CCminer(SP-MOD) Modded GPU kernels.

[DiCE1904]

Id donate for a faster blake-256 to mine VNL with 980Ti's. FPGAs or not

[sp_]

I have already done 100% faster opensource


https://github.com/sp-hash/ccminer/commits/windows/Algo256/blake256.cu

20 commits since april 2015

blakecoin 117% faster:

https://github.com/sp-hash/ccminer/commit/58d19f08bd34f9f5badc2b8dab86f3f1ea7a1163

Blakecoin 70% faster

https://github.com/sp-hash/ccminer/commit/4b9cb12c40f1e4e7bd35f82df2e424eadd1e6f45

[pallas]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

[DiCE1904]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

Same here, I reached out to two companies. One wants 900Eu for one and when asked on a deal for 10 didn't even get a email back. I guess that kind of order wasn't big enough to warrant a response. I guess lead time for the ztex is a few weeks but then I have to do a bunch of other shit I don't want to do or know how to do.
Thus my interest in it for GPUs

[pokeytex]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

Same here, I reached out to two companies. One wants 900Eu for one and when asked on a deal for 10 didn't even get a email back. I guess that kind of order wasn't big enough to warrant a response. I guess lead time for the ztex is a few weeks but then I have to do a bunch of other shit I don't want to do or know how to do.
Thus my interest in it for GPUs

I recently got lucky and purchased some CM1's on ebay.  The guy was also selling ztex 1.15y's like (15) of them.  I only purchased the CM1's - but if you check everyday they are there from time to time.  There is one ztex 1.15y up there right now. http://www.ebay.com/itm/ZTEX-1-15y-Quad-Spartan-4xSpartan-6-XC6SLX150-USB-FPGA-Bitcoin-Miner-btcminer-/131713140048?hash=item1eaab70550:g:Y-8AAOSwJb9WqZNX

I have GPU's but they are burning like 1300 watts of power 24/7 so I was thinking if I could get my hands on some FPGA's with low power consumption - I would be golden.

[tbearhere]

hmm.... may-be you should start by implementing the blockheader (technically the other part has been there for ages)

Not for Blake-256. Can be mined with fpga's. It's a loosers game. A 17% faster kernal will not be profitable.

Are you saying that Decred could be a losing coin because of fpga's?

[djm34]

hmm.... may-be you should start by implementing the blockheader (technically the other part has been there for ages)

Not for Blake-256. Can be mined with fpga's. It's a loosers game. A 17% faster kernal will not be profitable.

Are you saying that Decred could be a losing coin because of fpga's?

haven't understood either, it seems to me that sp was talking about vnl (while you were talking about decred)

[sp_]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

Private kernals. Asic designers can write a fast blake-256 in a couple of hours. And they have expensive equipment.. custom hardware. Vnc and decred might be profitable for a couple of weeks and then ... look at keccak and whirpoolx. Only mined by a few

[tbearhere]

hmm.... may-be you should start by implementing the blockheader (technically the other part has been there for ages)

Not for Blake-256. Can be mined with fpga's. It's a loosers game. A 17% faster kernal will not be profitable.

Are you saying that Decred could be a losing coin because of fpga's?

haven't understood either, it seems to me that sp was talking about vnl (while you were talking about decred)

Yes I was talking about decred.... but I got my answer ..... fpga's will or could overtake the coin.

[djm34]

hmm.... may-be you should start by implementing the blockheader (technically the other part has been there for ages)

Not for Blake-256. Can be mined with fpga's. It's a loosers game. A 17% faster kernal will not be profitable.

Are you saying that Decred could be a losing coin because of fpga's?

haven't understood either, it seems to me that sp was talking about vnl (while you were talking about decred)

Yes I was talking about decred.... but I got my answer ..... fpga's will or could overtake the coin.

fpga's have to exist in the first place and sometimes they aren't necessarily a lot faster than gpu (it is the case for skein, if I remember well)

[pokeytex]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

Stop looking for shitty ass Spartans and go for something like a Cyclone V, or if you like Xilinx, maybe an Artix-7.

@Wolf0 - are there bitstreams for those FPGAs you mentioned?

[GKar]

I said +30% in the quark algo. Quark doesn't have the SIMD512 and you should know it.

No shit - I assumed if you've got a private Quark, you've also got a private X11. If not, sorry, the question is pointless in that case.

Somebody does. X11 is not profitable with the public bins.

a wellcoded x11 should do 4-5MHASH on the 750ti

I'm getting 3.2MHASH with an EVGA 750ti w/o overclocking on X11 using your private miner.
It took about 3 weeks to get 0.1 DASH on suprnova. 

[bensam1231]

Oh yeah, that's why BTC is still being mined? Not every coin switches to PoS. It's only a 'phase' for some coins, for others it'll last the entirety of the coin. Not sure you guys understand what a race to the bottom is and definitely doesn't apply here, much like pump and dump does not apply to mining.

https://en.wikipedia.org/wiki/Pump_and_dump

http://www.oxforddictionaries.com/us/definition/american_english/a-race-to-or-for-the-bottom

I prefer the wikipedia take on the same expression, both on the economics/efficiency standpoint as well as the banana example, but I agree that my interpretation might be too liberal for some.
https://en.wikipedia.org/wiki/Race_to_the_bottom

I would summarize by stating that the mining competition naturally drives the overall profit opportunity downwards.
This happens both quantitatively (with more miners competing, there's less profits for each/all of them) as well as qualitatively (each new miner improvement - hardware or software - further denies profits to all the existing miners that do not enjoy the same improvement).

Perhaps, but increases in say efficiency in a kernel, will increase profits for those who have it. It also does not inherently reduce the value of a coin simply by having a more efficient kernel, because coin distribution is set. Usually a race to the bottom, means services and whatever sort of product offered will eventually reduce in quality in a effort to produce it as cheaply as possible... In this case there is no such thing. All else being equal, it's entirely possible for profits to increase as well as coin value rises, as higher hashrate does not mean more coins will be produced.

Miners generally speaking aren't supporters, they keep a coin alive.

That's an oxymoron bensam. If miners are keeping a coin alive, they are most definitely supporters.

Generally speaking, it's quite funny how all this mining talk casually ignores the fact that PoW is a race to the bottom. People complaining about small mining profits, would do better to just stop mining, and focus their energies elsewhere. Ah well, I'm probably just at the wrong thread...

Sure and it was meant to sound that way. Because they are a drain on the coin, but at the same time they make sure the network doesn't fall apart. It's a bought service. Miners are a oxymoron. XD

Not sure how GPU mining is characterized by a 'race to the bottom'. Just like miners aren't responsible for 'pump and dumps'. Coins can become more profitable while still minting the same amount of coins per day (relatively speaking). Minting is a controlled release.

Since you're here only for greed, and complain about how little you make, maybe you should get a job flipping burgers  

Perhaps because you're only here to help us out, you should make kernels for free? Oh... wait... double standards... Ah yeah... Miners are the only ones that can't make money. XD

Christ, you're stupid - mining will still be done, but only by those with the lowest costs. Secondly, P & D applies to mining because as soon as the price goes up, multipools and manual switch miners pile right onto it until it goes down again, or difficulty goes up again.

There isn't a drop in quality in order to offer lower costs and lower costs are relative and can only relate to electricity (which is still fixed depending on where you live). In the case of GPUs, the GPUs are a fixed cost and don't change unless you have some sort of insider connection with a GPU manufacturer. Efficiency of miners can increase, but of course you have to pay for development, which arguably is quality, unless you're asserting your masterpieces are shit-drivel that's pumped out of factories in china for cheap?

You're just operating under the assumption that because miners make less, it's a race to the bottom as more miners compete together. They don't lower the standards of their service and on the contrary, more miners means a raise in difficulty of whatever they're mining, therefore they're actually providing higher quality service the more of them there are. Myagui had the closest use and it was a metaphor. Nice try though.

Pump and Dump refers to putting a excessive amount of money into a market and then as other people are buying it up, the person who pumps it dumps for higher prices, essentially taking money of people who bought in during the pump. This is impossible for miners to do as mining income for a coin is fixed relatively speaking. They take part of someone else's pie, but that has nothing to do with a pump and dump. You're thinking of a literal interpretation of a 'pump and dump' rather then the actual definition, which may be confusing as miners throw the term around the forums randomly and they get misconstrued. Once again, completely wrong.

From my side of the coin, you're the one that's fucking retarded... But you know, anyone can fling around insults, just like they can act like they know what they're talking about.

do we really care about merge mining in anyway

You do if you want to help small coins out without the inherent problems of a small network due to no one mining it (because of low value). Basically everything I said, no one read and then Liquid said the same thing only while insulting someone else.

I don't want to be that guy...  ,  but can you move this discussion elsewhere... clearly off topic since a couple of pages...

Coming from someone who does this all the time... Yeah... and I believe we're talking about merged mining, which pertains to mining, which pertains to us mining with... say... CCminer...

First of all, you're confusing WHY a pump and dump matters - you're correct that the act itself has nothing to do with them. You're not seeing that it has an effect on OTHER SHIT. If the price goes up, miners will jump to that coin until the dump. Do I need to draw it in crayon?

As for costs only relating to electricity, you're being narrow minded. First, GPU mining will end for any successful coin, moving onto ASIC mining, at which point the people manufacturing the best ASICs at the lowest cost and having cheap power will dominate... it IS a race to the bottom, at the end of which you have a few companies who can manufacture these devices and operate private farms.

If you feel acting condescending makes you right, draw with that crayon all you want bro, seems it helps you more then I to get your big pictures out there.

You were talking about multipools, now you're particularly talking about miners (I assume you were referring to them as two seperate things before when you singled out multipools).

Alright so miners... now you're stating that miners basically have nothing to do with a pump and dump... why even talk about them in the first place? Good back pedal dude. In other words, miners have nothing to do with a pump and dump. This is either caused by people manipulating market values (which is a true pump and dump) or people designing shitcoins to be used just that way (also made around a pump and dump, large ipos, fast mining period).

A successful coin does not guarantee it will ever see ASICs (pretty certain you know this already). Notice how no new ASICs were made for Scrypt last year? The only new ones are in the planning phase this year. ASICs take a lot of capital to make and it entirely depends on how feasible it is to make. Since this is completely subjective, how big does a coin need to be before it's successful? DASH? Ethereum? Why aren't there any Ethereum ASICs? Oh yeah...

Kernels are about the closest thing you can get to ASICs for some coins due to feasibility of ASICs and there are definitely kernel developers, but they once again take a decent amount of capital before they'll spit anything out. So unless you mean a race in technology is the equivalent of a race to the bottom you're once again still trying to make something work that has nothing to do with one another.

Nice try dude, still wrong on both accounts... Even 'wronger' now that you're trying to maneuver around things by abandoning your original arguments to make yourself seem right.

hmm.... may-be you should start by implementing the blockheader (technically the other part has been there for ages)

Not for Blake-256. Can be mined with fpga's. It's a loosers game. A 17% faster kernal will not be profitable.

FPGAs and ASICs take awhile to develop (especially ASICs). Vanillacoin is still profitable for instance. Hardware is a completely different ballgame from software and production cycles take much longer and happen in smaller spurts, especially on here where people are basically making everyone recycle the wheel.

[sp_]

I said +30% in the quark algo. Quark doesn't have the SIMD512 and you should know it.

No shit - I assumed if you've got a private Quark, you've also got a private X11. If not, sorry, the question is pointless in that case.

Somebody does. X11 is not profitable with the public bins.
a wellcoded x11 should do 4-5MHASH on the 750ti

I'm getting 3.2MHASH with an EVGA 750ti w/o overclocking on X11 using your private miner.
It took about 3 weeks to get 0.1 DASH on suprnova.  

Why do you mine x11, it is lower payed than the other algos:

Current prices at nicehash:



You get 200% more BTC if you switch to quark/x15/neoscrypt/lyra2v2 (use a profit switcher software), and then you can buy Dash afterwards on an exchange.

With the current prices it will take 1 EVGA 750ti 7,9 days to mine 1 DASH (if you mine quark) (6.5MHASH with the private miner)

[s7icky]

I said +30% in the quark algo. Quark doesn't have the SIMD512 and you should know it.

No shit - I assumed if you've got a private Quark, you've also got a private X11. If not, sorry, the question is pointless in that case.

Somebody does. X11 is not profitable with the public bins.
a wellcoded x11 should do 4-5MHASH on the 750ti

I'm getting 3.2MHASH with an EVGA 750ti w/o overclocking on X11 using your private miner.
It took about 3 weeks to get 0.1 DASH on suprnova.  

Why do you mine x11, it is lower payed than the other algos:

Current prices at nicehash:



You get 200% more BTC if you switch to quark/x15/neoscrypt/lyra2v2 (use a profit switcher software), and then you can buy Dash afterwards on an exchange.

With the current prices it will take 1 EVGA 750ti 7,9 days to mine 1 DASH (if you mine quark) (6.5MHASH with the private miner)

He prob uses ur maths. 5% on everyyyythaaaaaang

[induktor]

Hi
I am trying to update my miners to CUDA 7.5 so I can run _sp's lattest version of ccminer.
runing lubuntu 14.04 x64 from a flash drive, driver was updated to 352.63
I can call nvidia-smi and watch parameters for all cards fine.

installed cuda 7.5 and created the symlink to /usr/local/cuda that points to the 7.5 folder.

LD_LIBRARY_PATH points to the correct cuda 7.5

ccminer won't start, ever....

now I pointed LD_LIBRARY_PATH to cuda 6.5 folder (which is still there) and I can mine perfectly

I am using _sp's ccminer 1.5.74
any ideas?, I probably missed something but at this hour (3 am) I can't tell

thanks!

[sp_]

1.5.74 is cuda 6.5.

1.5.78 is cuda 7.5.

Get the latest version.from git.

[chrysophylax]

I don't know how many compatible FPGAs you guys have, but I'm finding it really hard to get my hands on even a single one.
They aren't produced any longer and on ebay they are rare and very expensive.
Unless someone codes support for the new chips (newer than spartan6 and sold on multi-chip boards)...

Private kernals. Asic designers can write a fast blake-256 in a couple of hours. And they have expensive equipment.. custom hardware. Vnc and decred might be profitable for a couple of weeks and then ... look at keccak and whirpoolx. Only mined by a few

big deal ...

even if they do - the fact is that decred is being released very shortly - and you CAN but WONT implement the kernel changes to it ...

im a HUGE supporter of you sp ... but lately - its been quite disappointing for me as not only a miner - but a donor to you for even the products i CANNOT use from you ...

this is not good mate ... not good at all ...

and im starting to lose interest very quickly - im quite disappointed to say ... optimizations or not - you are losing my support and vote with the way you are handling this ...

#crysx

[sp_]

You can always switch to windows for a 5-10% boost in the most profitable algos.

Sp-mod private #6 is soon ready for the donators. with a 1% boost in the quark algo..

0.1BTC to subscribe.

[chrysophylax]

You can always switch to windows for a 5-10% boost in the most profitable algos.

Sp-mod private #6 is soon ready for the donators. with a 1% boost in the quark algo..

0.1BTC to subscribe.

i wont do it mate ...

windows is not for me - and if you continue along this path - thefarm will have to move elsewhere ... and its growing - just at a standstill right now due to my personal changes at the moment ...

when it gets into the new premises - it will grow from there - as it will have the room and the power expansion to do so ...

it will never go windows ... ever ... its too much hassle ... you continue to forget this is not just one or two machines here sp ...

i guess we know where your support will be - and i know exactly what to do with thefarm now ...

plans will start this week for other choices ...

its been good sp - but i wont follow you down that rabbit hole that means i have to switch to a cumbersome piece of garbage for an os - just to get a few percentage - AND - get a crap load of headaches ... no tanx ... linux just runs - and runs well ... with almost no maintenance issues ...

ill keep watching this thread - but i need to move on to a developer that is willing to support ME and all the others that are running linux ...

you know damned well i have donated more than 0.1btc - and the support you show me now is simply to switch to windows? ... nup - now this stops for me ...

its been good mate - and you are very talented and a good dev ...

tanx for all you prior support also ...

#crysx