CCminer(SP-MOD) Modded GPU kernels.

[smolen]

actually what is infected by gpl (the detail) ?

I'm at Stratum code ATM, one of reference implementations by Slush is under AGPL, all other I was able to find are under GPL.

[smolen]

Yeah, all infectious. I'm using this, and writing in straight C: https://mining.bitcoin.cz/user-manual/stratum-protocol#stratum

Also, keep in mind Jansson is MIT licensed - feel free to use it!

Actually, GPL is good - in the long term. I already released some code under it and plan to use it for one of my for-fun megaprojects - and be damn sure nobody will be able to just grab it an sell. And way too protected Stratum implementation was a good excuse to spare some time to finally play with boost::asio 

im looking at setting something up to allow for donations to be placed specifically FOR the development AND the developers such as yourself ...

Probably, money will not work, people don't donate much, but the reward already is here. I have my leaps sealed with some topics and do withhold information, that makes free discussion on the edge of current technology pretty amazing, a breath of fresh air in NDA and patent infested world.

Well, no, you don't have someone as good reversing it. Not because there is no one, but because you simply need to make it not worth their time to do so.

I'm implementing a Stratum pool client first. I also plan to support getblocktemplate.

Good solution would be a pool with custom precalculations of the first steps of hash and custom miner specially for it, that way reversing will be mostly harmless

[chrysophylax]

Yeah, all infectious. I'm using this, and writing in straight C: https://mining.bitcoin.cz/user-manual/stratum-protocol#stratum

Also, keep in mind Jansson is MIT licensed - feel free to use it!

Actually, GPL is good - in the long term. I already released some code under it and plan to use it for one of my for-fun megaprojects - and be damn sure nobody will be able to just grab it an sell. And way too protected Stratum implementation was a good excuse to spare some time to finally play with boost::asio 

im looking at setting something up to allow for donations to be placed specifically FOR the development AND the developers such as yourself ...

Probably, money will not work, people don't donate much, but the reward already is here. I have my leaps sealed with some topics and do withhold information, that makes free discussion on the edge of current technology pretty amazing, a breath of fresh air in NDA and patent infested world.

Well, no, you don't have someone as good reversing it. Not because there is no one, but because you simply need to make it not worth their time to do so.

I'm implementing a Stratum pool client first. I also plan to support getblocktemplate.

Good solution would be a pool with custom precalculations of the first steps of hash and custom miner specially for it, that way reversing will be mostly harmless

well - let me know if i can help in anyway ...

#crysx

[smolen]

Actually, they CAN just grab it and sell - just not enforce more restrictions on it, nor use it in non-GPL'd code. I may release a miner shell - that is, without algorithm implementations - under a different license.

In this case the area is too narrow and the reputation costs much more than quick money. Well, anyway that project is in deep lethargy.

And that is quite a good solution, but I would prefer my miner to also work with existing pools, and solo mining. I also want explicit support for the user's ability to fuck with the coinbase tx and add custom transactions - nonstandard or not, as long as they're valid.

Smelter almost got MIT licensed code from Bitcoin core for coinbase tx manipulations, but not-invented-here-syndrome took a victory over laziness

[sp_]

The total qubit hash in the world is around 30 ghash. With the current prices around 7.5Btc is generated every day.

A kernal that does 15 mhash on the amdn280x (300%) faster could have a fee of 33% and still ppl will use it.
10mhash to the miner. 5 to the developer.2.5 btc in my pocket every day 365 days a year.To avoid gpl,
 i just need to pay 1% to pooler and 1% to christian. Ccminer for amd.

[sp_]

1% of the total hashrate of the world is an offer they can't refuse 

[sp_]

1% of the total hashrate of the world is an offer they can't refuse 

1% of the total Qubit hashrate

and quark, fresh, whirlpoolx,x15 etc..

x11 and x13 is also possible, but you've made it very difficult to optimize more with your nice binaries

[ol92]

1% of the total hashrate of the world is an offer they can't refuse 

1% of the total Qubit hashrate

and quark, fresh, whirlpoolx,x15 etc..

x11 and x13 is also possible, but you've made it very difficult to optimize more with your nice binaries

Well I hope you won't release such an amd miner qubit optimized : qubit is one the most profitable algo with maxwell nvidia : this will end as soon as an optimized amd miner is out.

[smolen]

The total qubit hash in the world is around 30 ghash. With the current prices around 7.5Btc is generated every day.

A kernal that does 15 mhash on the amdn280x (300%) faster could have a fee of 33% and still ppl will use it.
10mhash to the miner. 5 to the developer.2.5 btc in my pocket every day 365 days a year.To avoid gpl,
 i just need to pay 1% to pooler and 1% to christian. Ccminer for amd.

I'm considering it, once my miner is ready - I've a decent Qubit.

Interesting times ahead. The winner takes half of hashrate advantage over the second best miner, all other developers get nothing.

[smolen]

Yeah, I've not seen sp_ do any AMD work. CUDA is pretty easy in comparison, so I'm more worried about you beating my current 10.75MH/s per 280X on Qubit, smolen.

Without taking any obligations and disclosing too much... Quite sure that you, pallas and some other developers here are ahead of me when it comes to OpenCL. dga, Supercomputing (and other) are great at math and CS - I truly enjoy playing here, but understand my limits. For fun, brawling and such there are simple algos; we have had a nice competition with Whirlpool, may be next arena will be Skein or something else, small both in code and money, as a challenge Qubit is just too big
Talking about money, I think I see a way to milk big farms and private miners without doing any GPU code - if they are at, say, wrong local Nash equilibrium. Perhaps this idea will fail, I just have to try. Another crazy idea, eating my spare time, is smart codegenerator. Not exactly compiler, not exactly SAT. Some good code, unpublished, perhaps patentable tricks, a perspective to profit not just from mining - and nothing ready for production yet.
You see, it's possible that I'll jump in and take your profits, but I'm driven not only by money

[AliMan]

Hey sp_, seeing that VNL has hit the sky, any tweaks you think I could try out to squeeze a little more hash with the whirlpoolx algo, if possible?

[MaxDZ8]

Again, what's up with Qubit? Lots of fuss about it recently.

[AliMan]

Again, what's up with Qubit? Lots of fuss about it recently.

Yea I don't really see it being profitable at all, DGB was the best of it, but that stopped being profitable sometime back.

[totoy]

Again, what's up with Qubit? Lots of fuss about it recently.

Yea I don't really see it being profitable at all, DGB was the best of it, but that stopped being profitable sometime back.

this one https://bitcointalk.org/index.php?topic=1006527.0

[Grim]

@ sp_ couldn't you optimize scrypt-jane algo nf 15+

you could probably sell it with

cryptonight
spreadx11

for 0.3 btc for all of them

 

[tbearhere]

Again, what's up with Qubit? Lots of fuss about it recently.

Yea I don't really see it being profitable at all, DGB was the best of it, but that stopped being profitable sometime back.

this one https://bitcointalk.org/index.php?topic=1006527.0

Yup I missed that coin. Qubit too 

[AliMan]

Again, what's up with Qubit? Lots of fuss about it recently.

Yea I don't really see it being profitable at all, DGB was the best of it, but that stopped being profitable sometime back.

this one https://bitcointalk.org/index.php?topic=1006527.0

Just took notice of this today.

[scryptr]

SP_   bmw fix build error --

Attempted to build sp_ 's latest mod (v1.5.44+, just now) and received this error:

"In file included from /usr/include/c++/4.8/cstdint:35:0,
                 from qubit/qubit_luffa512.cu:21:
/usr/include/c++/4.8/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support is currently experimental, and must be enabled with the -std=c++11 or -std=gnu++11 compiler options.
 #error This file requires compiler and library support for the \
  ^
make[2]: *** [qubit/qubit_luffa512.o] Error 1
make[2]: *** Waiting for unfinished jobs...."

My build platform is Ubuntu 14.04.2, and it has built all sp_ mods up to 1.5.44 succesfully.  I built with the command
"bash build.sh" within the git build directory.       --scryptr

[chrysophylax]

SP_   bmw fix build error --

Attempted to build sp_ 's latest mod (v1.5.44+, just now) and received this error:

"In file included from /usr/include/c++/4.8/cstdint:35:0,
                 from qubit/qubit_luffa512.cu:21:
/usr/include/c++/4.8/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support is currently experimental, and must be enabled with the -std=c++11 or -std=gnu++11 compiler options.
 #error This file requires compiler and library support for the \
  ^
make[2]: *** [qubit/qubit_luffa512.o] Error 1
make[2]: *** Waiting for unfinished jobs...."

My build platform is Ubuntu 14.04.2, and it has built all sp_ mods up to 1.5.44 succesfully.  I built with the command
"bash build.sh" within the git build directory.       --scryptr

that is strange scrptr ...

ive built it with fedora 19 x64 and it runs beautifully ...

though i do have both stdc++ and gnuc++ installed in the systems ...

#crysx

[scryptr]

SP_   bmw fix build error --

Attempted to build sp_ 's latest mod (v1.5.44+, just now) and received this error:

"In file included from /usr/include/c++/4.8/cstdint:35:0,
                 from qubit/qubit_luffa512.cu:21:
/usr/include/c++/4.8/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support is currently experimental, and must be enabled with the -std=c++11 or -std=gnu++11 compiler options.
 #error This file requires compiler and library support for the \
  ^
make[2]: *** [qubit/qubit_luffa512.o] Error 1
make[2]: *** Waiting for unfinished jobs...."

My build platform is Ubuntu 14.04.2, and it has built all sp_ mods up to 1.5.44 succesfully.  I built with the command
"bash build.sh" within the git build directory.       --scryptr

that is strange scrptr ...

ive built it with fedora 19 x64 and it runs beautifully ...

though i do have both stdc++ and gnuc++ installed in the systems ...

#crysx

GNUC++11  --

With "sudo apt-get install gnuc++", I installed the gnuc++ compiler.  I still get the same compile error, however, just as I quoted it above.  Do I need to edit the build.sh file, or use a command line option when running "bash build.sh"?

--scryptr