CCminer(SP-MOD) Modded GPU kernels.

[sp_]

If someone could point me to a good simple link for compiling for Lunix or windows I'd be glad to give it a shot.
I spent about 3 hours last night hunting and trying different walk threw, but most were not up to date ones.
So not much luck so far. I'm using Linux Mint 17.2 64bit for this rig, but only has a 660Ti in it at the moment.
Most of my rigs are open air so not much trouble changing video cards around for testing.

I have buildt release 63 for you to test.

[majinboo]

If someone could point me to a good simple link for compiling for Lunix or windows I'd be glad to give it a shot.
I spent about 3 hours last night hunting and trying different walk threw, but most were not up to date ones.
So not much luck so far. I'm using Linux Mint 17.2 64bit for this rig, but only has a 660Ti in it at the moment.
Most of my rigs are open air so not much trouble changing video cards around for testing.

I have buildt release 63 for you to test.

Looks like colors are broken on r63 but it seems to add around 100 khs on lyra2v2 for gtx 750.

[zTheWolfz]

If someone could point me to a good simple link for compiling for Lunix or windows I'd be glad to give it a shot.
I spent about 3 hours last night hunting and trying different walk threw, but most were not up to date ones.
So not much luck so far. I'm using Linux Mint 17.2 64bit for this rig, but only has a 660Ti in it at the moment.
Most of my rigs are open air so not much trouble changing video cards around for testing.

I have buildt release 63 for you to test.

Nice work!   Right at 300kh/s above what I was getting for Quark.
Edit: X11 5240kh/s, about the same as before.
Testing lyra2rev2 now 5857kh/s unsure of my numbers before on this, neoscrypt up next.

[bensam1231]

Yeah same... This is why it'd be great if there was just a percentage you paid into the miner. Everyone pays and reaps the rewards. If you guys are publicly disclosing your numbers, I'm sure there are big miners who have their own coders on their payroll.

Looking at it another way, if you release the miner with a 2% fee, how long would it take for miners who use it to reach your 'goal' for a donation bounty? Even if someone takes it and guts the fee, and assuming all miners are bad people and they use the fee less miner regardless of knowing it's supporting your work, you probably could make the donation amount before it happens.


May I suggest as a test case, releasing a version of CCminer with a 2% fee and 10% greater hashrate for Lyra2v2 and Quark. Then go from there? You aren't giving out all your cards. You can see how fast the miner is gutted and see how many people continue using the miner after a 'fee free' version pops up. It would give you some idea of the income you could generate with a fee and whether or not it's worth your time to pursue such an endeavor.

it was already done with vanillacoin, but it is a mess to code considering you have to send the fee somewhere...
ie a pool, then you have to add several pools (then restrict usage to those pool) etc...

probably better to do that with an existing (non optimized version)   rather than releasing new code to "test"... (nice try...  )

Also why 2% ? This would have to be scaled over the %age optimization and not some random number.
say something gives 100% increase (ie double the hashrate), I am pretty sure that it is worth 10% fee at least for a while

nb: the easiest way to impose a fee, would probably to find some arrangement with pool admin and make binaries working only on that pool with part of the fee going to the dev (like it was done for some prime number algos)

What sort of incentive is there to use a fee based miner when the current version is free, besides it's doing the right thing? And are you really bitching about having to do a little bit of work to earn money?It really isn't that hard to setup a pool and you know it, you just don't want to have to do the tiniest bit extra work to make it happen (which shows the whole reason we're stuck where we are right now).

Because people see a 10% fee and they hate you. Then they tear out your fee to spite you. It's supposed to be a agreeable amount for everyone. 2% is low enough where people will accept it and use it, even if a 'fee free' alternative is released. 2% is affordable. If a 'fee free' version comes out people will see 10% additional profit, instead of 2%. It then becomes about being competitive with other miners are earning more money and expanding much faster instead of doing the right thing.

Also a percent is a percent. If you have a 100% faster kernel, you will double how much you're making (if nothing else changes) through the donation. It scales. It isn't a flat fee. Continued improvements will reward the developer with increased profits. Ideally improvements such as the difference between quark public and quark private wont be that big, the only reason they are right now is because people have been holding back.

nb: the easiest way to impose a fee, would probably to find some arrangement with pool admin and make binaries working only on that pool with part of the fee going to the dev (like it was done for some prime number algos)

This has already been done multiple times by srcxxx (now dashminer).
I instead am supporting the idea of a multipool (based on yaamp) which sends a small percent to the most active developers.
People would simply mine on it like they usually do.

You can use whatever sort of pool you want to make it happen. A donation pool is really no different then a donation system right now. I guess some people prefer it, but if it came down to it, I'd just send some BTC. It's more inconvenient then just sending BTC if someone is more inclined. A fee is there because everyone then does it and there is straight income from the product. The developer improves their kernels, they see a improvement in the amount of income they get. It's incentive to continue to support the product and the end user doesn't need to do anything.

I'm pretty certain if a donation pool was setup, the current situation will not change and we'll go back to the same system where developers hold back because they feel they aren't getting adequate support for their miner.

[antonio8]

@sp_

Been having the craziest issue with all releases after 59. This is what I see when running the miner on just about any of the version and when trying to solo I always get rpc errors that I don't get with any other miner.

[bathrobehero]

What sort of incentive is there to use a fee based miner when the current version is free, besides it's doing the right thing? And are you really bitching about having to do a little bit of work to earn money?It really isn't that hard to setup a pool and you know it, you just don't want to have to do the tiniest bit extra work to make it happen (which shows the whole reason we're stuck where we are right now).

Actually, if the miner is good enough, this is kind of bad. One pool getting MASSIVE hashrate, or multiple pools controlled by one person? It's a very quick road to centralization.

I really don't think a united green team would case an issue related to centralization but even then, we could cross that bridge once we get there.
I think I'm just talking to the wall once again as I got used to being ignored when it comes to serious discussions but what the hell, I'll bite once again; so really as many people are working on optimizing (cuda based) miners there really isn't any improvements towards monetizing it. Personally, I'm using a closed source ccminer fork for about six months now so I'm not really bothered by what's going on with the open source scheme but still, I think there should have been some improvements regarding monetization. Unfortunately there was none.

There's always the argument towards open source fee based miners but there wasn't even a (major) release to at least try the freaking thing. In reality, we, in this thread are in the tiny minority when it comes to nvidia based mining and sites like cryptominingblog and several other sites are the ones distributing the software towards potentially thousands of miners who might not even know what BCT is or might not even speak English. Surely, an open source fee based miner would at least worth a try by now to see how much would it earn from both dedicated and imbecile miners - even if it's a compelte failure - but there wasn't any (major) release of it. Just look at how many people would donate to tsiv towards his/her lyra2rev2 miner.

And then there's yaamp which got open sourced not so recently and even that opportunity got ignored. Maybe I'm just oversimplifying it, but a yaamp based multipool with decent fees and full transparency would really motivate coders to improve code especially if their fee was based on their measured improvements (%) on a per algo basis after an established baseline.

I really think the coders of the community should put their heads together and come up with something reasonable because I'm sure appreciative fans wouldn't mind paying for improvements.

[GingerAle]

(snip)

IMO, closed-source fee-based miners without pool restrictions is best for the coin AND the coin communities.

Can you elaborate?

[djm34]

Yeah same... This is why it'd be great if there was just a percentage you paid into the miner. Everyone pays and reaps the rewards. If you guys are publicly disclosing your numbers, I'm sure there are big miners who have their own coders on their payroll.

Looking at it another way, if you release the miner with a 2% fee, how long would it take for miners who use it to reach your 'goal' for a donation bounty? Even if someone takes it and guts the fee, and assuming all miners are bad people and they use the fee less miner regardless of knowing it's supporting your work, you probably could make the donation amount before it happens.


May I suggest as a test case, releasing a version of CCminer with a 2% fee and 10% greater hashrate for Lyra2v2 and Quark. Then go from there? You aren't giving out all your cards. You can see how fast the miner is gutted and see how many people continue using the miner after a 'fee free' version pops up. It would give you some idea of the income you could generate with a fee and whether or not it's worth your time to pursue such an endeavor.

it was already done with vanillacoin, but it is a mess to code considering you have to send the fee somewhere...
ie a pool, then you have to add several pools (then restrict usage to those pool) etc...

probably better to do that with an existing (non optimized version)   rather than releasing new code to "test"... (nice try...  )

Also why 2% ? This would have to be scaled over the %age optimization and not some random number.
say something gives 100% increase (ie double the hashrate), I am pretty sure that it is worth 10% fee at least for a while

nb: the easiest way to impose a fee, would probably to find some arrangement with pool admin and make binaries working only on that pool with part of the fee going to the dev (like it was done for some prime number algos)

What sort of incentive is there to use a fee based miner when the current version is free, besides it's doing the right thing? And are you really bitching about having to do a little bit of work to earn money?It really isn't that hard to setup a pool and you know it, you just don't want to have to do the tiniest bit extra work to make it happen (which shows the whole reason we're stuck where we are right now).

Actually, if the miner is good enough, this is kind of bad. One pool getting MASSIVE hashrate, or multiple pools controlled by one person? It's a very quick road to centralization.

this what always happened with primecoin and some other coins... now obviously this isn't ideal either but at the moment, most of the alcoin only have a couple of pool in most of the case.
The question what would prevent the pool to play the game ? (clearly in that case no one is getting fooled, nor the user, nor the devs...)

[djm34]

(snip)

IMO, closed-source fee-based miners without pool restrictions is best for the coin AND the coin communities.

Can you elaborate?

well... honestly the best case is when the dev and the community contact a dev to develop miners and so on (and obviously dev get paid to do that... and in that case I have no objection in releasing the code.
But coding 2 weeks for a project for a couple of beers... I don't feel interested... (considering my code will be used in sp miner and in the final he will get the donation...

[GingerAle]

(snip)

IMO, closed-source fee-based miners without pool restrictions is best for the coin AND the coin communities.

Can you elaborate?

Well, imagine I develop the miner and the pool, and I lock my miner to one pool. If my miner is fast enough, people will switch to it, otherwise they will be at a severe disadvantage. This causes a miner monopoly, and since the pool is locked in the miner, it creates a pool monopoly. Now, one pool easily will have well over 50% hashrate, allowing them to attack the chain if they so choose.

yeah, I agree about not locking a miner to the pool, I was just interested about the closed source. I assume its to ensure receiving compensation for the effort of coding, which I also agree with, but I was just curious if there's anything more.

My only experience is the Monero, and its black and white comparing what has happened with nvidia vs. amd mining - claymore has updated his software constantly, whereas tsivs miner not so much. And, as he stated, its probably because the donations haven't been pouring in.

[bathrobehero]

What sort of incentive is there to use a fee based miner when the current version is free, besides it's doing the right thing? And are you really bitching about having to do a little bit of work to earn money?It really isn't that hard to setup a pool and you know it, you just don't want to have to do the tiniest bit extra work to make it happen (which shows the whole reason we're stuck where we are right now).

Actually, if the miner is good enough, this is kind of bad. One pool getting MASSIVE hashrate, or multiple pools controlled by one person? It's a very quick road to centralization.

I really don't think a united green team would case an issue related to centralization but even then, we could cross that bridge once we get there.
I think I'm just talking to the wall once again as I got used to being ignored when it comes to serious discussions but what the hell, I'll bite once again; so really as many people are working on optimizing (cuda based) miners there really isn't any improvements towards monetizing it. Personally, I'm using a closed source ccminer fork for about six months now so I'm not really bothered by what's going on with the open source scheme but still, I think there should have been some improvements regarding monetization. Unfortunately there was none.

There's always the argument towards open source fee based miners but there wasn't even a (major) release to at least try the freaking thing. In reality, we, in this thread are in the tiny minority when it comes to nvidia based mining and sites like cryptominingblog and several other sites are the ones distributing the software towards potentially thousands of miners who might not even know what BCT is or might not even speak English. Surely, an open source fee based miner would at least worth a try by now to see how much would it earn from both dedicated and imbecile miners - even if it's a compelte failure - but there wasn't any (major) release of it. Just look at how many people would donate to tsiv towards his/her lyra2rev2 miner.

And then there's yaamp which got open sourced not so recently and even that opportunity got ignored. Maybe I'm just oversimplifying it, but a yaamp based multipool with decent fees and full transparency would really motivate coders to improve code especially if their fee was based on their measured improvements (%) on a per algo basis after an established baseline.

I really think the coders of the community should put their heads together and come up with something reasonable because I'm sure appreciative fans wouldn't mind paying for improvements.

IMO, closed-source fee-based miners without pool restrictions is best for the coin AND the coin communities.

At this point I think anything is better. I mean we have nothing but the random beer system and even though I try to donate frequently to most of the devs it's a really crappy and inconsistent system and I don't think I'm saying anything new when I say that some of the devs are also not very happy about it...

Anyway, I didn't mean pool restrictions but if you take into consideration how much issue something as simple as the difficulty difference caused with lyra2rev2 while the solution was as simple as -f 2 or -f 0.5 which should be pretty obvious for seasoned miners, I'm pretty sure a lot of people would just join a multipool that is guaranteed to work with the latest ccminer releases. Some potential centralization there but that could be fixed somewhat easily.
And if it would be transparent and not bot-ridden I'm sure more miners would join. And of course knowing that it makes the coders happy would also motivate people to join.
And motivating new coders to try and join the race would also be pretty good for everyone. Or just motivating the usual coders to implement something new that could earn some decent money (*khm*... hashimoto) would also be beneficial.

I think unfortunately this will remain a child's dream as it would require a lot of work and collaboration but one could hope as I think something like this would be needed to solidify the future of open source nvidia mining.

[sp_]

@sp_
Been having the craziest issue with all releases after 59.

I submitted a fix for this. The problem is that the pool is switching the difficulty, and the miner submit results for the old difficulty. I am now aborting the threads and force a quick restart when the pool changes the diff..

[sp_]

The fix didn't work.. I have reproduced the issue on p2poop.pl. Will work on it...

[mendoza1468]

ccminer-1.5.62-git-spmod  (Gpu clock=1366 mhz) (Mem clock=3004 mhz)

Quark
ccminer.exe -i 22.9 -r 5 -R 5 -m --cpu-priority 5 -q -a quark -o stratum+tcp://quark.usa.nicehash.com:3345 -u x -p x

2 X Gtx 970 = 33 100 kh/s
Gtx 970 = 16 550 kh/s per card average
0.0132 btc/day atm

Lyra2v2
ccminer.exe -i 19.4 -r 5 -R 5 -m --cpu-priority 5 -q -a lyra2v2 -o stratum+tcp://lyra2rev2.usa.nicehash.com:3347 -u x -p x

2 X Gtx 970 = 19 825 kh/s
Gtx 970 = 9 912 kh/s per card average
0.00793 btc/day atm

ccminer-1.5.63-git-spmod  (Gpu clock=1366 mhz) (Mem clock=3004 mhz)

Quark
ccminer.exe -i 22.9 -r 5 -R 5 -m --cpu-priority 5 -q -a quark -o stratum+tcp://quark.usa.nicehash.com:3345 -u x -p x

2 X Gtx 970 = 33 350 kh/s (+250 kh/s)  
Gtx 970 = 16 675 kh/s per card average
0.0133 btc/day atm

Lyra2v2
ccminer.exe -i 19.4 -r 5 -R 5 -m --cpu-priority 5 -q -a lyra2v2 -o stratum+tcp://lyra2rev2.usa.nicehash.com:3347 -u x -p x

2 X Gtx 970 = 19 900 kh/s (+75 kh/s)  
Gtx 970 = 9 950 kh/s per card average
0.00796 btc/day atm

[sp_]

-i 22.9 is not the best setting for gtx 970. -i 22.9 is good on 2gb cards like 750ti and 960 (and the 1gb 750) uses around 800 MEG of gpu memory.

-i 24 is the default for quark on 970/980

[sp_]

In 24-oct-2014 the gigabyte 970 was mining quark at around 10.8 MHASH

in sp-mod release 63 the same card is doing 16.7MHASH.

Small increases, becomes big when you add them up...

http://cryptomining-blog.com/page/2/?s=gtx+970+mining+performance

[sp_]

Without the sp-mod:

quark
2 X Gtx 970 = 21600  kh/s](-11500Kh/s)
Gtx 970 = 10 800 kh/s per card average
0,0086 btc/day atm

[restless]

on last .63 build my 750Ti has broken colors (have to run with --no-color) and I see no speedup compared to .61
All stock, only the CPu is relatively old&slow (E5200), but this was no issue so far

[pallas]

one of my rigs has 5 970s and a single core shitty sempron, still ccminer only uses some percent of cpu time :-)

[sp_]

on last .63 build my 750Ti has broken colors (have to run with --no-color) and I see no speedup compared to .61
All stock, only the CPu is relatively old&slow (E5200), but this was no issue so far

Wich algos are you mining?

You should get faster rate in quark and lyra2v2