Nicksasa
|
|
February 26, 2013, 09:20:59 PM |
|
Hello Burnside,
Love your pool to death! Trying to do my best on this end to deliver optimize my rigs.
I have a question about the best way to run cgminer 2.10.5. Specifically I am trying to determine if I should run failover-only or the default failover setting.
My setup is
4 x 5850's running about 1.4 Mhash 1 x 5850 running about 380 Khash 1 x 5870 running about 410 Khash
All of the above equipment is running Windows 7 - 64bit, 4GB ram, and cooled to <72. My hardware is totally stable and runs non-stop.
The problem I am trying to solve: High stales and a lack fault tolerance (your pool is highly available, no slight intended. I just want to be sure my gear keeps running!)
I tried running cgminer with the -load-balance setting but it sent about 50% of the work to the other pool I use. So I stopped doing that.
Then I set cgminer to run with -failover-only. In this mode 99%+ of my hashing went to your pool.
The Litecoin yield was great but I also had very high stales. I hit refresh a few times and saw the stales rate change, so I think the high stales might have been related to one of your background processes starting up and bogging down the server for a few seconds.
In an effort to reduce the stales I changed cgminer to run the default failover mode. In the default failover mode, about 20% of my hashing bleeds over to the failover pool. I have a hard time accepting that so much of my hashing is best sent to the backup pool, but then again the settings in cgminer have proved very reliable in my other applications. If this was best, the yield from the backup pool should be 25% of the daily burnsides pool yield. The yield is only about 15% of the Burnside yield, so I think that tells me something.
What should I do to optimize my setup?
I wondered maybe I should download your Stratum proxy and run that locally for my group of three machines. I thought that might be a small improvement.
Should i just run cgminer as -failover-only and stop worrying about the stales? Or should I run in the default failover mode and let CGminer bleed over some of my work to pool 2 because there really are times the Burnside pool cannot supply enough work.
Thank you in advance for taking a few minutes to read over my questions and thank you for the well implemented pool!
Planman
I hate to say it, but I have zero experience with cgminer. I need to play with it soon to get some stratum support going though! I do wonder... if the stales are mostly from the block boundaries, would failing over to another pool be helpful? Or would they also be seeing dips around the same block boundaries? Every pool has the "stale period" when there's a new block on the network, so no it wouldn't help. It's not too bad in stratum, but looking back at it .. it's terrible on getwork (it's where most of the rounds stales come from)
|
|
|
|
kosmokramer
Sr. Member
Offline
Activity: 266
Merit: 250
The Assman: CEO of Vandelay Import/Export, Inc.
|
|
February 26, 2013, 10:55:57 PM |
|
Hello burnside, first off as a fellow Oregonian, its good to see a local pool!, been lovin it so far Ack, "burnside". Why didn't I catch that before this!?! PDX representing!
|
|
|
|
burnside (OP)
Legendary
Offline
Activity: 1106
Merit: 1006
Lead Blockchain Developer
|
|
February 26, 2013, 11:12:25 PM |
|
Hello burnside, first off as a fellow Oregonian, its good to see a local pool!, been lovin it so far Ack, "burnside". Why didn't I catch that before this!?! PDX representing! Haha, awesome! Dare I say it? .... Go Beavs!
|
|
|
|
planman
Newbie
Offline
Activity: 56
Merit: 0
|
|
February 27, 2013, 06:26:14 AM |
|
Every pool has the "stale period" when there's a new block on the network, so no it wouldn't help. It's not too bad in stratum, but looking back at it .. it's terrible on getwork (it's where most of the rounds stales come from)
Ok Nickasa! Thanks for pointing that out. It makes sense when you explain it that way. I just edited my batch files and am back to --failover-only mode. This should climb me from 1869 khash to around 2300 khash since work won't be bleeding to my backup pool anymore. If nothing else, this should bump up my share of the pool. Next question: Are we running stratum here in Burnside's pool? If no, who do we tip?
|
|
|
|
burnside (OP)
Legendary
Offline
Activity: 1106
Merit: 1006
Lead Blockchain Developer
|
|
February 27, 2013, 11:13:44 PM |
|
Every pool has the "stale period" when there's a new block on the network, so no it wouldn't help. It's not too bad in stratum, but looking back at it .. it's terrible on getwork (it's where most of the rounds stales come from)
Ok Nickasa! Thanks for pointing that out. It makes sense when you explain it that way. I just edited my batch files and am back to --failover-only mode. This should climb me from 1869 khash to around 2300 khash since work won't be bleeding to my backup pool anymore. If nothing else, this should bump up my share of the pool. Next question: Are we running stratum here in Burnside's pool? If no, who do we tip? I'll get it going soon as I can. Then if you want to tip, you can plug in a higher donation percentage.
|
|
|
|
planman
Newbie
Offline
Activity: 56
Merit: 0
|
|
March 01, 2013, 04:49:08 AM |
|
Every pool has the "stale period" when there's a new block on the network, so no it wouldn't help. It's not too bad in stratum, but looking back at it .. it's terrible on getwork (it's where most of the rounds stales come from)
Ok Nickasa! Thanks for pointing that out. It makes sense when you explain it that way. I just edited my batch files and am back to --failover-only mode. This should climb me from 1869 khash to around 2300 khash since work won't be bleeding to my backup pool anymore. If nothing else, this should bump up my share of the pool. Next question: Are we running stratum here in Burnside's pool? If no, who do we tip? I'll get it going soon as I can. Then if you want to tip, you can plug in a higher donation percentage. ? Higher donation percentage? Uh, oh... I was planning cocaine and hookers... but if all it takes to make you happy is a few more litecoins...
|
|
|
|
Zedster
|
|
March 01, 2013, 02:18:27 PM Last edit: March 01, 2013, 08:49:36 PM by Zedster |
|
Yeah I too am getting crushed by stales after found blocks. Is there anything I can do from my side?
EDIT: Switching from cgminer to Reaper seemed to help immensely
|
|
|
|
planman
Newbie
Offline
Activity: 56
Merit: 0
|
|
March 01, 2013, 11:50:46 PM |
|
Yeah I too am getting crushed by stales after found blocks. Is there anything I can do from my side?
EDIT: Switching from cgminer to Reaper seemed to help immensely
Hey Zedster, Thank you for the tip... what version of Reaper are you running? Planman
|
|
|
|
planman
Newbie
Offline
Activity: 56
Merit: 0
|
|
March 02, 2013, 01:25:36 AM |
|
Yeah I too am getting crushed by stales after found blocks. Is there anything I can do from my side?
EDIT: Switching from cgminer to Reaper seemed to help immensely
Has anyone else experienced less stales running Reaper instead of CGminer? I have always been able to match Reaper hash rates running CGminer and I have fully debugged my setup using CGminer. Reaper looks like a whole new learning curve and it doesn't look like reaper handles overclocking internally, so I will have to get software and figure some way to script the overclocking... Anybody have a quick and dirty GUI front end for reaper they suggest? Planman
|
|
|
|
Zedster
|
|
March 02, 2013, 07:46:55 AM |
|
Yeah I too am getting crushed by stales after found blocks. Is there anything I can do from my side?
EDIT: Switching from cgminer to Reaper seemed to help immensely
Has anyone else experienced less stales running Reaper instead of CGminer? I have always been able to match Reaper hash rates running CGminer and I have fully debugged my setup using CGminer. Reaper looks like a whole new learning curve and it doesn't look like reaper handles overclocking internally, so I will have to get software and figure some way to script the overclocking... Anybody have a quick and dirty GUI front end for reaper they suggest? Planman Reaper V13 beta4 64bit the newest one I am pretty sure. I am very new to this all so I am little help but these may help https://bitcointalk.org/index.php?topic=93357.0http://wiki.solidcoin.info/wiki/Reaper
|
|
|
|
planman
Newbie
Offline
Activity: 56
Merit: 0
|
|
March 02, 2013, 02:10:52 PM |
|
Thank you Zedster!
|
|
|
|
vindimy
|
|
March 03, 2013, 08:47:59 AM |
|
Hey burnside, First off, thanks for great work I've been on-and-off in BTC mining, and now that it's getting pointless to mine BTCs with GPU, I've decided to give LTCs a try. So far I'm having lots of fun with my 7950 and reaper I'm here with a perhaps unusual offer. I'm a Linux Sysadmin / DevOps with about 8 years experience and a B.S. in CS from a good university, in my mid-20's, based in the U.S. I'd like to get a lot more into cryptocurrency, learn what are the challenges of operating a real pool, contribute a few fixes/improvements to infrastructure, etc. If you need a person like me to help with whatever it is you're working on, let's chat off-thread sometime. Rather than stumbling in the dark learning it all on my own, I'd be more than happy to help out an existing project. If that sounds good to you, let's chat off-thread. Thanks!
|
|
|
|
serraz
|
|
March 03, 2013, 11:11:32 AM |
|
hey so i derped and locked my address. I was wondering if it was possible to change it i now have a new address but i cant change it in my account. Thanks guys
|
|
|
|
Bogart
Legendary
Offline
Activity: 966
Merit: 1000
|
|
March 03, 2013, 04:28:48 PM Last edit: March 03, 2013, 08:35:15 PM by Bogart |
|
Yeah I too am getting crushed by stales after found blocks. Is there anything I can do from my side?
EDIT: Switching from cgminer to Reaper seemed to help immensely
I was having this same issue, and today I updated to cgminer 2.11.0 (from 2.10.5), and it seems somewhat better. After running for about 20 minutes now (12 new blocks on the network), my stats show A:402 R:29, or 7.21% rejected. That's improved from 2.10.5 where it was showing about A:6000 R:800, or 13.33% rejected. Edit: After a longer period, I'm now back at A:3864 R:608 NB:130. I guess I didn't fix it after all.
|
"All safe deposit boxes in banks or financial institutions have been sealed... and may only be opened in the presence of an agent of the I.R.S." - President F.D. Roosevelt, 1933
|
|
|
Balthazar
Legendary
Offline
Activity: 3108
Merit: 1359
|
|
March 03, 2013, 04:49:25 PM |
|
The stale and unknowns are WAY too high, what's going on? I thought this was being addressed? 19% stale is not on my end, I've got 50MB fibre with a 32ms ping time, and I keep getting pool not providing work fast enough. If this can't be addressed it makes no sense to stay.
[2013-02-18 08:32:44] LONGPOLL from pool 0 detected new block [2013-02-18 08:32:47] Rejected 006bb902 Diff 608/31 GPU 0 (prevhash-stale) [2013-02-18 08:32:47] Rejected 0445a806 Diff 59/31 GPU 1 (prevhash-stale) [2013-02-18 08:33:02] Rejected 01f4d841 Diff 130/31 GPU 0 (unknown-work) [2013-02-18 08:33:03] Rejected 062b5f13 Diff 41/31 GPU 1 (unknown-work) [2013-02-18 08:33:05] Rejected 01e16d7e Diff 136/31 GPU 1 (unknown-work) [2013-02-18 08:33:08] Rejected 037ed51c Diff 73/31 GPU 0 (unknown-work) [2013-02-18 08:33:09] Rejected 048fa180 Diff 56/31 GPU 0 (unknown-work) [2013-02-18 08:33:10] Rejected 025d4668 Diff 108/31 GPU 1 (unknown-work) [2013-02-18 08:33:19] Rejected 03c5e4c7 Diff 67/31 GPU 1 (unknown-work)
Maybe you need to adjust -E option.
|
|
|
|
vindimy
|
|
March 03, 2013, 09:58:26 PM |
|
I'm going to post my reaper 13 beta 4 config for Radeon HD 7950 that's been working well for over 24 hours now, in case anyone else is trying to get this up. This is on stock voltage, CPU, and memory clock: kernel reaper.cl save_binaries yes enable_graceful_shutdown yes long_polling yes platform 1 device 0 cpu_mining_threads 0
mine litecoin
host ltc.kattare.com port 9332 user YOUR_USER_NAME pass YOUR_PASS
protocol litecoin
worksize 256 vectors 1 aggression 20 threads_per_gpu 1 sharethreads 32 lookup_gap 2 gpu_thread_concurrency 24576
reaper reports 578 kH/s with very little variance, stales are at 5-9%. Looking at "my stats" page, I see hourly kH variance, the range of values shown is 500 kH - 700 kH. Note: be sure to give reaper the right "address" of your graphics card: in my case it's "platform 1, device 0". Default config reaper comes from is "platform 0, device 0" which would be Intel CPU for me. Don't let reaper mine on CPU - GPU hash rate drops due to that.
|
|
|
|
Bogart
Legendary
Offline
Activity: 966
Merit: 1000
|
|
March 04, 2013, 03:19:21 AM |
|
Setting -E 15 didn't help at all, but I noticed something.
The pool still reports essentially the full hashrate for the worker. I think a lot of the stales may be duplicate submissions.
|
"All safe deposit boxes in banks or financial institutions have been sealed... and may only be opened in the presence of an agent of the I.R.S." - President F.D. Roosevelt, 1933
|
|
|
Rawted
|
|
March 04, 2013, 06:54:58 PM |
|
i brought in about 1mh today and my stales are terrible. im hovering between 15-25% personal stales, and the pool is sitting at 81% efficiency. I noticed some big boys came into the pool to play today - things a little haywire or can you offer some tips to reduce stales? using cgminer currently with 4 5830s.
|
|
|
|
Zedster
|
|
March 04, 2013, 08:52:43 PM |
|
i brought in about 1mh today and my stales are terrible. im hovering between 15-25% personal stales, and the pool is sitting at 81% efficiency. I noticed some big boys came into the pool to play today - things a little haywire or can you offer some tips to reduce stales? using cgminer currently with 4 5830s.
I run 5830s and have to use reaper it helps immensely as I posted before.
|
|
|
|
Rawted
|
|
March 04, 2013, 11:13:46 PM |
|
mind sharing your config? ive played with reaper a dozen times and just about every switch i could think of, and it always craps out after a few hours.
|
|
|
|
|