Yeh mine dropped aswell
ok I will make that memory access scheme configurable...
Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 971. (Read 3426921 times)
January 30, 2014, 04:14:10 PM
ok I will make that memory access scheme configurable...
January 30, 2014, 03:50:02 PM
Yeh mine dropped aswell
January 30, 2014, 03:48:44 PM
I just compiled the latest github, and can tell you that the performance dropped using the same parameters. From 4.11khash to 3.93khash...
Trying some other -L now...
Trying some other -L now...
January 30, 2014, 03:43:39 PM
Holy hell! I wish I had that hashing power
Let us know how it goes
Let us know how it goes
January 30, 2014, 03:41:46 PM
autotune also surprised me once today with way too low hash rates.
I am going to throw 6 MHash at a MRC pool today. Let's see what that amounts to...
I am going to throw 6 MHash at a MRC pool today. Let's see what that amounts to...
January 30, 2014, 03:21:39 PM
Can I call shenanigans on that 2014-01-29 660 TI added to the spreadsheet? I find it very unlikely that a stock 2GB card running the 12-18-2013 build is going to get 409Kh/s (79Kh/s faster then the top 660 TI)
Possibly.
The hashrate spreadsheets are pretty outdated and sometimes weird results are being uploaded and to be honest I completely lost interest in scrypt. I'll keep maintaining them like once a week but don't expect any upgrades from me on my own. However, if you suggest something useful I'll gladly do it.
The reason I abandoned scrypt completely is that I have a low-mid range card, which is a GTX 660 and these cards have much better scrypt-jane performance compared to scrypt performance.
So for example a 780 Ti hits like, I don't know, let's say 700+ khashes/s doing scrypt and 5 khashes doing scrypt-jane while a 660 gets 260 in scrypt and 3.4 in jane so a 660 is 2,7 times slower in scrypt but only 1,4 times slower in scrypt-jane which means mid cards are better off with scrypt-jane and high end cards should probably go with scrypt.
On another note, I'm sticking with the old autotune method because the new one seems to do weird things. Could be because of the 1% backoff value instead of 12, not sure.
Edit: Oh, the microcoin difficulty stopped dropping and started climbing back up (3.59 -> 2.2). Found 10 blocks these two days with roughly 360-400 kH/s.
January 30, 2014, 03:07:20 PM
Question of the day: What to mine with my nVidia cards?
Yacoin -> this coin is falling (failing?)
MRC -> no luck with solo mining at all. Trying a pool now (mrc.easy-mining.net pool seems to be fixed at last).
VertCoin -> who is currently into this coin? if you are, why?
middlecoin -> the DOGE rush is over, right?
I think what would really be helpful is a profitability calculator for various scrypt-jane coins, taking one's hash rate at different N-factors into account.
Christian
Yacoin -> this coin is falling (failing?)
MRC -> no luck with solo mining at all. Trying a pool now (mrc.easy-mining.net pool seems to be fixed at last).
VertCoin -> who is currently into this coin? if you are, why?
middlecoin -> the DOGE rush is over, right?
I think what would really be helpful is a profitability calculator for various scrypt-jane coins, taking one's hash rate at different N-factors into account.
Christian
Middlecoin.com at ~600 khash is making me about 0.0125 BTC every 2 days currently. So far today, Jan 30 2104, I'm set to do 0.103 in a single day due to some of the BTC news coming out of China at the moment. Worth at least keeping in mind, since good days are great but "meh" days are still a consistent source of BTC at decent values even after the Doge craze.
YACoin is definitively on it's way out of relevance right now. Is it dying? Not really, just not the latest and greatest.
January 30, 2014, 12:24:13 PM
Hi to all.
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?The easiest way would be to create a scheduled task in Windows which starts every Day at 0:00 and runs for 23,59 hours.
There is a setting where you can terminate a scheduled task after a given amount of time. So cudaminer wont be executed twice.
January 30, 2014, 11:42:24 AM
Hi to all.
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?
Here you have a little manual, not sure if it works with cudaminer, but it's worth a try:Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?http://social.technet.microsoft.com/Forums/windowsserver/en-US/6bd27cac-e402-4bf2-9c7c-c329f42262fb/run-a-batch-job-as-windows-service?forum=winservergen
January 30, 2014, 11:41:07 AM
Hi to all.
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?
You could try running it as a service (don't ask me how to do that though...)Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?Try going into the service manager, create new service, fill in where the app is and don't forget the command line switches, enable interact with desktop, select local service for account. That is where I would start. For more information check Technet.Microsoft.com
January 30, 2014, 11:13:11 AM
Hi to all.
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?
You could try running it as a service (don't ask me how to do that though...) Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations? January 30, 2014, 10:48:19 AM
Hi to all.
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations?
Please, help me to solve one problem with cudaminer. I'am trying to run it in the environment, where no user is logged on windows - machine is in Active Directory, so no auto-login (I know, that technically this is possible
but administratively this is not allowed). Is there any way to run cudaminer without user logon? Or there some nvidia driver limitations? January 30, 2014, 08:52:19 AM
In what do we get paid out if we use that chunky pool (the 6666 port one)?
It's sadly not mentioned in the FAQ there.
It's sadly not mentioned in the FAQ there.
It is currently in the beta mode, sorry that I forgot to mention that.
You can setup automatic payout every coin which supported by the pool to your wallet, and then point your miner to port 6666. It will automatically mines the most profitable coin. This kind of service is not new, I know. The reason I like about this pool it is because it supports new coin and the new coin are in most case very profitable before the multipool & co knows about it
Only the pool owner knows. Every pool owner has his self-made profit calculation method. But in general I assume it is all the same like coins mined per hour * current exchange rate. Of course many other factors are included. I even have my self-made profit calculator.
January 30, 2014, 08:49:28 AM
@AizenSou
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
cudaminer cannot even switch to a different N-factor coin, as one currently gives the N-factor parameters (or coin name) at program launch. Furthermore, switching to a different N-factor would require a complete re-initialization of all memory buffers and require a different set of kernel launch parameters.
hmm, so I do not think multipools for scrypt-jane are currently feasible with cudaminer.
One would have to define extensions to the stratum protocol to signal different startT, Nfmin/Nfmax parameters (or alternatively the current nfactor) and redesign parts of cudaminer so it can switch the Nfactor at run time.
Christian
I didn't say that this pool is scrypt-jane multipool. It is pure scrypt multipool like many others. But they support new potential coin IMO. And I only report my experience.
@Christian: yeah I need to manually kill cudaminer in all of my machines and restart it again. At first it is pure pain in *ss but I have written a bash script for this task, so it's ok now. And I always wait 5min before start cudaminer again, for safety.
January 30, 2014, 08:42:23 AM
@AizenSou
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
cudaminer cannot even switch to a different N-factor coin, as one currently gives the N-factor parameters (or coin name) at program launch. Furthermore, switching to a different N-factor would require a complete re-initialization of all memory buffers and require a different set of kernel launch parameters.
hmm, so I do not think multipools for scrypt-jane are currently feasible with cudaminer.
One would have to define extensions to the stratum protocol to signal different startT, Nfmin/Nfmax parameters (or alternatively the current nfactor) and redesign parts of cudaminer so it can switch the Nfactor at run time.
Christian
January 30, 2014, 08:36:09 AM
@AizenSou
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
The pool wont be able to switch between scrypt-jane and scrypt so you have to pick one or the other to do. As their different algorithms
January 30, 2014, 08:31:02 AM
In what do we get paid out if we use that chunky pool (the 6666 port one)?
It's sadly not mentioned in the FAQ there.
It's sadly not mentioned in the FAQ there.
It is currently in the beta mode, sorry that I forgot to mention that.
You can setup automatic payout every coin which supported by the pool to your wallet, and then point your miner to port 6666. It will automatically mines the most profitable coin. This kind of service is not new, I know. The reason I like about this pool it is because it supports new coin and the new coin are in most case very profitable before the multipool & co knows about it
January 30, 2014, 08:13:48 AM
In what do we get paid out if we use that chunky pool (the 6666 port one)?
It's sadly not mentioned in the FAQ there.
It's sadly not mentioned in the FAQ there.
It is currently in the beta mode, sorry that I forgot to mention that.
You can setup automatic payout every coin which supported by the pool to your wallet, and then point your miner to port 6666. It will automatically mines the most profitable coin. This kind of service is not new, I know. The reason I like about this pool it is because it supports new coin and the new coin are in most case very profitable before the multipool & co knows about it
January 30, 2014, 07:51:52 AM
In what do we get paid out if we use that chunky pool (the 6666 port one)?
It's sadly not mentioned in the FAQ there.
It's sadly not mentioned in the FAQ there.
January 30, 2014, 07:36:51 AM
I will download and give it a test :-)
Edit: ok so it took 7 attempts and 3 downloads to get the latest build to compile. Bare in mind I have compiled every other release in 1 attempt. It was giving different errors constantly and even sometimes doing nothing
Finally got it to compile and found a drop in performance.
4.77khash/s for the latest
5.04khash/s on the one I built around 12 hours ago
Thats on yacoin with the same configuration as I posted earlier. It was not left to autotune I just ran the settings I did earlier
Edit: ok so it took 7 attempts and 3 downloads to get the latest build to compile. Bare in mind I have compiled every other release in 1 attempt. It was giving different errors constantly and even sometimes doing nothing
Finally got it to compile and found a drop in performance.
4.77khash/s for the latest
5.04khash/s on the one I built around 12 hours ago
Thats on yacoin with the same configuration as I posted earlier. It was not left to autotune I just ran the settings I did earlier
For scrypt-jane I am seeing mixed results. GT 640 (Compute 3.5) seems a little slower, GTX 780 TIs on Linux slightly faster.
Maybe I can later offer an option to switch between these two (totally different) memory access schemes to the scratchpad.
I am also wondering if you guys still see non-validating shares in conjunction with the lookup gap after this change?
I can't see much activity when solo-mining, but the pool miners will notice.
Christian
Hi Christian,
I still don't have time to research the profitability of scrypt-jane coins, so I still stick my cudaminer to scrypt coin. Earning steadily over 0.02 BTC/1Mhs daily now (as most as double as middlecoin & multipool co&). If you don't have time to research the new coins, just point your miner to pool.chunky.ms port 6666. I'm currently supporting and working with the pool owner to improve the profit calculator for the new coins.
Check pool.chunky.ms for more information
Jump to: