CCminer(SP-MOD) Modded GPU kernels.

[sp_]

If I recall correctly, cryptonight uses a 2MB buffer, which means you need MORE than 2MB of cache to fit it entirely and avoid using main gpu ram. Moreover, the miners must work on many cards so they probably make use of gpu ram anyway. Cache helps, but there will be a lot of misses. Thus, cache size and speed matters but so does global memory latency.

If your rig has 6 gpu's they have a total of 12 MB cache. instead of computing the complete hash each of the gpu's, it might be possible to spread the work.

[pallas]

If I recall correctly, cryptonight uses a 2MB buffer, which means you need MORE than 2MB of cache to fit it entirely and avoid using main gpu ram. Moreover, the miners must work on many cards so they probably make use of gpu ram anyway. Cache helps, but there will be a lot of misses. Thus, cache size and speed matters but so does global memory latency.

If your rig has 6 gpu's they have a total of 12 MB cache. instead of computing the complete hash each of the gpu's, it might be possible to spread the work.

not sure the internal state can be split... maybe someone with better knowledge on cryptonight, like wolf0, can comment.

[sp_]

In spreadcoin it would be possible if the buffer was a bit smaller.
)
The bottleneck in the spreadcoin hash is a 200 000 table sha256 runthrough. (800kb)

The 980 has 96kb shared mem (level 1 cache)

So you would need 9 980 cards in a rig.

card 1 hash the first 96KB, passes the result to card2 wich calculates the next 96kb block, then card3 etc..

Each card will store the partial result in their sharedmem/level1 cache.

[Grim]

So why is the 750ti performing best (hash per watt) with a rather low gddr5 bandwith? (cryptonight)
The only thing it offers is a huge level 2 cache (2mb).

[chrysophylax]

sp could you ask them how much electricity cost?  plz 

i too would be interested in shifting the farm there ... i mean - interested in electricity costs ... cough cough ...
...
#crysx

Today it was 38 degrees celcius outside. This is not a good place to mine.

here in adelaide it gets to 45C ...

so its not an issue mate - just got to know how to manipulate the airflow ...

#crysx

[tbearhere]

sp  how much does electricity cost there?

[Amph]

i don't remember now but, can this miner mine cryptonite?

[chrysophylax]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

[coinits]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

[chrysophylax]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

i cant exactly remember - but it was ok ...

from memory - tsiv's miner ( which sp optimized - and correct me if im wrong sp ) was close but didnt hash as much as sp's optimized miner ...

i need sleep now - so i could be way off the mark - but i can test it again for you tomorrow to get an exact hashrate if you like ...

#crysx

[myagui]

cryptonite =! cryptonight 

[coinits]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

i cant exactly remember - but it was ok ...

from memory - tsiv's miner ( which sp optimized - and correct me if im wrong sp ) was close but didnt hash as much as sp's optimized miner ...

i need sleep now - so i could be way off the mark - but i can test it again for you tomorrow to get an exact hashrate if you like ...

#crysx

It is a cost-benefit thing. If there is enough extra hashpower to cover the cost relatively shortly, through mining XMR, then it would be worth the cost. Depends on the cost though.

I would be more happy with miners that took a percentage fee from the hashpower vs a one time charge for it. Sort of a willing botnet where the dev keeps getting paid to infinity and makes it worthwhile to keep developing and improving.

[pallas]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

i cant exactly remember - but it was ok ...

from memory - tsiv's miner ( which sp optimized - and correct me if im wrong sp ) was close but didnt hash as much as sp's optimized miner ...

i need sleep now - so i could be way off the mark - but i can test it again for you tomorrow to get an exact hashrate if you like ...

#crysx

It is a cost-benefit thing. If there is enough extra hashpower to cover the cost relatively shortly, through mining XMR, then it would be worth the cost. Depends on the cost though.

I would be more happy with miners that took a percentage fee from the hashpower vs a one time charge for it. Sort of a willing botnet where the dev keeps getting paid to infinity and makes it worthwhile to keep developing and improving.

the problem with miner and a percentage fee is that, to be honest, a developer should code his own miner (not use a modified opensource one). It also moves the risk from the miner to the developer.

[smolen]

the problem with miner and a percentage fee is that, to be honest, a developer should code his own miner (not use a modified opensource one). It also moves the risk from the miner to the developer.

As an author of such miner I'd say that getting rig of GPL'ed code is hard, but well worth the time. And the main risk is to forget to look at new coins mined with it, still have plenty of useless FZ sitting somewhere in blockchain under my private key

[ol92]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

i cant exactly remember - but it was ok ...

from memory - tsiv's miner ( which sp optimized - and correct me if im wrong sp ) was close but didnt hash as much as sp's optimized miner ...

i need sleep now - so i could be way off the mark - but i can test it again for you tomorrow to get an exact hashrate if you like ...

#crysx

It is a cost-benefit thing. If there is enough extra hashpower to cover the cost relatively shortly, through mining XMR, then it would be worth the cost. Depends on the cost though.

I would be more happy with miners that took a percentage fee from the hashpower vs a one time charge for it. Sort of a willing botnet where the dev keeps getting paid to infinity and makes it worthwhile to keep developing and improving.

+1 : I am a small miner : I paid 0.1 btc for the optimized spreadcoin ccminer, but with two cards only, there is no way to recover the cost for cryptonight. I paid once because I am using the work of sp_ (and others) for others algo too in the open source ccminer.
The percentage thing like Claymore is far better for small miner. For the dev, I think it could be more profitable if the coin is good and keep value (xmr and vanillacoin should !).

[chrysophylax]

i don't remember now but, can this miner mine cryptonite?

no mate ... but sp has his private miner that is optimized for that algo ...

#crysx

What kind of hash for a GTX 750 ti? Will 660 ti work?

i cant exactly remember - but it was ok ...

from memory - tsiv's miner ( which sp optimized - and correct me if im wrong sp ) was close but didnt hash as much as sp's optimized miner ...

i need sleep now - so i could be way off the mark - but i can test it again for you tomorrow to get an exact hashrate if you like ...

#crysx

It is a cost-benefit thing. If there is enough extra hashpower to cover the cost relatively shortly, through mining XMR, then it would be worth the cost. Depends on the cost though.

I would be more happy with miners that took a percentage fee from the hashpower vs a one time charge for it. Sort of a willing botnet where the dev keeps getting paid to infinity and makes it worthwhile to keep developing and improving.

+1 : I am a small miner : I paid 0.1 btc for the optimized spreadcoin ccminer, but with two cards only, there is no way to recover the cost for cryptonight. I paid once because I am using the work of sp_ (and others) for others algo too in the open source ccminer.
The percentage thing like Claymore is far better for small miner. For the dev, I think it could be more profitable if the coin is good and keep value (xmr and vanillacoin should !).

mining xmr was a fruitless exercise for us ...

the farm was powering away - only to bring in a small amount of xmr due to the ridiculous difficulty ... the farm needs to be really LARGE to make any decent coinage from xmr ...

im guessing there is still a great deal of optimization left in the cryptonight algo before we can say its a 'worthy' algo for the farm ...

believe it or not - even at these prices - spr is a hell of a lot more profitable with regards to the amount of coins and its conversion to btc ... im never comparing it to fiat ( which most do when talking about 'profitability' ) ...

we still lean towards x11 algo as far as mining and conversion ( profitability ) is concerned ... and the opensource ccminer that sp optimizes on a regular level is by far the better one for that ...

#crysx

[sp_]

The best rates I could get in version 1 of my private mod of the XMR miner was 305KHASH on the 750ti and 475KHASH on the 970. The 960 was down to 280KHASH (1mb level2 cache?)

Tuning the clocks was nessesarry. You need to find the sweetspot Core/mem speed.

That's why I will implement a -gpuclock / mem parameter in ccminer just like in sgminer.

But my vacation is still not finished. A darkcoin masternode is worth more than a yearly university graduate thai salary here.. Stil got some satoshis to spend.

he-he

Happy mining

[5w00p]

Subscribed.

I might be interested in the optimized Cryptonight (Monero) miner, if it can substanatially improve the speeds of my 750Ti and 970.

[scryptr]

Tomb Sweeping, Easter, the Mystical Orient, etc...

This is a holiday weekend for many reasons, it seems, depending on your people.  I hope that sp_ returns safely from his vacation, soon.  His well-followed (and referenced) thread has been wandering the desert while he has been scaling the mystic oriental heights in search of the best Bitcoin beer and perhaps racking up an expense tab on a crypto-mogul.

I've been reading about "Coolbits" and Linux overclocking.  From a post on another blog, I've found and re-written an xorg.conf file.  I haven't tried it yet.  I am going to place it in /usr/share/X11/xorg.conf.d/ on my Ubuntu 14.04 Linux box, with the filename "50-nvidia.conf":

Section "Device"
 Identifier "Device0"
 Driver "nvidia"
 VendorName "NVIDIA Corporation"
 BoardName "GeForce GTX 750 Ti"
 Option "Coolbits" "28"
 Option "TripleBuffer" "True"
EndSection

The page on which the original configuration file was found is at "https://devtalk.nvidia.com/default/topic/820497/-solved-coolbits-without-xorg-conf-/", and the post was from user "manorba", posted 23/MAR/15.  I believe that his system has a single 650ti.

The "Coolbits" option is set at "28"; this number being derived from "4" + "8" + "16".  If "Coolbits" is set at "4", fan control is enabled, if set at "8", clock adjustments are enabled, if set at "16", power adjustments are enabled.  Nvidia Linux driver 346 + drivers are required for all three of these user-controls to be enabled.  I have read that power adjustments are command-line interface (CLI) only.  Also, a setting of "12" enables only fan+clock controls.

I am posting this here so that performance-minded Linux users can perhaps come up with working nvidia configurations for multi-card mining boxes.  In a preliminary experiment, I was only able to enable the nvidia-settings GUI fan and clock controls for my device0 card in my 6x750ti rig.

I hope, because sp_ is performance minded, that this is not too far off-topic.  On my Windows box, I discovered that EVGA PrecisionX 16, their overclocking program, is limited to a maximum of 4 cards.  I am looking for a better program, one that will manage 6 cards, or I may revert to Linux.  And, for those that overclocked in Linux years ago, a "Coolbits" setting of "1" only works with older Nvidia cards and drivers.

I hope that all enjoy a holiday weekend, wherever they may be in the world!       --scryptr

[sp_]

I have studied the opencl code of sgminer, and I see many possible speed increases. Quark/qubit. I have picked up a couple of AMD cards cheap(280x,290 and 290x), and will try to
rewrite and optimize quark and qubit. I belive 10-15 MHASH quark per card is possible on the 280x (up from todays 5MHASH)

300% faster.