the bug was introduced, which is why I was using it.

Why not use 11.11?

I don't know why not, was the 100% CPU utilization for multiple GPU's fixed in that version?  When I was messing with this installation  before I had a terrible time get later Catalyst versions installed and functioning at all which was why I was using 10.7 and 10.9 at that time to begin with.
Thanks,
Sam

[DeathAndTaxes]

I don't remember which version it was fixed in.

>11.6 and <=11.11 

I know it is fixed in 11.11.  As pointed out above you can even use 12.1 just make sure to install Runtime/SDK FIRST and then do CUSTOM SETUP and the uncheck OpenCL Runtime.  Once installed 2.6 is a pain to get rid of.  That is why I suggested 11.11.  It is a version I know has no 100% CPU bug and it doesn't have 2.6 so no chance of accidentally installing it.

[os2sam]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

Why not use 12.1? As long as you don't install the SDK runtime that comes with it, it's working 100% with my 5850's.

What SDK are you using with it?

I guess I'll have to give that a try.  I'm still perplexed why the 11.6, the last know good version, doesn't work on Win7, or is just my Win7 installation?
Thanks,
Sam

[os2sam]

I don't remember which version it was fixed in.

>11.6 and <=11.11 

I know it is fixed in 11.11.  As pointed out above you can even use 12.1 just make sure to install Runtime/SDK FIRST and then do CUSTOM SETUP and the uncheck OpenCL Runtime.  Once installed 2.6 is a pain to get rid of.  That is why I suggested 11.11.  It is a version I know has no 100% CPU bug and it doesn't have 2.6 so no chance of accidentally installing it.

Sounds like my best shot.  I'll give a whirl over the weekend.
Thanks again,
Sam

[DeathAndTaxes]

What SDK are you using with it?

Generally speaking.

2.1 is best for 5000 series cards.
2.4 or 2.5 is best for 6000 series cards.
2.6 is best for 7000 series cards.

That being said the difference between 2.1 and 2.4/2.5 is minimal.

2.6 is a huge performance hit w/ low memclocks I mean 20%+ so avoid it like the plague unless you have a 7000 series card.

Really no reason to use 2.2 or 2.3 for anything.

[Vbs]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

Why not use 12.1? As long as you don't install the SDK runtime that comes with it, it's working 100% with my 5850's.

What SDK are you using with it?

I guess I'll have to give that a try.  I'm still perplexed why the 11.6, the last know good version, doesn't work on Win7, or is just my Win7 installation?
Thanks,
Sam

I'm using the SDK runtime 2.5.793.1 that comes with the 11.11 drivers (it's the newest version of 2.5 sdk available), since I'm using the phatk kernel. If I wanted to use the poclbm kernel, I would use SDK 2.1.

I did a quick test sometime ago, and the phatk GPU ISA code generated with SDK 2.4.650.9 was the same as with SDK 2.5.793.1, so they "should" have identical performance, at least for the phatk kernel.

What SDK are you using with it?

Generally speaking.

2.1 is best for 5000 series cards.
2.4 or 2.5 is best for 6000 series cards.
2.6 is best for 7000 series cards.

That being said the difference between 2.1 and 2.4/2.5 is minimal.

2.6 is a huge performance hit w/ low memclocks I mean 20%+ so avoid it like the plague unless you have a 7000 series card.

Really no reason to use 2.2 or 2.3 for anything.

Damn you! You're increasing my itch of comparing the low-level code of poclbm + 2.1 vs phatk + 2.5...

On a side note, on series 5xxx and 6xxx, using poclbm with any sdk above 2.1 is a performance loss, as is using phatk with anything below 2.4.

[cablepair]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

Why not use 12.1? As long as you don't install the SDK runtime that comes with it, it's working 100% with my 5850's.

What SDK are you using with it?

I guess I'll have to give that a try.  I'm still perplexed why the 11.6, the last know good version, doesn't work on Win7, or is just my Win7 installation?
Thanks,
Sam

sam: install 11.9 if your using win7 with 5xxx series cards - it does not get better than that and believe me I have done the testing

[os2sam]

OK, I pretty sure this is ancient history, but here goes anyway.

I've been using Windoze XP with Catalyst 11.6 with a 5830 and 5770 and had low CPU utilization with that combination.  I went back to my Win7 install where I was using 10.9 to get low CPU utilization prior to CGMiner 2.0.  I updated this Win7 to Catalyst 11.6 and now have 100% CPU utilization.  I had thought Catalyst 11.6 had low CPU utilization across all platforms?

Any thoughts?  Or is this just the way it works with Windoze 7?

IIRC was a lot later than 11.6 before 100% bug was solved in Windows.  Why not use 11.11?

Why not use 12.1? As long as you don't install the SDK runtime that comes with it, it's working 100% with my 5850's.

What SDK are you using with it?

I guess I'll have to give that a try.  I'm still perplexed why the 11.6, the last know good version, doesn't work on Win7, or is just my Win7 installation?
Thanks,
Sam

sam: install 11.9 if your using win7 with 5xxx series cards - it does not get better than that and believe me I have done the testing

Well, I'll run it up the flagpole and see what happens.
Thanks,
Sam

[Dyaheon]

On a side note, on series 5xxx and 6xxx, using poclbm with any sdk above 2.1 is a performance loss, as is using phatk with anything below 2.4.

Really? Using 960/300 clocks on a 5870, -g 1 -I 8 (p2pool), tried with and without -w 256 -v 2, I get much worse performance on poclbm + 2.1 compared to phatk + 2.1.

Phatk + 2.1 and phatk + 2.4 is about the same though, with 2.4 taking a very slight lead (<1%) .

What would you recommend as settings for poclbm? I do remember it being better back in the day too, but haven't played much with 2.1 lately. Also, keeping memory speeds low is a must 

[Vbs]

On a side note, on series 5xxx and 6xxx, using poclbm with any sdk above 2.1 is a performance loss, as is using phatk with anything below 2.4.

Really? Using 960/300 clocks on a 5870, -g 1 -I 8 (p2pool), tried with and without -w 256 -v 2, I get much worse performance on poclbm + 2.1 compared to phatk + 2.1.

Phatk + 2.1 and phatk + 2.4 is about the same though, with 2.4 taking a very slight lead (<1%) .

What would you recommend as settings for poclbm? I do remember it being better back in the day too, but haven't played much with 2.1 lately. Also, keeping memory speeds low is a must 

Phatk was specifically written for SDK 2.4+, so compiling it with anything lower than that is a performance loss, since it has specific tweaks for the SDK 2.4+ compilers.

I don't recommend anything for poclbm, since I don't use it. I have two 5850's on two different machines (free power ), both running at stock voltage, one at 970/150 (~400MH/s) and the other at 980/150 (~404MH/s), using phatk V2 W256 I9.

Both are on Win7 x64, Catalyst 12.1 drivers and SDK runtime 2.5.793.1 (comes on the 11.11 drivers).