CCminer(SP-MOD) Modded GPU kernels.

[chrysophylax]

I added some boost in x11. Compile the latest and run with -i 19 -g 3 -a x11 and overvolt your card.

x13,x14,x15 will also run bether with multiple threads now. I reduced the constmem usage in the AES kernals. (shavite,echo,fugue,whirlpool)

done sp ... running now with those settings and the latest git pull ...

you can see the results on yaamp - still mining the donation link ...

quark seems to have a hard time staying mining - due to the stratum extranonce2 errors ...

we dont seem to have this issue with x11 ...

i need rest now - so ill check it all when i wake tomorrow ... but i find it weird how quark runs for a while fine - then e2 errors go on for a while before it settles again ...

this happens whether connected directly or through the proxy ... an issue with yaamp? ... or ccminer? ...

btw - i cant overvolt / undervolt any of these cards ... but they seem to be running well - for now ... ...

#crysx

[scryptr]

hi all ...

the donation stratum address is now active and can be mined directly ...

looks like yaamp had a major hiccup and dummy spit - with extranonce2 size errors - as can be seen by the graphs of sp's btc address on yaamp ... this issue looks like it affected all the miners - and was caused by yaamps stratum ... but it all seems to be going ok now ... we will see how it goes for the next few hours and check it again when i wake ...

so now we have sp donation - x11 and quark ...

just point miners to the following address and ports if you wish to donate in this way to sp - any username / any password ...

x11 - donate.granitecoinx.com:7999 ...
quark - donate.granitecoinx.com:7919 ...

remember these are still test environments and the address will more than likely change to - donate.granitecoin.com:portnumber - when it is setup permanently ...

just for examples sake - our mining settings for gigabyte 750ti oc lp cards in linux scripts or commandline are as follows ...

./ccminer -o stratum+tcp://donate.granitecoinx.com:7999/ -O chrysophylax.test:x -a x11 -i 20.6
./ccminer -o stratum+tcp://donate.granitecoinx.com:7919/ -O chrysophylax.test:x -a quark -r 3 -R 10 -i 23.7

./ccminer would be replaced with ccminer.exe to be used for windows batch files if you have the same cards ( probably with different parameters also ) ...

we hope this is ok so far ... the other algos will be setup tomorrow and the list of algo / port messaged here ...

#crysx

HAMPSTER ON THE WHEEL FOR YA, MATE!

I just hooked up to your donation proxy.  Great implementation!  I'll leave the critter there a while.  Any way we can view the action online?    

EDIT:  I just looked up the donation wallet on Yaamp.  All is good!    --scryptr

[bathrobehero]

:loop
ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-west01.miningrigrentals.com:3333 -u me -p x  -d 0,1,2
taskkill -t -f /im ccminer*
timeout -t 1
ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-central01.miningrigrentals.com:3333 -u me  -p x  -d 0,1,2
goto loop

Should prevent the crash.

Thanks bathrobehero. But it was crashing at the goto loop. And said at that point it couldn't find ccminer ..

Yes, I misunderstood the problem. I couldn't replicate it though.

[sp_]

I submitted a fix for the cpu usage in multiple threads.

x11 750ti settings:

-i 19 -g 3 -a x11

seems to run faster than default.

around 150Khash faster per 750ti.

Look at the power graph, only 60-90% tdp. With overclocking/voltage the hashrate gets much higher.

[Slava_K]

WhirpoolX not work whith -g.  Hash is slowing 15-20%. GTX960.

X11, X13 is faster without -g. X15 is slower.

[sp_]

WhirpoolX not work whith -g.  Hash is slowing 15-20%. GTX960.
X11, X13 is faster without -g. X15 is slower.

You need to combine the -g with the -i parameter. (for whirlpoolx and skein there is no gain)

And you need to download and build the latest version on github.

x11:

-i 19 -g 3

for quark:

-i 22 -g 5

[rednoW]

commit #753
Tried to run quark and x11 without any parameters. Huge cpu usage (totally grabs one cpu core), system unusable with high priority.
p.s. win 7 64 bit, single gtx750, athlon x2 3ghz

[sp_]

Skein is optimized

0 memory and full boost on the gpu. full load, and 98%TDP

100MHASH on the 50watt 750ti.

[Slava_K]

with -g 4 and -i 22 hash on my 3 960 is 25M (-g 5 stop with errors). With only -i 22 hash 29M!
Now i downloading new build. trying later today.

[sp_]

commit #753
Tried to run quark and x11 without any parameters. Huge cpu usage (totally grabs one cpu core), system unusable with high priority.
p.s. win 7 64 bit, single gtx750, athlon x2 3ghz

Thanks. I have fixed it now. Should be back to almost 0.

[flipclip]

Can the g parameter be used per device, example two cards set differently with -g 4,3 ?  (per if you are running a mix of cards)

[sp_]

I am building 48 now. so you can help me test.,

[sp_]

Can the g parameter be used per device, example two cards set differently with -g 4,3 ?  (per if you are running a mix of cards)

you have to do

-d 3,4 and then -g x

[Slava_K]

last commit - high cpu load (2 cores from 8 ) on single thread (without -g) on x11 and quark...

[sp_]

Ifixed the cpu load 10 minutes ago.

[Slava_K]

Sorry. Compiling just now.

[flipclip]

Can the g parameter be used per device, example two cards set differently with -g 4,3 ?  (per if you are running a mix of cards)

you have to do

-d 3,4 and then -g x

thanks

[sp_]

- Faster quark on compute 5.2 (gtx 980 + 800Khash)
- Faster skein
- Improved multiple threads perfomance (the -g parameter) on quark,x11,x13,x14,x15,x17 etc


1.5.48(sp-MOD) is available here: (30-april-2015)

https://github.com/sp-hash/ccminer/releases/tag/1.5.48

The sourcecode is available here:

https://github.com/sp-hash/ccminer

[go6ooo1212]

Which improvement we could expect on QUBIt with GTX970/750Ti

[sp_]

Which improvement we could expect on QUBIt with GTX970/750Ti

With -g 2 I get this. (2 750ti oclock)

(deleted)

Qubit is not fixed for multiple threads. with -g you get invalid hashes and lower rates on the pool