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

[Nadia]

It's in the readme.

Wow thanks. I suppose it works for all parameters. Gonna try it. Thank you. 

[dbabo]

Yes, solo submits any share >64k on litecoin mining.

CK,
 does it mean i have to limit the submitting value? If so, what value it should be and how to limit it?
thank you

It's just reporting information. It's as if you went up to the counter to pay $88,800 for a new toy that costs $31,621,939 and the clerk said "NO!"

_{I think}

ok. So why does it do it occasionally then?

[collapse]

Compiled for 64bit windows, someone knows if cgminer has better performance or consumes less resources that 32bit?

[Marrs]

Compiled for 64bit windows, someone knows if cgminer has better performance or consumes less resources that 32bit?

With a 64-bit binary, performance will be the same because all of the work happens on the GPU/FPGA/ASIC and it will use more resources in terms of disk space and memory (although only trivially more).

[collapse]

Compiled for 64bit windows, someone knows if cgminer has better performance or consumes less resources that 32bit?

With a 64-bit binary, performance will be the same because all of the work happens on the GPU/FPGA/ASIC and it will use more resources in terms of disk space and memory (although only trivially more).

in scrypt, Cgminer puts hexacore 3930K 4.11Ghz at 34% of full power..., mining in GPU

[Eric Rye]

the last two characters of the .bin file changed from l8 on Linux to l4 on windows, otherwise they're the same

The compiled opencl binary file name for scrypt is defined as:
name + kernelname + g + lg + lookup_gap + tc + thread_concurrency + w + work_size + l + sizeof(long) + .bin

l4 indicates 32-bit (4 byte) longs whereas l8 indicates 64-bit (8 byte) longs. This is unlikely to be related to your speed discrepancy.

Thanks for the reply. All I can really try is rolling back the drivers I suppose.

[Peleus]

As crazy as it sounds, I tend to get different performance if I start the miner in person vs over VNC.

In person it usually averages about 500 kH/s per card.

Over VNC it drops to around 400 kH/s per card.

Would it have something to do with the screen that it's using when starting? Should I take Display=:0 out of the script?

Any thoughts would be interesting, cheers.

[daemondazz]

Would it have something to do with the screen that it's using when starting? Should I take Display=:0 out of the script?

If you take the DISPLAY=:0 out of your script, you won't be doing any mining, that's how the program gets access to the graphics card

I have noticed with my laptop (5450M) that if I start cgminer on the console, it seems to push the card harder and causes it to overheat and stop after ~10 minutes. If I start it via RDP, then it doesn't push it quite so hard, so doesn't overheat and hence keeps running indefinitely.

[2112]

Competing with myself in mining doesn't make any sense.

You aren't "competing with yourself."  Difficulty is programed to adjust periodically to a target of 6 blocks per hour.  Mining slower to only get 6 blocks per hour will prevent a difficulty change, but won't change how often you get a block vs mining faster once difficulty is done adjusting.  If you want difficulty to stay low, you could CPU mine (although not with cgminer).  If you need more than 6 blocks per hour, maybe modifying the code for your testchain would make more sense.  If you want to use less power, then it probably makes sense to "detune" your video card.  In that case, I would recommend starting with the tuning suggestions in the readme and trying to use logic to get the results you want (IE, underclock the video card).

Well, underclocking the video card is a quite involved process, especially on a laptop. It seems like the AMD drivers keep the set of clock settings per power state of the OS. So when underclocked while plugged in the machine upclocks when unplugged and put on the battery power. And vice versa: when underclocked on the battery it upclocks when plugged in.

The goal is to be able to mine at least one blockchain without triggering the increase of the difficulty and without raising the temperatures of the CPU & GPU to trigger the internal fan of the laptop.

With the fan working I could mine at least three competing branches of the blockchain without raising the difficulty; all on the low-end laptop. This is really helpful for the reorganization testing, but not really required for the usual tests.

I absolutely want to mine against the official binaries without changing anything or even a recompilation. This is an explicit goal of the test process: to successfully interface with the official Satoshi clients.

Mining on the CPU even with a single core always results in the fans spinning and produces blocks at the annoyingly slow rate that would needlessly prolong the test runs.

Yes, I fully understand that cgminer is 99.99% designed for the production mining and not for the development mining. I have to rethink how to cover the remaining 0.01% use case: R&D on the testnet.

I'll maybe need to think about a creative use of cgminer API to slow it down or maybe come up with a simple slow-it-down mining proxy server.

[-ck]

Well I tried to search for this but came up empty. Currently solo farming for feathercoin when I see this:

Code:

Rejected 000001f9 Diff 8.49M/5432796 BLOCK! GPU 1 pool 0

What does this mean? I solved for higher than the current diff. Did some one else beat me to the punch?

If blocks come in quickly on your altchain, and you were mining solo on your *coind then it doesn't have a way of telling you blocks have changed (like longpoll or stratum), so you risk mining on old blocks and increasing rejects. Lots of ways to improve this, someone else might have time to tell you.

[Tia]

Anyone have an idea what this might be indicative of? Suggestions?
This machine is W8 Pro, 7850 + 7970 and CGminer 2.11.4
Pool 0: eligius.st:3334 Pool1: mint.bitminter.com:8332 Pool2: eligius.st:80 Pool3: eligius.st:8337 And I believe confirmed this log was while I was running with Pool Load Balancing Failover only setting, while running on Pool 1. I am wondering if it was just CGMiner trying to re-check the availability/viability of Pool2 and essentially failing, then trying again (see this error multiple times).
 [2013-05-02 20:07:53] ALL (5s):1.043G (avg):989.3Mh/s | A:125  R:6  HW:0  U:12.6/m  WU:13.6/m
 [2013-05-02 20:07:53] GPU1  73.0C 3392RPM | (5s):716.7M (avg):661.3Mh/s | A:94 R:6 HW:0 U:9.5/m I:15
 [2013-05-02 20:07:55] GPU0  73.0C 2047RPM | (5s):330.9M (avg):331.6Mh/s | A:31 R:0 HW:0 U:3.1/m I:15
 [2013-05-02 20:07:57] Setting GPU 0 fan percentage to 48
 [2013-05-02 20:07:57] Setting GPU 1 fan percentage to 74
 [2013-05-02 20:07:58] ALL (5s):1.029G (avg):989.4Mh/s | A:125  R:6  HW:0  U:12.5/m  WU:13.5/m
 [2013-05-02 20:07:58] GPU1  73.0C 3406RPM | (5s):743.5M (avg):660.8Mh/s | A:94 R:6 HW:0 U:9.5/m I:15
 [2013-05-02 20:07:59] Setting GPU 0 fan percentage to 47
 [2013-05-02 20:08:03] ALL (5s):950.3M (avg):987.9Mh/s | A:125  R:6  HW:0  U:12.4/m  WU:13.4/m
 [2013-05-02 20:08:03] Setting GPU 0 fan percentage to 48
 [2013-05-02 20:08:03] Setting GPU 1 fan percentage to 73
 [2013-05-02 20:08:04] GPU1  74.0C 3398RPM | (5s):722.5M (avg):660.4Mh/s | A:94 R:6 HW:0 U:9.3/m I:15
 [2013-05-02 20:08:05] Testing pool http://eligius.st:80
 [2013-05-02 20:08:05] HTTP request failed: necessary data rewind wasn't possible
 [2013-05-02 20:08:05] GPU0  74.0C 2173RPM | (5s):337.0M (avg):331.1Mh/s | A:31 R:0 HW:0 U:3.1/m I:15
 [2013-05-02 20:08:05] HTTP request failed: necessary data rewind wasn't possible
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
 [2013-05-02 20:08:05] Setting GPU 0 fan percentage to 47
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
 [2013-05-02 20:08:05] Closing socket for stratum pool 2

[nathanrees19]

Anyone have an idea what this might be indicative of? Suggestions?
This machine is W8 Pro, 7850 + 7970 and CGminer 2.11.4
Pool 0: eligius.st:3334 Pool1: mint.bitminter.com:8332 Pool2: eligius.st:80 Pool3: eligius.st:8337 And I believe this log was while I was running with Pool Load Balancing setting
 [2013-05-02 20:07:53] ALL (5s):1.043G (avg):989.3Mh/s | A:125  R:6  HW:0  U:12.6/m  WU:13.6/m
 [2013-05-02 20:07:53] GPU1  73.0C 3392RPM | (5s):716.7M (avg):661.3Mh/s | A:94 R:6 HW:0 U:9.5/m I:15
 [2013-05-02 20:07:55] GPU0  73.0C 2047RPM | (5s):330.9M (avg):331.6Mh/s | A:31 R:0 HW:0 U:3.1/m I:15
 [2013-05-02 20:07:57] Setting GPU 0 fan percentage to 48
 [2013-05-02 20:07:57] Setting GPU 1 fan percentage to 74
 [2013-05-02 20:07:58] ALL (5s):1.029G (avg):989.4Mh/s | A:125  R:6  HW:0  U:12.5/m  WU:13.5/m
 [2013-05-02 20:07:58] GPU1  73.0C 3406RPM | (5s):743.5M (avg):660.8Mh/s | A:94 R:6 HW:0 U:9.5/m I:15
 [2013-05-02 20:07:59] Setting GPU 0 fan percentage to 47
 [2013-05-02 20:08:03] ALL (5s):950.3M (avg):987.9Mh/s | A:125  R:6  HW:0  U:12.4/m  WU:13.4/m
 [2013-05-02 20:08:03] Setting GPU 0 fan percentage to 48
 [2013-05-02 20:08:03] Setting GPU 1 fan percentage to 73
 [2013-05-02 20:08:04] GPU1  74.0C 3398RPM | (5s):722.5M (avg):660.4Mh/s | A:94 R:6 HW:0 U:9.3/m I:15
 [2013-05-02 20:08:05] Testing pool http://eligius.st:80
 [2013-05-02 20:08:05] HTTP request failed: necessary data rewind wasn't possible
 [2013-05-02 20:08:05] GPU0  74.0C 2173RPM | (5s):337.0M (avg):331.1Mh/s | A:31 R:0 HW:0 U:3.1/m I:15
 [2013-05-02 20:08:05] HTTP request failed: necessary data rewind wasn't possible
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
 [2013-05-02 20:08:05] Setting GPU 0 fan percentage to 47
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
 [2013-05-02 20:08:05] Closing socket for stratum pool 2

Looks like a pool problem.

[Tia]

[2013-05-02 20:07:53] ALL (5s):1.043G (avg):989.3Mh/s | A:125  R:6  HW:0  U:12.6/m  WU:13.6/m
 [2013-05-02 20:08:05] HTTP request failed: necessary data rewind wasn't possible
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
 [2013-05-02 20:08:05] Setting GPU 0 fan percentage to 47
 [2013-05-02 20:08:05] JSON decode failed(1): '[' or '{' expected near '<'
[/quote]
Looks like a pool problem.
[/quote]

Well, to honest this is the first time I've tried port 80 on Eligius. I've been plagued with problems on Eligius pool lately. Had been running pretty solid via :8332, then out of the blue myself and many others ran into problems there with terrible hashrates, stales, unable to retrieve work, and so on. Was just playing with the options of other ports there to see if anything works better. So far I'd say the stratum pool at :3334 seems best but not very good for me. But I digress, not trying to get off topic for this thread, sorry. I was wanting to make sure I wasn't missing something in my config with CGminer.

Here is my call via batch file: cgminer -o http://eligius.st:3334 -u xXXxxXXxxx -p x -I 14,15 --kernel diablo --gpu-memclock 1180,1190 --gpu-engine 800-1100,900-1100 --auto-fan --gpu-map 1:0,0:1 --gpu-powertune 18,18 --expiry 150 --per-device-stats --sharelog %userprofile%\desktop\minedshares.log 2>>%temp%\CGminer.log

And keep in mind the alternate pools are stored in my Config file. I'm open to suggestions for further tweaks as well...

[nathanrees19]

Pool2: eligius.st:80

Should be

Code:

mining.

in front of that hostname, but the pool is broken anyway. Try another pool.

[sgrunger]

I have a question and potentially a future feature to implement (maybe !)

I have a scrypt question. I have a rig with very bastardly cards, 6970, 7970, 5970 and 5870. I can get everything working fine -but- ! 7970 works best in scrypt with -g 2, but when I turn it on it makes my ATI driver crash, I reboot, etc, sometimes PC after reboot goes nuts, safe mode, reboot again. If I run -g 1 all works well but 7970 isn't mining as optimal as it should (175KH less than optimal super hashing speed).

I tried the -d option, but cgminer still compiles a scrypt*.bin for the scrypt setting and doesn't run properly. If let's say I run --gpu-fan at like 55, it sets 55 on ALL cards, even if I only tell it to mine from -d 1. I was thinking of running one instance of cgminer on 7970 card with -g 2, then another instance with -g 1, but I don't seem to be able.

Is there anything else I can do ?

[Tia]

Pool2: eligius.st:80

Should be

Code:

mining.

in front of that hostname, but the pool is broken anyway. Try another pool.

Hey, sharp eyes! That was the issue with the port 80 pool. It needed to be: mining.eligius.st:80
Thanks for your response nathanrees19!

[SubNoize]

Is there any reason i can't use a thread concurrency of 1600? My gpu posts 720kh/s with it but all as errors. as soon as i use 8192 it only posts 550ish max. Perhaps this is why ghz card perform so badly?

7970

[Tia]

I have a Bitcoin machine running W8 Pro, Radeon 7850 + 7970 and CGminer 2.11.4
I am looking for finer tuning suggestions, and wanted your thoughts as to what you think I may have overlooked or should consider trying. Of the configs I've tried the following seems to run the best, but I'm sure there are more refinements I've not tried, including worksize.

Here is my Config:
"intensity" : "15,15",
"vectors" : "1,1",
"worksize" : "256,64",
"kernel" : "diablo,diablo",
"lookup-gap" : "0,0",
"thread-concurrency" : "0,0",
"shaders" : "0,0",
"gpu-engine" : "800-1100,900-1100",
"gpu-fan" : "0-85,0-85",
"gpu-memclock" : "1180,1190",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "18,18",
"gpu-vddc" : "0.000,0.000",
"temp-cutoff" : "95,95",
"temp-overheat" : "85,85",
"temp-target" : "75,75",
"api-port" : "4028",
"auto-fan" : true,
"expiry" : "155",
"failover-only" : true,
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "2",
"hotplug" : "5",
"log" : "5",
"no-pool-disable" : true,
"per-device-stats" : true,
"queue" : "1",
"scan-time" : "60",
"temp-hysteresis" : "3",
"verbose" : true,
"shares" : "0",
"kernel-path" : "/usr/local/bin"

[techwtf]

cgminer 3.1.0, I have found blocks with best share > diff & the pool has confirmed my share. but the counter of found blocks stayed 0. it may introduced in around 2.11.4 with your new block checking logic? (I'm running inside openwrt)

[Krak]

Is there any downside to running cgminer as root? I'm getting ready for when my Jalapeno arrives and I can't find the 01-cgminer.rules file anywhere. Ah, I just found it. Still, running as root is much easier.