Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 783. (Read 5805728 times)

Strange, for me each core on my 5970's can be disabled seperately and correctly ([g] [d] 0, [g] [d] 1).

I believe I read it's supposed to already. Could be a bug or older version?

Windows? Make shortcut and size the font down or increase the lines height from 25 to 50.

You can, comma's, just have to know which ones are enumerated which. (--gpu-clock 1000,440,750,950) for example.

CARDbased, would be nice to see, but there'd have to be a way to check by manufacturer and such too, I know some 5970's by like MSI have different heatsink/fan combinations than say PowerCooler or something, and they might not be tolerant the same way but still show up as 5970 by their ID. Don't know how you'd be specific with setting that maybe with grouping (--gpu-group {5970@950+300+1.5})?

A range with a dash in it would be cool. (--gpu-memory 200-700)!

Well if he's already parsing IN the config file, maybe if you specify a config, and change something inside it can be exported back out in addition tacking on these extra settings in the file? (--config myconfig.json) could get a line with card serial numbers or something unique? ({"_Safe_CARDID": "CLOCK,MEMORY,FAN"})

Well of course.  Doesn't everyone?

Still running OS/2 Warp 4 on a really old Thinkpad.  The current versions are now called eComStation and I have that on my Personal Laptop and on a VPC on my company laptop.  Having trouble finding a bitcoin miner though .
Thanks for asking,
Sam

+1

p.s. os2sam: do you really still run os/2?

I remember getting OS/2 Warp for Christmas one year when I was like 12 or 13, too bad for IBM win'95 came out that summer ;/

+1

Makes sense...

Default should be 0, where:
0 = disabled function, no exit after GPU failure

At Windows this batch file would make it:

I have a feature request.
Every once in a while i have a GPU that can't be restarted...
but restarting the CG miner works fine.

I was wondering if you could add a cmd switch that exits after a GPU failure.
i.e --gf # ( where # is number of GPUs to fail before exiting the CGMiner.exe)

this would save me on constantly having to check on the miner to make sure that all GPU's are constantly mining

Thanks.

In Windoze I created a shortcut and changed the layout properties to 55 rows (Window height) instead of the default 25.  Now I can see the info for both of my GPU's at once.

You can also modify the pixels per character under the Fonts tab in conjunction with the layout window height to get more window realestate.

Sam

Are you having trouble in that department, or did you just miss this?

Sorry if I'm stating the overly obvious, but in the ATI CCC did you unlock the overclocking page?  When I first started messing with this stuff I looked at the overclock page a bunch of times before I realized that the lock was actually a button.  I was really irritated about it being grayed out too.
Sam

This should probably identify which GPU it's talking about, and maybe have some kind of throttling added to it, if a card's temperature is wiggling around the threshold.

^^^ That's the ticket.  Plus keep track of how long it was able to run at that clock rate and use that info to drive the adjustments.  I've got a bunch of cards that don't like to run for more than 20-30 mins at elevated clocks, and it can take a couple days to dial them in.

The new GPU features are awesome! A few suggestions/requests:


1) Dual GPU cards don't get adjusted/disabled correctly. For example, a 5970 will only disable the one GPU that has the temp sensor:


2) It'd be awesome if cgminer would record the current clocks on startup, and restore them on exit if the auto tuning changed them at all.

3) Pressing "G" when you have more than a couple of cards makes it impossible to read the output, because the window the output is displayed to is too small unless you have a HUGE screen.

4) I'd love a way to specify different temperature thresholds per GPU on the command line. If I have different model cards in there, they have different points where they're happy. 5770s get crashy above 90-95, where 5970 and 6990 cards idle near there at times.

5) My ideal dream would be a way of somehow saying "Any 5970 cards you see, set the temperature thresholds to X/Y, the voltage to Z, etc. Any 5770 cards, the temperature threshold is..." so that I don't have to look up which cards are in which system, just to pass that along to cgminer.

6) Temperatures >100C should be allowed, no matter how bad of an idea that sounds. We have some cards that go up to 105-107C without issue.

7) Specifying an overclock/underclock range that cgminer is allowed to adjust the clock in would be handy.

One step further, having it attempt to determine (maybe even saved into a local file) how high the clock was able to go without problems, and self-tuning the max clock rate while under the threshold temperature.

Can we not test this with some clever CUDA/OpenCL code? run a light gpu loop thats heavy on memory operations for X amount of seconds, count ops/sec and write back out of gpu to the cpu to write to disk, then throttle down the memory more and repeat, you should be able to see a bottleneck at some point?

(in a slightly off topic but related issue, I can't clock my card, I got drivers installed and can mine, but EVERY clocking app {cgminer, overdrive, ccc (wont even start), clock tool} either doesn't run or has all sliders grayed out... nobody's helping in the technical forums D:> tried different drivers and ccc versions 11.5 thru 11.8 and it can't be messed up installs, I got frustrated and even reformatted and installed windows over and fresh installed the drivers for two of the versions .6 and .

When I build 2.0.0 (with the original or newer -1 source), it works fine as long as I don't add the ADL header files to the ADL_SDK folder.  When I do that, I get this at the end of the make step:This is running Fedora 15 with GLIBC 2.14-4.  I'm guessing that is the problem, and I'm guessing it is due to the AMD header files and outside of ck's control, but wanted to report it just in case.  I may look for some repo with a newer GLIBC at some point if I can find time, but in the meantime, FYI:  It would appear that you have to have a pretty recent version of GLIBC to compile the gpu monitoring support.

EDIT:  I actually resolved this issue by defining LDFLAGS to point at ..../ati-stream-sdk-v2.1-lnx64/lib/x86_64/

I guess that means my guess was wrong.

For further clarification:  To get this running on Fedora 15 x86_64, I am running ./configure, gathering the CFLAGS and LDFLAGS settings from Makefile, and re-running ./configure with CFLAGS using those settings plus a -I..../ati-stream-sdk-v2.1-lnx64/..../includes/ setting and LDFLAGS using those settings plus a -L..../ati-stream-sdk-v2.1-lnx64/lib/x86_64/ (where the .... sections indicate paths that may vary per machine that I don't remember and can't see at this very moment).  Finally, because Fedora 15's JSON library is too old, I am editing the Makefile to use the source's included JSON instead of my installed JSON (I haven't tried setting JSON_INCLUDES for .lconfigure yet).

Well for one thing ATI CCC is completely useless IMO anyway.

But
you may have an interesting point on that cgminer cannot downclock the memory, is it because it cant downclock past 300? You could be right. I was trying to downclock my mem to 180 which is the ideal spot for the cards I was testing this on.

I dont think it has anything to do with power though, this rig has a corsair 1200 watt PSU

but I bet your right on with the downclock thing

ckolivas: can you verify that CGMINER can't downclock mem past 300? This would be a feature I would very much like. I know the main stream thought is 300 is the sweet spot for mem, and I think this myth exists because it is the lowest point you can downclock in some early versions of software.

In any case I have 9 GPU's and have been mining since BTC was worth 0.85 USD - my point being?

I have thoroughly tested these cards every which way possible, and although 1 of my cards prefers 300 for memclock (my xfx 5830 which I hate BTW) all my 6870's and 5870's LOVE 180 plus I am getting some energy savings there (even if its not a lot it adds up)

I'm using WinXP with 11.6 and am using it to overclock a Radeon 5770 and 5830 to 950Mhz and underclock the memory to 300MHz and I verified that the settings took with GPU Shark.  I could not do that with the ATI CCC.  And my hash rate has improved by 60 to 70Mhs for the pair of GPU's.
Sam

Cgminer cannot down clock the memory as much as MSI Afterburner or Trixx on some cards. So if that is happening and you do not have enough power you may be exceeding the power you need to get a steady Mhash out of you cards. That is what happened to me anyway.

The developer tells me it is because he uses the ATI stuff to change settings. MSI Afterburner and Trixx bypass the ATI stuff and change some settings on their own directly.

It's while the GPU code is executing that the CPU usage is high due to the driver consuming useless cycles. Sleeping when it comes back to the CPU will do nothing for that.

just out  of curiosity... can't a Sleep(1); be added into each thread? this would fix the 100% cpu bug?
1ms shouldn't affect the mining at any signifigant level

Ooopss! Thanks a lot!