Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 706. (Read 2347641 times)

Alexis have been busy. The last improvement was submitted 19 hours ago..

https://github.com/alexis78/ccminer/commits/windows

I think I need to do version #7 with more hash for the donators.

Why don't you compare my modded kernal with the 1.7.2 version. (the first version) How many percent faster is it?..

My modded cryptonight kernal is using less memory so that you can run with optimal intensities on windows. I have done more. It is faster. What launch config are you using?

he didn't fix shit, my version bought from him with cryptonight speed increase was a fraud, no real speed increase, i just have the same speed with the tsiv version by increase the intensity i tried it LOL

please SP refund my 0.05

And how much faster is the latest sp-modded version than the first version found here: (vanilla and decred)

http://cryptomining-blog.com/6619-updated-ccminer-1-7-2-fork-by-tpruvot-with-vanillacoin-and-decred-support/

Try to run alexis version without the -i parameter in the Bat.

edit: link updated.

Some Tests Decred#8+Vanilla#6 VS Alexis78 Lastest Git compiled version(cuda 7.5)
2 cards (MSI 970 and MSI 960) overclocked to +200Mhz Core. Intensy not changed and  set from SP_version Bat files
Blake
SP-1680+1038=2718
Alexis78 - 1836+1135=2971
Vanilla
SP - 3441+2098=5539
Alexis78 - 3467+2114=5581
What can I say, the version from Alexis78 free and the speed on it is the same or even slightly faster than the paid version of SP. SP - there is work to do!
I do not regret that I bought the paid version, SP constantly improving its Assembly. Most importantly he was not involved in collecting money for the new "algorithms" with each new version of "ALL in one miner"
For those who haven't bought his version of the miner, there is reason to think of expediency of purchase
Some screens:
SP_Version on Vanilla:

Alexis78 on Vanilla:

The idiots are making alot of bitcoin with my optimized kernals. That's why they donate. To increase their profit.
I just opensourced decred #4, why don't you build it and test it. It works, and was very profitable when it came out. In the beginning of decred  a 980ti could earn up to 10$ a day mining with my private bins..

always insane promises... every one knows, you haven't done shit for month (except a few idiots here who buy from you sorry for the idiots, but they should be realistic if all other dev tell them there is no speed increase, while we don't really say that to each other, it might be we are right...

Currently SP you are just running a scam.

no - i didnt lose passwords or anything to the email ... just having issues with it currntly ... the email will be fixed when i can get my isp online to help with the issue ...

its their end so i think when they can resolve it when they can figure out what is happening on their server ... im just a little impatient - thats all ...

i havent even got the test machien running yet - and thefarm is down until i find a new home for it ... the small warehouse i was looking at didnt go through - so im left with keeping thefarm offline ... so no mining on this end in any big way - just a few small gpus ...

im selling off the vast majority of gb 750ti oc lp cards ( i was going to sell all - but i may keep a few ) so that i can upgrade thefarm to much more powerful cards ...

anyway - i guess i can wait ... its just that i CAN test the latest kernels on my gb 750ti oc lp card i have mining currently - but its only ONE card at the moment ...

as for the issues with speeds and miners ... i think a stock test would suffice to show the hashrates of the miners ...

i think you know how i work with testing - and know how i do this ... fedora 23 x64 with cuda 7.5 ...

and after 3 years of being on here ( and on the way to becoming legendary ) THEN having to start my profile here again on a newbie account - im not impressed with the way bct themselves have been 'helping' ( or the lack thereof ) with my issue ... so ill be changing the way i keep posts - on my own forum ( currently granite - but that will change to the umbrella name for all my projects ) ...

so this week will be my 'setting up the test miner and the beast miner' week ...

two machines ...

test - 1 x gb 750ti oc lp + gb 980ti g1 ...
beast - 3 x gb 980ti g1 + 2 x gb 980ti extreme ...

beast is my friends machine ( the one i built that STILL has the 6 card issue - so only 5 cards mine ... if you remember from my chrysophylax profile ) ... this is the machine i will be using to mine in a couple of days - with the test machine actually doing the tests on the kernels ...

alexis78 is NOT a newbie here - just his profile - as mine is ...

i will be testing all the kernels i have and will be placing the results on pastebin as we did earlier ( before i locked myself out of the chrysophylax profile ) ...

my email 'should' be functional again in a day or two - so ill email you direct from there so you know its 'me' ...

tanx ...

#crysx

Bether kernal?
Pallas just did a 10% bether decred kernal but when I build it, I get +1%  (windows? linux?)
My opensource decred #4 is 14% faster than the pallas kernal.
Decred #8 private more than 20% faster.

(750ti)

Alexis used my blake-256 implementation in his vanilla mod and opensourced it with a slow default intensity. -i 24 on windows.!  
When I build his git unmodified, I get a 30% speed loss. Since alexis have broken the windows project file, you need to copy and rename the vcash.cu to vanilla.cu into the algo256 folder of 1.7.5. And you need to rename vcash to vanilla inside the file, and correct the free method name. (This guy really don't want us to use windows)

Yes there is a lot of shit like that in my code that makes my latest decred 14 round blake-256 300% faster than the first blake 256 kernal implementation in ccminer. (from nov 2014)
I just need to put some more shit to gain 400%, and maybe you can help me. You seem to be full of shit.

My Decred #4 bin was stolen and put on this thread. Just like your bins. Open up your sourcode!

I just did with decred #4

There is a difference. I keep my kernals private outside the internet. The percentage fee miner can be downloaded by everybody.

The lost password to the bitcointalk account is ok, but your email as well? What happened?

because the powers that be on bct havent got back to me AT ALL ...

the only one that did was an admin that couldnt reset my account ... all he could do was allow a little more access to this crap newbie account ... theymos has a ton of emails before mine to deal with - but have not heard anything from him - and badbear aparently is awol with no indication as to when he will be back ... ive emailed a few others - but like i said - NOT ONE of the mods ( only one admin ) has got back to me ... and its been a couple of weeks now since my screw up of the account password and question ...

i cant believe the laid back method that the mods of this forum have towards accounts that are liek mine - hero turning legendary ...

aparently even the admin who helped me told me 'not to hold my breath' ...

anyone here will be most welcome to open a thread on the granite forum - but ill be changing that to the new umbrella name and open a new thread here also - just get back control over my own threads again ( granite and dbm - and now ozziecoin as well since ive taken over ) ...

i completely understand the mistrust of newbie accounts - but pm me a question that only YOU and I know - and ill answer it ... especially regarding all this mess with the kernels and such ...

ill be fixing the email issue tonight i hope ( if my isp can do so ) and will be back up and running ...

i can prove a great deal with regards to how genuine i am as being chrysophylax ...

want to see how serious ( and accurate i am with any of the questions ) - chat me real time in skype ( chrysophylax69 ) ...

try me ...

#crysx

A lot of butt hurt newbie fucktwats in town.

Just a friendly reminder, SP is selling speed increases, they sometimes are broken, but he does fix them and he goes out of his way to try and improve them.

What sort of competition is there? Who is offering improvements besides him?

Pretty much anything is better then nothing and that's what other developers are offering... But other developers feel the need to come in here and tell him how it's done when they see him earning money.

You don't want someone sourcing your code and making a profit off of it? Don't open source it. It's that simple.


This is just another case of jealousy and someone else wants a piece of the pie... and more so, epeen recognition.

Speaking of recognition. I was the one that pointed out the Decred client/pool hash bug and poked around till SP fixed it. ...but you know, not like that matters.

Vcash is 8 rounds of blake and based on the same code as the decred mod(14 rounds of blake). (blake-256 sp-mod) (tpruvot Vanilla fork used my optimized blake kernal without giving me the credits.).

Run my opensource 1.7.79 compiled with cuda 7.5.

ccminer -a blakecoin --benchmark.

First of all, the above header is from vcash. Not Decred.
The fork was from tpruvot not you.

Second, Of course the decred source will be released.
I was going to do it yesterday. One day after the fee release, but you showed up your source and i needed some time to test it on my machine. I have only one pc, Cant do much on it.
I just need some time to merge the decred algo on the public repo, never said i'm going to keep it for my self.

Here you admit that you based your work on my optimized blake-256 maxwell kernal.
And now you have made a Fee version out of my hard work that everybody can download.
CCminer is under GPL licence and your fee miner is violating GPL. Please show us your sourcecode.

And please change your title in the opensource sourcecode.

It was not based on the implementation from 2014.

* Based upon Blake-256 implementation of Tanguy Pruvot - Nov. 2014

It was based on my 100% faster code from 2015 in the sp-modded maxwell kernals. The  Tanguy Pruvot - Nov. 2014 release did 225MHASH. My version from may 2015 did 460MHASH. (blake-256 14 rounds)
Basicly you steal my work without giving me any credit and add a few percent, and then come to my thread and spread shit... Then you release a fee miner that everybody can download, and violate GPL.
Morron.

Here is the credits on your opensource kernal. Where is my name?:

https://github.com/alexis78/ccminer/blob/a1db906a9b43494dd710af2d33b695ba148539dd/Algo256/blake256_8round.cu

1. Alexis how i can run your miner with decred algo ? i want run tests and compare it with SP version
2. I compile your lastest GIT version in CUDA 7.5 (right?) cuda 6.5 not compiled and shows error.
You have the final version with all additions and changes to the test ?
I have a Maxwell card

I never said you didn't fork and improved tpruvot's version.
I just said that you forked mine (which was a forked, already 100% faster version of blake, never denied your past work, and which i did manage to further improve by 12 (750ti) and 15-16(9xx series) %).
And you came again claiming another 30-15-8% faster code without anyone confirming it.
And two people who tested it and replied back said that the code you sell as 30-15-8% faster reports the same speeds (if not less) from the open sourced one.
And yet, here we are still, with you claiming an 8% faster than mine, code, and me asking for some real proof.

Am i asking something irrational?