Dear ckolivas & kano!

I have an unusual request: I use cgminer to mine testcoins. The problem I'm having is that your software is too good and too fast even on a low-end laptop and causes an increase of difficulty. Competing with myself in mining doesn't make any sense.

Thus far I've used "--shares" switch to periodically mine a handfull of blocks, but the problem with this solution is that it causes thermal cycling and cycling of the fans on and off.

I was wondering if you could suggest a way to quickly detune the CGminer to a slower speed: I know how many miners I run and I culd just make sure that they in total don't go above 7Mhash/s (or the equivalent for scrypt).

If this is currently impossible to do, please let me know where to inserts the appropriate waits/sleeps in your code. I could then rebuild my own version.

Thank you for any suggestions you may have.

It's in the README ...

i'm looking for the best driver verisons for 2x7950
driver & SDKs.

thanks

--auto-fan and --auto-gpu are the commands that tell CGMiner to control you clocks and temps to maintain your target temp.  After these command you can then set your desired thresholds that CGMiner then attempts to maintain.  Rarely does CGMiner need to drop the clocks.  My target temp for one card is 75c and the other is 60c and they never have had the engine clock dropped.

Read the top post of this thread paying close attention to the executive summary examples.  I hear, unsubstantiated rumors that there is a readme file, or several, that are actually included with the software.  I have been too lazy to verify this for you but it may be worthwhile to look.

Good Luck,
Sam

README (hint: driver and SDK choice)

I have a puzzler. I am dabbling into Lightcoin mining and after setting up all major CPUs on my rigs to go for LTC, I decided to try one of my GPUs.
LTC on CPUs:
i7-3930K, @ 4.1GHz: ~80Khs (11 threads)
Q6600 @ 2.4GHz: ~19KHz

This is my reference for LTC speeds. GPUs should do better than CPUs, right?

My cards, currently mining for BTC are:
2xHD7970 (1150MHz, 300MHz, 1.140v, CrossFire): ~1300 Mhs together
2xHD6670 (960MHz, 400MHz, two different machines): ~127Mhs each
HD6570 (900MHz, 400MHz): ~117MHs
GTX 560 Ti (980MHz, 2400MHz, 1.062v): ~100MHs

All machines have Catalyst 13.1 with the bundled SDK installed.
2D Driver: 8.01.01.1266

Output from cgminer -n
 [2013-04-18 12:31:32] CL Platform 0 vendor: NVIDIA Corporation
 [2013-04-18 12:31:32] CL Platform 0 name: NVIDIA CUDA
 [2013-04-18 12:31:32] CL Platform 0 version: OpenCL 1.1 CUDA 4.2.1
 [2013-04-18 12:31:32] Platform 0 devices: 1
 [2013-04-18 12:31:32]  0       GeForce GTX 560 Ti
 [2013-04-18 12:31:32] CL Platform 1 vendor: Advanced Micro Devices, Inc.
 [2013-04-18 12:31:32] CL Platform 1 name: AMD Accelerated Parallel Processing
 [2013-04-18 12:31:32] CL Platform 1 version: OpenCL 1.2 AMD-APP (1084.4)
 [2013-04-18 12:31:32] Platform 1 devices: 2
 [2013-04-18 12:31:32]  0       Turks
 [2013-04-18 12:31:32]  1       Turks
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] Failed to ADL_Adapter_ID_Get. Error -1
 [2013-04-18 12:31:32] ADL found less devices than opencl!
 [2013-04-18 12:31:32] There is possibly more than one display attached to a GPU
 [2013-04-18 12:31:32] Use the gpu map feature to reliably map OpenCL to ADL
 [2013-04-18 12:31:32] WARNING: Number of OpenCL and ADL devices did not match!
 [2013-04-18 12:31:32] Hardware monitoring may NOT match up with devices!
 [2013-04-18 12:31:32] 2 GPU devices max detected
 [2013-04-18 12:31:32] USB all: found 12 devices - listing known devices
 [2013-04-18 12:31:32] No known USB devices

On PL 1: #0 is HD 6670, #1 is HD 6570
The screen is on nVidia; AMD cards are headless, used solely for mining.

Now that you have the reference speeds, here is what I experience with CGMiner, and Scrypt. I am trying things out on the HD6570 card, so as to see how things turn out and to not impact the ongoing BTC mining too much. I also see here that HD6570 can be pushed to give ~55Khs

When I run cgminer with almost all default setting, targeting this card, I get ~16Khs:
-T -l 1 --scrypt -g 2 -w 256 --lookup-gap 2 -I d --gpu-platform 1 -d 1

Adding --auto-gpu --thread-concurrency 5760 --gpu-memclock900 I get a whopping 18Khs, which is nowhere near the expected 55

The same story is on the other cards - they perform way below what I would expect to see, based on the tables in the link above.

Am I missing something? To me it looks like it is more profitable to use the cards for BTC mining and leave LTC to my CPUs...

Hey tiktoc,


Nope it runs dual 5870Ms stock speeds the cards should be getting 150mhash each and OCed 185ish mhash.

bump for help.... when i specify to use one card of the other, it seems only the left card gets used as the right side dosnt even get warm.. not sure if its a cgminer glitch, driver issue or hardware problem :-/ still trying to work it out.

just upgraded pc and now cgminer cant find cards.. other miners can

says all devices disabled.


using cgminer 2.4.4
amd drivers 13.1
sdk 2.5.. tried 2.8 as well


i'm guessing it is due to the intel opencl  on here?

If you use stratum and make over 1 diff1 share per second it will rise automagically.
You can addn or +n at end of username to get share diff=n
It is useful only on high power or you need lower data stream miner<>pool.

Oops, can`t compile latest git on windoze https://github.com/ckolivas/cgminer/commit/d5981f83804492a70a4aa309f8f83a200f88f478
ps. why "issues" are disabled on git?