[ANN][LTC][Pool][PPLNS][STRATUM] - ltc.kattare.com - burnside's Mining Pool - page 24.

turvarya

hero member

Activity: 714

Merit: 500

Quote from: Rawted on March 07, 2013, 06:53:01 PM

turvarya, maybe these two threads will help. both have multiple posts from people mining with cgminer or reaper on 6870s. looks to me your concurrency and a few other settings might be off.

https://bitcointalksearch.org/topic/cgminer-optimization-procedure-for-hd6870-102376

https://bitcointalksearch.org/topic/hd6870-maximum-hashs-4114

ok, thanks, helped me a lot
still it's a pain in the ass to try stuff out, since it often nearly freezes my desktop.
I got about 360kh/h top and avg of about 300kh/h over night, when i am afk, but haven't found a configuration, yet, that gives me good kh/h and let me still work on it. But I guess it#s a matter of more tries. Never hat real problems with btc-mining

setting for my best result, if anyone needs it(the worksize of 64 is strange, but it works)

Code:

cgminer --scrypt ^
-o http://ltc.kattare.com:9332/ ^
-u turvarya.1 -p x ^
--thread-concurrency 6720 ^
-I 16 ^
-g 1 ^
-w 64

Rawted

hero member

Activity: 742

Merit: 500

turvarya, maybe these two threads will help. both have multiple posts from people mining with cgminer or reaper on 6870s. looks to me your concurrency and a few other settings might be off.

https://bitcointalksearch.org/topic/cgminer-optimization-procedure-for-hd6870-102376

https://bitcointalksearch.org/topic/hd6870-maximum-hashs-4114

Rawted

hero member

Activity: 742

Merit: 500

Quote from: reb73 on March 07, 2013, 06:24:17 PM

Quote from: c0ldfusi0nz on March 07, 2013, 04:42:35 PM

Well, that is certainly nice to know! Sure would have saved me some time if it had been outlined in the manual or simply printed out as "GPU Error Rate"

After significantly lowering my config values, now I seem to actually be processing work!

Had a similar issue myself when getting started with gpu mining on my NVidia GT555M, had to bring down the GPU thread concurrency down from 6144 down to 2048 initially to get it hashing properly and then steadily got up to 4096 where its at now, though the stale share count is pretty high on gpu mining (between 40-50%)..

This is apparently linked to the number of shaders and the thread queue depth for each shader - the shader count is woefully low on NVidia GPUs - mine is 192. Wondering if the magic number is 192 shaders multiplied by sharethreads 18 = 3456? Will have a stab at this to see if it helps..

i have a pair of sli'd 560ti's in an old gaming rig that i mine on for one pool. i know sli slows it down a bit, but with each having 384 shaders, i sit around 150mh/s combined.

reb73

newbie

Activity: 37

Merit: 0

Quote from: c0ldfusi0nz on March 07, 2013, 04:42:35 PM

Well, that is certainly nice to know! Sure would have saved me some time if it had been outlined in the manual or simply printed out as "GPU Error Rate"

After significantly lowering my config values, now I seem to actually be processing work!

Had a similar issue myself when getting started with gpu mining on my NVidia GT555M, had to bring down the GPU thread concurrency down from 6144 down to 2048 initially to get it hashing properly and then steadily got up to 4096 where its at now, though the stale share count is pretty high on gpu mining (between 40-50%)..

This is apparently linked to the number of shaders and the thread queue depth for each shader - the shader count is woefully low on NVidia GPUs - mine is 192. Wondering if the magic number is 192 shaders multiplied by sharethreads 18 = 3456? Will have a stab at this to see if it helps..

turvarya

hero member

Activity: 714

Merit: 500

I tried Reaper and cgminer. cgminer nearly freezes my desktop and gives me 31 kH/s(that's the value burnside shows me), with this command

Code:

cgminer --scrypt -o http://ltc.kattare.com:9332/ -u turvarya.1 -p x --thread-concurrency 4800 -I 18 -g 1 -w 128

when I use Reaper i just get about 20 kH/s, but can do other stuff, with this configuration

litecoin.conf

Code:

protocol litecoin

worksize 256
aggression 10
threads_per_gpu 2
sharethreads 18
lookup_gap 2
gpu_thread_concurrency 4800
long_polling yes

reaper.conf

Code:

kernel reaper.cl
save_binaries yes
enable_graceful_shutdown no
long_polling yes
platform 0
device 0

mine litecoin

Is that even near of what I should get? Any suggestions?
(I have a 6870)

Rawted

hero member

Activity: 742

Merit: 500

Ya, that's just errors. If a single nvidia could pull 1gh/s, we'd all be using them! Wink

Glad you got it fixed.

c0ldfusi0nz

newbie

Activity: 52

Merit: 0

Well, that is certainly nice to know! Sure would have saved me some time if it had been outlined in the manual or simply printed out as "GPU Error Rate"

After significantly lowering my config values, now I seem to actually be processing work!

msm595

full member

Activity: 185

Merit: 100

Quote from: c0ldfusi0nz on March 07, 2013, 04:03:23 PM

Hm, Reaper's output shows: GPU 956.7MH/s, shares: 0|0, stale 0%, ~0 kH/s, 906s

Once again, this is not your GPU hashrate, this is the rate at which your GPU is erroring. drop down thread concurrency, work size, aggression, etc.

c0ldfusi0nz

newbie

Activity: 52

Merit: 0

Hm, Reaper's output shows: GPU 956.7MH/s, shares: 0|0, stale 0%, ~0 kH/s, 906s

I bumped up the aggression and threads per GPU which really helped the initial number shoot up, but the actual process rate is still showing 0 kH/s.

FYI, I'm running Ubuntu 12.10 on a NVS 4200M that tells me it has 48 CUDA cores, with the official Nvidia drivers from http://www.nvidia.com/object/unix.html

I'm not having much luck finding Nvidia-related Reaper discussions :-/

Rawted

hero member

Activity: 742

Merit: 500

Quote from: c0ldfusi0nz on March 07, 2013, 03:19:03 PM

I'm leaning toward it being a problem with my Reaper setup. If I run my cpu miner on this machine, it pulls work fine and the worker shows up as active. Running Reaper instead pulls no work and the worker shows as inactive. The only thing I can think of that might be wrong is that I'm running this on an NVIDIA chipset rather than ATI; I thought Reaper is supposed to support both, just that ATI is a noticeably better performer.

litecoin.conf

Quote

host ltc.kattare.com
port 9332
user c0ldfusi0nz.gpuworker
pass therightpass

protocol litecoin

worksize 256
vectors 1
aggression 8
threads_per_gpu 1
sharethreads 18
lookup_gap 2
gpu_thread_concurrency 6144

reaper.conf:

Quote

kernel reaper.cl
save_binaries yes
enable_graceful_shutdown no
long_polling yes
platform 0
device 0

mine litecoin

Whats your hash rate? nvidia typically is very slow, and the site wont update your worker stats until it receives a share. Also, some of those settings are for amd gpus only.

c0ldfusi0nz

newbie

Activity: 52

Merit: 0

I'm leaning toward it being a problem with my Reaper setup. If I run my cpu miner on this machine, it pulls work fine and the worker shows up as active. Running Reaper instead pulls no work and the worker shows as inactive. The only thing I can think of that might be wrong is that I'm running this on an NVIDIA chipset rather than ATI; I thought Reaper is supposed to support both, just that ATI is a noticeably better performer.

litecoin.conf

Quote

host ltc.kattare.com
port 9332
user c0ldfusi0nz.gpuworker
pass therightpass

protocol litecoin

worksize 256
vectors 1
aggression 8
threads_per_gpu 1
sharethreads 18
lookup_gap 2
gpu_thread_concurrency 6144

reaper.conf:

Quote

kernel reaper.cl
save_binaries yes
enable_graceful_shutdown no
long_polling yes
platform 0
device 0

mine litecoin

burnside

legendary

Activity: 1106

Merit: 1006

Lead Blockchain Developer

Quote from: LittleDigger on March 07, 2013, 01:23:49 PM

Quote from: reb73 on March 07, 2013, 01:01:50 PM

+1

Close to 1.5 million shares submitted over 6 hours without yielding a block, something looks amiss..

7 Hours 17 minutes now.. and I'm paying 23c per Kwh..
Huh

The simplecoin cronjob that runs once a minute stopped running last night. (~8 hrs ago) Naturally, it has to break while I'm sleeping.

The way it works, it takes out a lock, then only unlocks when it finishes. So something happened that it didn't finish and release the lock. From there on out, no more can run until I clear the lock manually. I've looked at the logs and see no reason that it should have failed. There's no db errors or script errors or anything. On a hunch I've added a set_time_limit(0); call at the top. I really hate it when things break with no idea why. Angry

You can see on the blocks and payouts though that when I released the lock it caught up with the solved blocks and payouts.

Cheers.

Rawted

hero member

Activity: 742

Merit: 500

coldfusionz, burnside's pool was having some issues this morning. It literally just came back up a few minutes ago. Can you check again and let us know?

c0ldfusi0nz

newbie

Activity: 52

Merit: 0

Should I expect my miner to be pulling shares right now? I'm trying to get Reaper up and running and after successfully compiling and configuring it I'm not getting any errors, but neither am I pulling any work to process... also, my worker is not showing up as "active" on the pool site, yet I don't receive any authentication or connection errors from Reaper.

Quote

Compiling kernel... this could take up to 2 minutes.
Binary size: 951564 bytes
Program built from source.
LTC buffer size: 384MB.
Creating 1 GPU threads
1...done
Longpoll url -->/LP<--
ffffffffffffffffffffffffffffffffffffffffffffffffffffffffff070000
Activating long polling.
Long polling URL: [/LP]. trying to parse.
LP Path: /LP
Getwork took 5.893 s! 0|0, stale 0%, ~0 kH/s, 46s
Got LP7MH/s, shares: 0|0, stale 0%, ~0 kH/s, 153s
Getwork took 9.208 s! 0|0, stale 0%, ~0 kH/s, 191s
Got LP7MH/s, shares: 0|0, stale 0%, ~0 kH/s, 254s
Getwork took 9.236 s! 0, stale 0%, ~0 kH/s, 269s
Got LP695MH/s, shares: 0|0, stale 0%, ~0 kH/s, 309s
Getwork took 9.411 s!
Got LP691MH/s, shares: 0|0, stale 0%, ~0 kH/s, 357s
Getwork took 9.244 s! 0|0, stale 0%, ~0 kH/s, 431s
Got LP68MH/s, shares: 0|0, stale 0%, ~0 kH/s, 502s
GPU 2.671MH/s, shares: 0|0, stale 0%, ~0 kH/s, 652s

LittleDigger

sr. member

Activity: 462

Merit: 251

Quote from: reb73 on March 07, 2013, 01:01:50 PM

+1

Close to 1.5 million shares submitted over 6 hours without yielding a block, something looks amiss..

7 Hours 17 minutes now.. and I'm paying 23c per Kwh..
Huh

reb73

newbie

Activity: 37

Merit: 0

+1

Close to 1.5 million shares submitted over 6 hours without yielding a block, something looks amiss..

Rawted

hero member

Activity: 742

Merit: 500

Wasn't trying to be a dick, was short on time, apologies if i came off that way.

Also, pool is having some serious issues, or the site is at least. no blocks in 6 hours, last 10 blocks sitting at same confirms, no payouts or balance increases.

thepwnorbpwnd

sr. member

Activity: 462

Merit: 250

Quote from: Rawted on March 07, 2013, 09:08:33 AM

Just about everything in those is wrong. Please read over and redo your confs according to https://bitcointalksearch.org/topic/consolidated-litecoin-mining-guide-for-5xxx-6xxx-and-7xxx-gpus-117221

You didn't have to be a dck about it. He's got a little more than half of the settings in the right direction just not at the perfect spot.

Rawted

hero member

Activity: 742

Merit: 500

Just about everything in those is wrong. Please read over and redo your confs according to https://bitcointalksearch.org/topic/consolidated-litecoin-mining-guide-for-5xxx-6xxx-and-7xxx-gpus-117221

turvarya

hero member

Activity: 714

Merit: 500

ok, thanks
I have no idea what u/p means ^^

reaper.conf looks like this

Code:

kernel reaper.cl

enable_graceful_shutdown yes
long_polling yes
platform 0
device 0
threads_per_gpu 2
cpu_mining_threads 0
aggression 13

mine litecoin

litecoin.conf

Code:

protocol litecoin

worksize 128
aggression 10
threads_per_gpu 2
sharethreads 4
lookup_gap 2
gpu_thread_concurrency 6144
long_polling yes

isn't there a site with good suggested configs?

also, can somebody explain, when the litecoin.conf is used and when the reaper.conf and which one is used, when I e.g. but aggression in both

Topic: [ANN][LTC][Pool][PPLNS][STRATUM] - ltc.kattare.com - burnside's Mining Pool - page 24. (Read 118848 times)