Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 599. (Read 2347641 times)

Yeah color me unimpressed as you're not a coder. You aren't the only that's mined Cryptonote over the years. We all know the hurdles and are talking about them. There is no reason to hate on trying to improve things unless you're trying to keep things lucrative for CPU mining.

There are plenty of algos that haven't been looked at or thought to be maxed out... or that they're CPU algos and there is no way GPUs could actually do anything meaningful. Lyra is a example of one that basically had its performance doubled by Nanashi and if you mentioned that before hand to any other developers they would've said there isn't anyway or it would've already been done by now.

guys .... cryptonight is CPU algo .......  cpu are more efficient  at it than gpu. i used to mine monero back in the days with a full cluster of servers.... 84 cpus brought me around 14 khashs back in the day... was lucky it pumped at 1 cent  around august of 2014....

sold all my coins ... got GOOD profit... then it was burst time... ahh another coin at at beginning servers were so nice.... but now again monero is profitable only on new cpus .... if coder could give profitable hasrates on cryptonote he would have done already... better to program a closed source software with fee than  keep it for yourself and buy hardware... unless of course you are not already rich to take over on the hashrate.... r9 290 on cryptonight gives 600 h /s  no way it will get you to 1200 .... the only card taht can do such is 295x and 7990  but they are dual gpu and smal ovens.... .... and at current difficulty and price it is not profitable.... r9 290 will consume at 700 h /s  250w , 0.34 XMR in day .... around 1.7 USD .... so forget abt it... same will be nvidia... i suppose with an nvidia we can get around 800 h/s with an optimized miner ( 250 / 260h tsiv miner that was optimized for gtx 750 to... same hash rate was anyway on gtx 560 .... but of course 560 was a small fournance ... 750 was kind on the wattage)

300 watt is seriously bad because for a mere rig you are looking at 2kw, which is crazy, require huge cooling solution if your have a farm of those...

for 390 to do 38mhs eth you will need core clock > 1250, optimized timings at 1500 mem and increased voltage. It will consume > 300 watt
I know only one card capable to do so - Nitro. And they all die very fast ))

Yeah, I've been watching, it would have to double again. Market cap doesn't matter, volume matters for mining. That's liquidity and how well it tolerates sells/buys.

Mining Monero with a 970 will get you .00107 BTC AT 800H/s (which I don't think is even right, it's just what I have put into a calculator). That's $.63. Pretty certain it's closer to 450H/s if I remember right.

So double and you get Monero to 800H/s on a 970... Still not likely to happen. Someone needs to go Nanashi on things and figure out a way of making it work better for GPUs... or it needs to be dual mined, somehow.

Monero has already doubled in just 4 days.

The Mcap is 70MUSD (8th largest altcoin after Bitcoin)

Yeah, so it would need to be a dual miner to be worth while. Even if Monero doubles again, it would barely be on par with Ethereum.

Wonder if it's purely bandwidth related or a latency thing like with Ethereum. For instance GDDR5X on the 1080 has generally been a downside to it.

Either way, definitely need some greener pastures right now. ROI is going into the stratosphere. Even Ethereum is in decline due to the massive China farms building on it right now.

yes the 390 can do 38 but at what wattage? 200? a 1070 can do 31.5 at 90w, no comaprison

i can fill a rig full of 1070 with just a 750w psu while you need expensive psu, more cooling, more electricity cost etc...

With optimal settings the The hawaii (390) can mine etherum@38MHASH. The 1070 can at best do around 32MHASH.
512bit / vs 256bit memory interface. Cryptonight is limited by the speed of the memory. I guess 800H/s could be possible with optimal BIOS/CODE/OC. on the 1070
But 600-700 is more realistic.

I think we will see blockchains that can give the miners big profits in the years to come. The Big banks are investing in blockchain technology now...

https://www.bloomberg.com/gadfly/articles/2016-08-24/blockchain-is-banks-secret-sauce

Yup, Cryptonote is a untapped market... if GPUs can finally overtake the botnets in any meaningful manner. I've asked a few different times for this, but the devs I've talked to said it was too hard to say double performance for GPUs. It currently doesn't even match Claymore's miner and hasn't for years.

a R9 290 was able to obtain 840KH/s in early 2014, now it should be up to 1200. With 3650KH/s you've got the same profitability of ETH+SIA ...

Do not bother

Cryptonight is limited to random memory access just like etherum. If you look at etherum a 970 can do 22MHASH and the gtx 1070 around 30.

So perhaps 700 is possible with some recoding and tuning.

even 500 is bad because i remember that a 970 can do 480, a 1070 should do almost 1000

My private is doing around 500 on the 970. If the price double again, I might take a look.

can you not released just the update for spreadcoin to donators at least? i was a donators for that algo


cryptonote algo is pretty bad on a 1070 doing 350khs only, maybe you can look at it

Monero is up over 50% today.

http://coinmarketcap.com/currencies/monero/

Unless it can be dual mined, Lbry is pretty much dead. Probably not going to see much action till after the reward block hits 500... Who knows what will happen then. Coins that have been on a downhill decline for over a month usually don't come back.


Yup, which is why I was wondering why you were trying to improve Spread... You throw a couple 970s on there and the difficulty explodes.

SPR has "decentralized development" to one person since other developers left. It has been months and months of "stay tune" whenever someone ask for update on progress and road map. The decentralized developer has his hands on lots of ad hoc things that interest him. There is no cut off to deliver something.

Not so much. The spreadx11 sourcecode is based on my x11 opensource from 1.5 years ago. Some of the kernals have been optimized in my sp-mod private (All nicehash algos optimized), so I guess a 5% increase from #9 should be possible without alot of work.. Just merge the new kernals into the old code, but then no opensource anymore and no linux support. And I might ask for more donation for the pascal-mod or a 2% fee.

24hours volume of 0.21 BTC is a bit low. Spreadcoin is not as good as it used to be..