Pages:
Author

Topic: [POOL][Scrypt][Scrypt-N][X11] Profit switching pool - wafflepool.com - page 97. (Read 465716 times)

sr. member
Activity: 490
Merit: 250
Where can i see full list of Wafflepool API ??
hero member
Activity: 693
Merit: 500

It should pass N-factor to opencl kernel as well as other parameters, thus when you want to change the N-factor it do it simply with change of a function call parameter.

It does - to the opencl kernel, but the parameter within the main program does not know when to change because they just take nfactor as a command  line parameter instead of starttime, min n, and max n...

Right, I'm going to assume nfactor is compiled as part of the kernel (haven't done enough research, why are these being called kernels, are they not shader-based?), so changing from stratum would require a kernel recompile for the first time an nfactor is seen, and after that would require a kernel swap (not sure how long these take - but makes profit-switching significantly more difficult)

NFactor is a parameter passed to the opencl "program".  So, if the main program knows when it changes, it can pass the correct value.  The project I've been working on for a while is to get it so YACMiner can swap between coins of different N factors and mine them successfully without intervention by the user - in essence, the basis for allowing a multipool that can mine scrypt/scryp-chacha and N-scrypt much like middlecoin/wafflepool/etc are just for scrypt.

I've completed the auto-tuning portion now, and will be digging through what SGMiner did for the kernel hotswap - the way I've been doing it has some issues.  

The remaining issue is exactly what phzi mentioned - it requires an extension to the stratum protocol to be able to control what the mining software is mining.  Very do-able... just a matter of finding time to work on it.  It is somewhat silly to have it on the miner-side, not just for Scrypt-chacha or NScrypt, but for ANY mining algorithm.  If a miner connects to a server with software incapable to hashing that algorithm, it should disconnect them.  This solves the problem with people specifying the wrong algorithm in their miner config as well
hero member
Activity: 700
Merit: 500
Right, I'm going to assume nfactor is compiled as part of the kernel (haven't done enough research, why are these being called kernels, are they not shader-based?), so changing from stratum would require a kernel recompile for the first time an nfactor is seen, and after that would require a kernel swap (not sure how long these take - but makes profit-switching significantly more difficult)
Bang on summary.  And kernels swaps are bloody fast in my tests (almost no downtime observed, definitely less then half a second on my rigs).

My read-thru of the sgminer code last night seemed to suggest that pool provided nfactor would be reasonably trivial.  Most of the functionality is already there, it would now involve taking extra params from stratum notify messages for algo/algoparams, and attaching this to the work struct and handling it in the new get_work_prepare_thread func.  Or, a cleaner approach might be a new stratum method that updates pool algo/nfactor (relevant functions are also already in place) and clears the work queue.

It does - to the opencl kernel, but the parameter within the main program does not know when to change because they just take nfactor as a command  line parameter instead of starttime, min n, and max n...
I think a cleaner approach in the long run would be to have the mining software unaware of when it should change N, and have this information provided by the pool only when it occurs.  Making it miner-side just seems silly.
sr. member
Activity: 322
Merit: 254

It should pass N-factor to opencl kernel as well as other parameters, thus when you want to change the N-factor it do it simply with change of a function call parameter.

It does - to the opencl kernel, but the parameter within the main program does not know when to change because they just take nfactor as a command  line parameter instead of starttime, min n, and max n...

Right, I'm going to assume nfactor is compiled as part of the kernel (haven't done enough research, why are these being called kernels, are they not shader-based?), so changing from stratum would require a kernel recompile for the first time an nfactor is seen, and after that would require a kernel swap (not sure how long these take - but makes profit-switching significantly more difficult)
hero member
Activity: 693
Merit: 500

It should pass N-factor to opencl kernel as well as other parameters, thus when you want to change the N-factor it do it simply with change of a function call parameter.

It does - to the opencl kernel, but the parameter within the main program does not know when to change because they just take nfactor as a command  line parameter instead of starttime, min n, and max n...
sr. member
Activity: 322
Merit: 254
EU server is dead?

Looks to be up and running to me?  SEA was down for about an hour just now, but EU has been fine
sr. member
Activity: 490
Merit: 250
sr. member
Activity: 490
Merit: 250
Somewhat awesome dev news, IMO, albiet slightly unrelated to WafflePool.

The latest sgminer git now supports pool-specific algorithm/nfactor and kernel hot-swapping!  As of today's commits, everything seems to be working great (mode relevant code was originally committed on the 7th, but a few bugfixes just came down today).  I can now hot swap between EXECoin solo-mining and WafflePool without restarting the miner.

https://github.com/veox/sgminer/commits/master

Maybe we're going to see nscrypt profit switching soon! *crosses fingers*  Come on stratum extension to pass nfactor.

Having not tested it, what is the performance cost of switching kernels?  Does it take half a second, 3 seconds?  I assume since that is in place (kernel swapping I assume is the hard part), it should be pretty trivial to swap nfactors via stratum param...  My guess is it just considers it two different kernels, one with nfactor=X and one with nfactor=Y, similar to switching between scrypt/scrypt-n.

Very interesting news indeed Smiley

It should pass N-factor to opencl kernel as well as other parameters, thus when you want to change the N-factor it do it simply with change of a function call parameter.
hero member
Activity: 693
Merit: 500
Having not tested it, what is the performance cost of switching kernels?  Does it take half a second, 3 seconds?  I assume since that is in place (kernel swapping I assume is the hard part), it should be pretty trivial to swap nfactors via stratum param...  My guess is it just considers it two different kernels, one with nfactor=X and one with nfactor=Y, similar to switching between scrypt/scrypt-n.

Very interesting news indeed Smiley
If it has to compiled the new kernel, it takes a while just like start-up. But, if you've run that nfactor before and the kernel is already compiled, the swap seems to happen in under half a second.  It's impressively fast (to they eye/ear anyway, and compared to what I expected).

The new code looks at the new pool's nfactor and algorithm, and if they are different then the current settings, hotswaps to the needed kernel.

I was just looking at this actually - I had written that functionality into a branch of YACMiner a while ago to support failing over from a scrypt-chacha pool to a scrypt or NScrypt pool, but had to shelve it for other priorities.  It's using the same command line switch I did, but without the needed arguments as they handle N Factor differently (hard coding it, so your miner requires a restart when N switches) - have to give some thought as to whether to keep the same parameter even though it's used vastly differently...
hero member
Activity: 700
Merit: 500
Having not tested it, what is the performance cost of switching kernels?  Does it take half a second, 3 seconds?  I assume since that is in place (kernel swapping I assume is the hard part), it should be pretty trivial to swap nfactors via stratum param...  My guess is it just considers it two different kernels, one with nfactor=X and one with nfactor=Y, similar to switching between scrypt/scrypt-n.

Very interesting news indeed Smiley
If it has to compiled the new kernel, it takes a while just like start-up. But, if you've run that nfactor before and the kernel is already compiled, the swap seems to happen in under half a second.  It's impressively fast (to they eye/ear anyway, and compared to what I expected).

The new code looks at the new pool's nfactor and algorithm, and if they are different then the current settings, hotswaps to the needed kernel.
sr. member
Activity: 322
Merit: 254
Somewhat awesome dev news, IMO, albiet slightly unrelated to WafflePool.

The latest sgminer git now supports pool-specific algorithm/nfactor and kernel hot-swapping!  As of today's commits, everything seems to be working great (mode relevant code was originally committed on the 7th, but a few bugfixes just came down today).  I can now hot swap between EXECoin solo-mining and WafflePool without restarting the miner.

https://github.com/veox/sgminer/commits/master

Maybe we're going to see nscrypt profit switching soon! *crosses fingers*  Come on stratum extension to pass nfactor.

Having not tested it, what is the performance cost of switching kernels?  Does it take half a second, 3 seconds?  I assume since that is in place (kernel swapping I assume is the hard part), it should be pretty trivial to swap nfactors via stratum param...  My guess is it just considers it two different kernels, one with nfactor=X and one with nfactor=Y, similar to switching between scrypt/scrypt-n.

Very interesting news indeed Smiley
hero member
Activity: 700
Merit: 500
Somewhat awesome dev news, IMO, albiet slightly unrelated to WafflePool.

The latest sgminer git now supports pool-specific algorithm/nfactor and kernel hot-swapping!  As of today's commits, everything seems to be working great (mode relevant code was originally committed on the 7th, but a few bugfixes just came down today).  I can now hot swap between EXECoin solo-mining and WafflePool without restarting the miner.

https://github.com/veox/sgminer/commits/master

Maybe we're going to see nscrypt profit switching soon! *crosses fingers*  Come on stratum extension to pass nfactor.
full member
Activity: 182
Merit: 100
Would it be possible to have a setting added where you could get paid out in other coins that just btc? I mean if you are feeding through an exchange api to do the trades (I'm assuming that is the case) then in theory could you set it up so we could pick from a small list of payout coins?
member
Activity: 77
Merit: 10
Yipee!!  Cheesy  Cheesy  Cheesy It's quite a BIG day for me!!  Grin   Grin  Grin I earned my FIRST EVER BITCOIN today!! Yay!! Thanks a bunch for my very first payout, poolwaffle!! I hope to keep earning @ WafflePool for a long time. Smiley

P.S. Now at least my bitcoin wallet has something in it....  Grin


Congratulations Smiley

Don't spend it all at once  Grin

Yes, of course!! Smiley I'm saving it all so that I could buy some good hardware to gain more BTC through alt coins. I've free energy and internet, but not much money to put some good hardware into mining work. Hence I'm saving each and every penny from wherever I can, so that I could buy some good hardware that will make me gain more bitcoins. Grin

P.S. That  is why I even updated my signature today. You noticed that?   Wink
sr. member
Activity: 322
Merit: 254
@PoolWaffle: could you change WP's CSS by chance?  The forced widths are making the page look funny on my 1080x1920 monitor (1080P in portrait orientation).

#pool_stats { width: 1200px; }
#wrapper { width: 1200px; }

to:

#pool_stats { max-width: 1200px; }
#wrapper { max-width: 1200px; }

Done.  Had bumped it up for admin-screen purposes (needed more horizontal space) Smiley
sr. member
Activity: 322
Merit: 254
How often are payouts?  As soon as 0.01 BTC is mined?  Or is it at a time interval (e.g every 24 hours)?

Twice daily, roughly 12hrs apart.
legendary
Activity: 3654
Merit: 8909
https://bpip.org
Yipee!!  Cheesy  Cheesy  Cheesy It's quite a BIG day for me!!  Grin   Grin  Grin I earned my FIRST EVER BITCOIN today!! Yay!! Thanks a bunch for my very first payout, poolwaffle!! I hope to keep earning @ WafflePool for a long time. Smiley

P.S. Now at least my bitcoin wallet has something in it....  Grin


Congratulations Smiley

Don't spend it all at once  Grin
member
Activity: 77
Merit: 10
Yipee!!  Cheesy  Cheesy  Cheesy It's quite a BIG day for me!!  Grin   Grin  Grin I earned my FIRST EVER BITCOIN today!! Yay!! Thanks a bunch for my very first payout, poolwaffle!! I hope to keep earning @ WafflePool for a long time. Smiley

P.S. Now at least my bitcoin wallet has something in it....  Grin
full member
Activity: 138
Merit: 100
You're right.  I'm a tard.  Should of looked at the 'faqs'.  Thanks though.
hero member
Activity: 700
Merit: 500
How often are payouts?  As soon as 0.01 BTC is mined?  Or is it at a time interval (e.g every 24 hours)?

You could try looking in the obvious place, before asking an obvious question:

Quote from: wafflepool.com/faq
When am I paid?

Payouts are processed multiple times a day, and anyone with a balance over 0.01 BTC is automatically paid in order of when you reached the threshold. On Sundays we pay everyone with a balance over 0.001.

---

@PoolWaffle: could you change WP's CSS by chance?  The forced widths are making the page look funny on my 1080x1920 monitor (1080P in portrait orientation).

#pool_stats { width: 1200px; }
#wrapper { width: 1200px; }

to:

#pool_stats { max-width: 1200px; }
#wrapper { max-width: 1200px; }
Pages:
Jump to: