Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1082. (Read 2347601 times)

duplicate share dirty trick. Let's ban him

10 Mh/s on 750Ti?? That's really great!!
will you merge the mods, make a new repo or sell it?

Yes could be a thread lock issue... I try to fix this kind of problems since months...

Some other problems seems to be on the pool... Not easy to fix that for all the possible cases.

In my fork of ccminer I Submit stale shares. This was because the hashrate on some pools dropped alot if I didn't do this.
If you are on a slow connection, this might leed to a higher reject rate.

I will make an option for submit stale shares like we have in sgminer..

Agreed, not a major issue and likely a pool issue. Just noticed the pattern and thought
I'd mention it.

The cards are EVGAs, OCed at the factory but not by me. Thanks.

we get that here too ...

but i think its more a pool issue than miner ... i could very easily be mistaken ...

not a major issue though - especially if it stays on mining well ...

ive seen 'job not found' and other rejected share reasons with a few algos also ...

btw - nice hashes on the cards ...

#crysx

kool ...

i asked due to the chat about some new coins that will be coming out with blake2b algo - specifically looking at sia ...

its better to be ready for the coins than wait till they come out and scramble to try and get in before the major players do ... thats all im saying ...

so even if it a simple non-optimized implementation - its better than not having it ...

but if the -a blake option will process blake2b - then is all good ...

thought id ask - thats all ...

you guys are the experts with this ...

#crysx

I noticed the following error a couple of times while mining quark at westhash
with ccminer 1.5.50-SP, default parameters, seen on both Linux and Windows.
I don't know if it's the same reject reason everytime but it's always the first
share submitted after the stratum set nonce. They seem to be very rare.

[2015-05-20 18:04:02] GPU #0: GeForce GTX 970, 15904
[2015-05-20 18:04:02] accepted: 6820/6863 (99.37%), 27426 khash/s yay!!!
[2015-05-20 18:04:03] GPU #1: GeForce GTX 780 Ti, 11438
[2015-05-20 18:04:03] accepted: 6821/6864 (99.37%), 27426 khash/s yay!!!
[2015-05-20 18:04:04] Stratum set nonce 77ffffc2f2 with extranonce2 size=3
[2015-05-20 18:04:14] GPU #0: GeForce GTX 970, 15922
[2015-05-20 18:04:14] accepted: 6821/6865 (99.36%), 27425 khash/s booooo
[2015-05-20 18:04:14] reject reason: Share above target.
[2015-05-20 18:04:16] stratum.westhash.com:3345 quark block 1678360
[2015-05-20 18:04:16] GPU #1: GeForce GTX 780 Ti, 11475
[2015-05-20 18:04:16] GPU #0: GeForce GTX 970, 16062
[2015-05-20 18:04:17] GPU #1: GeForce GTX 780 Ti, 11450
[2015-05-20 18:04:18] accepted: 6822/6866 (99.36%), 27416 khash/s yay!!

I also get an initial retry message when connecting to  the same pool, most
of the time but not every time. It always connects after the timeout. It's odd
because there is no associated error message, just the retry message and the
30 second wait. Again seen on both Linux and Windows.

I don't know if they are related or are related to the CTRL-C crash but they all
seem to be recent regressions. Seems more likely they would be related to
tpruvot's changes that SP's.

Edit: I don't recall ever seeing the first error on yaamp. I khow the second error
did not occur on yaamp. The CTRL-C crash occurs every time anywhere.

Any thoughts?

Update: a rejected share does not occur every time there is a stratum set nonce.
I have also seen job not found following stratum set nonce

At last !!! )) enough for offtopic discussions ))) .@Sp At this time is there any update  )))

Just started a dedicated thread so we can stop hijacking SP's with this discussion:
https://bitcointalksearch.org/topic/how-to-choose-a-coin-to-mine-1066450

I have already optimized the blake implementation in Neoscrypt (from djm34's version). the total hashgain was around 7% on the 750ti. So the latest Sp-mod is around 7% faster than the optimized djm34 version when mining neoscrypt.

else.. on the other branch... multi pools config and pool failover/time switching is almost ready.

I can say it will be in my next release.. 1.6.4

sp ...

just out of curiosity - will you implement blake2b algo in ccminer? ...

ive been seeing a number of chats about the algo - which usually leads to new coins being created to make use of the algo ...

#crysx

Indeed. The codebase is pretty large with 30+ hashing algorithms. This meens that it's still possible to find small speedups everywhere.
It just takes alot of time... A code marathon....

If you look at the first sp-mod from october 2014  the x11 was doing 6,5MHASH on the gtx 980x (scroll to the first page).
In the latest Sp-mod (50+) an overclocked gtx980 will do 11-12 MHASH.

That's a 70% - 85% increase of hashrate in 7 months. (free and opensource)

There are other algos that have a higher increase.. Try blake... over 150% faster


The problem with modern gpu's is that they change so quickly that the software developers don't have time to make optimal code
before a new chip is comming.

tanx sam ...

very few devs will go out of their way to do these things - let alone release all the source code ( charged for or not ) ... sp is one of those that will ...

so the focus is simple ... give back a percentage ( in our case 7% ) of the mining on a regular basis - which is about one day out of seven for as long as the farm is around ...

if westhash DO whitelist our ip's - then the donation links will be permanent ... or at least as permanent as we are around ...

we want to support the devs that support the community who support the devs ...

so any ideas would be great - if we can implement them of course ...

maybe start a new thread for this? ...

#crysx

Its great to see that you are going out of your way to keep this just and support the dev. We can play capitalism with you for sure.
All the Best.

well - it just so happens that all the nvidia miners are offline due to maintenance at the moment ... only the amd miners are online at the moment ...

this is so we can do the full upgrade to the new frames we have just designed - and the upgrade to the new systems in the following months ( making the farm larger ) ...

after all that is done - a portion of that will be DEDICATED to mining for donations for our devs ... so its funny you mentioned that ... hehehe ...

all updates are on our thread ( in the sig on granitecoin ) - and will start a new thread very shortly ( if westhash are kind enough allow our ip ( hence our and everyones miners ) into the whitelist ) ...

contacting nvidia today ( when i have 15 mins ) with regards to the overclocking under linux also ... so with teh optimizations that sp is making - we should have a pretty unique system to mine with ...

really looking forward to it all being functional in the next few weeks ... am full time with this as of next week also - so should be interesting times ahead ...

#crysx

No problem, you can offer me hash and miners for my generosity

OMG - hahaha ...

two posts that just put a gaping hole in my face - and laughter coming out ...

im in a good mood now after being woken by heavy rain and cold ...

tanx guys ...

#crysx

hahaha ...

i sense a little Ice-T around the place ...

...

#crysx