Hardware errors with cgminer and ltc mining?

[Gomeler]

Is it normal to have so many reported hardware errors?

Here are my cgminer settings: cgminer --scrypt -o http://LHPool.eu:8080 -u Gomeler.1 -p x -w 256 -I 18 -g 1 --lookup-gap 2 --thread-concurrency 1120

Drivers: 13.1 x64

OS: Ubuntu 12.04 x64

and it running a bit. This is a HD 6870x2 with 1120 shaders. Setting --thread-concurrency to higher than 1120 causes it to throw some bizarre openCL error about memory being greater than available. Top shows ~1.1GB free.

cgminer version 2.10.4 - Started: [2013-02-01 01:22:24]
--------------------------------------------------------------------------------
 (5s):487.8K (avg):456.8Kh/s | Q:10  A:19  R:2  HW:0  E:190%  U:5.0/m
 ST: 0  SS: 0  DW: 1  NB: 4  LW: 24  GF: 0  RF: 0
 Connected to LHPool.eu diff 15 with LP as user Gomeler.1
 Block: 33dae6ae1d602569...  Diff:1.33M  Started: [01:24:40]  Best share: 275
--------------------------------------------------------------------------------
 [P]ool management [G]PU management [ S ]ettings [D]isplay options [Q]uit
 GPU 0: 243.1K/241.9Kh/s | A:10 R:2 HW:34 U:2.64/m I:14
 GPU 1: 243.1K/214.9Kh/s | A: 9 R:0 HW:41 U:2.37/m I:14
--------------------------------------------------------------------------------

 [2013-02-01 01:21:56] Started cgminer 2.10.4
 [2013-02-01 01:21:57] Probing for an alive pool
 [2013-02-01 01:22:24] Long-polling activated for http://LHPool.eu:8080/LP
 [2013-02-01 01:22:44] Rejected 0cb20944 Diff 20/15 GPU 0  (prevhash-stale)
 [2013-02-01 01:22:44] LONGPOLL from pool 0 detected new block
 [2013-02-01 01:22:48] Rejected 072b736f Diff 35/15 GPU 0  (prevhash-stale)
 [2013-02-01 01:22:56] Accepted 0e7e1c78 Diff 17/15 GPU 1
 [2013-02-01 01:22:56] LONGPOLL from pool 0 detected new block
 [2013-02-01 01:23:01] Accepted 095d0243 Diff 27/15 GPU 1
 [2013-02-01 01:23:16] Accepted 03b91ba9 Diff 68/15 GPU 1
 [2013-02-01 01:23:20] Accepted 0588028e Diff 46/15 GPU 1
 [2013-02-01 01:23:21] Accepted 06f464e0 Diff 36/15 GPU 0
 [2013-02-01 01:23:22] Accepted 0d398263 Diff 19/15 GPU 0
 [2013-02-01 01:23:38] Accepted 00ee0f22 Diff 275/15 GPU 0
 [2013-02-01 01:24:06] Accepted 085aae8c Diff 30/15 GPU 0
 [2013-02-01 01:24:27] Accepted 06abc42b Diff 38/15 GPU 1
 [2013-02-01 01:24:40] Accepted 0e21986c Diff 18/15 GPU 0
 [2013-02-01 01:24:40] LONGPOLL from pool 0 detected new block
 [2013-02-01 01:24:56] Accepted 0a849586 Diff 24/15 GPU 1
 [2013-02-01 01:24:56] Accepted 0488b40e Diff 56/15 GPU 0
 [2013-02-01 01:25:03] Accepted 03d595f3 Diff 66/15 GPU 0
 [2013-02-01 01:25:03] Accepted 097c66d4 Diff 26/15 GPU 1
 [2013-02-01 01:25:17] Accepted 0372c6fd Diff 74/15 GPU 1
 [2013-02-01 01:25:18] Accepted 0e1c6ccb Diff 18/15 GPU 0
 [2013-02-01 01:25:25] Accepted 0b376340 Diff 22/15 GPU 0
 [2013-02-01 01:25:38] Accepted 049bbd06 Diff 55/15 GPU 1
 [2013-02-01 01:25:43] Accepted 0f036fea Diff 17/15 GPU 0

[Gomeler]

Bump.

[tacotime]

As stated in my mining guide (see sig), HW errors are mainly caused by thread-concurrencies that are too low (the RAM buffers for LTC start to overwrite one another and you get apparent data corruption)

reaper doesn't check for memory usage errors thrown by the cl kernel, so I would try using reaper instead before doing anything else.

[Gomeler]

Ok, I'll dump the card into another rig with more ram and try increasing thread-concurrency.

[nethead]

i seem to have some problem with cgminer too. I downloaded 2.10.4 to re-try soloing LTC and i get this after opening:

"Connected to 127.0.0.1 diff 1.24M without LP as user nethead"

I get full gpu usage and it seems to mine (no blocks yet from this version) but why is the diff so high??

[-ck]

i seem to have some problem with cgminer too. I downloaded 2.10.4 to re-try soloing LTC and i get this after opening:

"Connected to 127.0.0.1 diff 1.24M without LP as user nethead"

I get full gpu usage and it seems to mine (no blocks yet from this version) but why is the diff so high??

README last FAQ.

[Gomeler]

Gave up on cgminer. Compiled reaper, plugged in the following just to get running..

worksize 256
vectors 1
aggression 20
threads_per_gpu 1
sharethreads 32
lookup_gap 2
gpu_thread_concurrency 8192

and I'm up and running at ~600 KH/s reported from reaper and the pool. cgminer was a mess trying to find a combination of thread-concurrency, lookup-gap and aggression that produced low hardware errors and high utility. Highest I achieved with cgminer was ~440 KH/s but lots of hardware errors. Best utility was 26 shares/minute compared to ~31 shares/minute with reaper. Not sure why I couldn't run a thread_concurrency higher than lookup_gap * 1024 but I'm happy now.

[gnarcoinguy]

I seem to be having a similar problem.  On Windows cgminer works great with the same cards, but under linux I can't seem to get cgminer to work, close to 100% hardware errors with any setting.  Its very strange.  I have dual 6950s on linux and have tried just about everything without any luck.  

I'd like to get cgminer working as its worked so well in the past for me but I'm sort of stumped at this point.  

Where did you get reaper?  The solidcoin download links all seem to direct and I can't find the source.

Edit: found a copy of reaper on github.  Its now mining okay without any errors.  Very strange.  I highly prefer cgminer it seems like a far superior software package (granular, stable, recently updated, etc), but I'm at a loss how to get it to work on linux at the moment.