Author

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 593. (Read 2347641 times)

sr. member
Activity: 420
Merit: 250
As someone with 1 GPU, this is not a worth it deal. I can just do and dual mine ether and decred.

If you made a donation free miner that did 5 or even 6 minutes of mining for you instead of 2, it would be worth while for owners of only 1-2 GPU's to use, where as all the farm owners can go and pay 0.1 btc for the reduced miner fee. 6 minutes an hour is 10% of all mining from a user. Assuming this hashes 15% higher it remains profitable for the miner.



newbie
Activity: 39
Merit: 0
Sp_ please ditch the 0.1 fee, we subscribed a few times for that amount for different miners but you always come up with something else we should subscribe to.

I am curently testing the donation code in the LBRY sp-mod #1.

I plan to release Optimized Pascal lyra2v2,skein,pentablake,x17 etc.. as well. free for the the donators in the past(but with a donation %). The sp-mod ultimate.

You are late to the party with LBRY. And what kind of boosts are we going to see? One percent? Two percent?
legendary
Activity: 2716
Merit: 1094
Black Belt Developer
20% of miners hold 80% of hashing power;
80% of developer market is in 20% of miners;
The rest 80% of miners are going to whine on forums when treated "unfairly" Grin

Looks a lot like a Pareto distribution ;-)
legendary
Activity: 1512
Merit: 1000
quarkchain.io
Yeah , you are right, I wish to kindly suggest you , to consider lowering that fee...
I have another model of 970s , will test it on them too...

How about keeping the fee and add 12% on the pascal 1070 and 1060 cards as well. Smiley Don't think about the fee. Think about the increased profit, Increased speed and Increased support in future releases..

Sounds quite good , yes , waiting to see it as well Smiley
Edit: I don't like cheap cards such as 1060. My opinion is to focus on the top cards 1080 and 1070....
member
Activity: 72
Merit: 10
20% of miners hold 80% of hashing power;
80% of developer market is in 20% of miners;
The rest 80% of miners are going to whine on forums when treated "unfairly" Grin

Anyway it would be nice to have a subscription miner with lower fee and a free miner with higher fee.
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Yeah , you are right, I wish to kindly suggest you , to consider lowering that fee...
I have another model of 970s , will test it on them too...

How about keeping the fee and add 12% on the pascal 1070 and 1060 cards as well. Smiley Don't think about the fee. Think about the increased profit, Increased speed and Increased support in future releases..
legendary
Activity: 1512
Merit: 1000
quarkchain.io
My 980Tis  Palit Super Jetstream , give me some improvement with the new miner 255MH on 1467MHz clock , compared with 228MH with the public miner. But the integrated fee will cover that improvement , I think..
I'm waiting to the next version...

12% faster, 3.3% fee...

Yeah , you are right, I wish to kindly suggest you , to consider lowering that fee...
I have another model of 970s , will test it on them too...
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
My 980Tis  Palit Super Jetstream , give me some improvement with the new miner 255MH on 1467MHz clock , compared with 228MH with the public miner. But the integrated fee will cover that improvement , I think..
I'm waiting to the next version...

12% faster, 3.3% fee...
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Tested many combinations , I didn't see that "vcap reason" My GPU-Z shows PerfCap reason , end the state is Pwr.
I could gain 160 MH max , with power limit 106% (the max you can tweek in AB).

I will look into this in release #2.
legendary
Activity: 1512
Merit: 1000
quarkchain.io
My 980Tis  Palit Super Jetstream , give me some improvement with the new miner 255MH on 1467MHz clock , compared with 228MH with the public miner. But the integrated fee will cover that improvement , I think..
I'm waiting to the next version...
legendary
Activity: 1764
Merit: 1024
If 1070 will make 340, I am ready to pay 0.1 BTC + 1 minute or 2 minutes only (not  0.1 BTC)

Claymore will make dual eth+lbry and who you will pay the price?

Do not be greedy.

Algos can't just be turned into a dual miner and they're automatically better. Some algos use more memory then GPU resources and vice versa, that's why they work out in a dual miner setup. If they have to fight for resources, they aren't going to be better. Fucking ethbabies.

"It cost 0.1BTC AND it comes with 2minute donation mining per hour."

isn't that too much? shouldn't be one or the other?

0.1 + 3.3% fixed (with 1% of the pool and more with the exchange), means that the final amount it's more or less the half of what advertized, and frankly speaking with the added 0.1 it's too much for my taste. The one or the other, not both.

Keep in mind that the frequency of a 1070, as you know, it's related to the boost clock (1750/1800, even 2k in some cases.. Stock clocks)

Pool fee being included with miner fee? Oh yeah? I hear developers are also responsible for transaction fees, exchange fees, and BTC withdrawl fees. Silly goose.

0.1 with fee is insane for the 1070 owner, unless you can pull off a crazy 20% or something, and i really doubt you can it's not optimizable anymore, pallas wolfo and epsylon all agree on this...

Because 20% is crazier then 15%. "I know nothing about coding, but I'm going to flat out say you can't improve this algo." You're dumb.

sp, i respect your work and i´ll gladly consider buying it when you optimize the 1070 part.

but, please, dont make your numbers on miner´s profit. Weight your work, not our income.

keep on the good work!


Translation: This guy owns one BIG ASS FARM and doesn't want it taxed. In other words, you'll make more then .1 BTC off him in the long run easily.

stock 1070 give 280 mhs on the latest miner. GPU 2050, mem -800

This is a heavy OC. Most people wont be stable at 2050. Looking closer to 1950, SPs numbers are quite realistic. Also it wouldn't be 280Mhs, it's around 270. I have a couple GPUs running at these clocks.

stock 1070 give 280 mhs on the latest miner. GPU 2050, mem -800

ccminer 1.8.1 gives more than 270 mhs on 1070 GPU 1960

While I see no reason to buy your miner.





Yeah, that's a 2080mhz OC, maybe even higher... Probably also crashed after he took the picture.

You realize people actually mine Lbry so they know rates like this aren't normal, right?



So all this shit aside, good SP is finally trying to make a long term solution here... There is literally no one else doing this except Claymore. Funny all the people bitching about fees. I do agree though that dev fee and flat fee combined are kind of silly though. The whole reason for a miner fee is for long term support and so you can get all the smaller miners in there as well.

A lot of people are now running 1070s. You probably wont see that many people purchasing this till there is some headway made with those units.


What was also mentioned about dual mining profit is true. I really don't think many Nvidia miners have tried out Claymores dual miner yet. Before the recent Lbry pump, 1070 and 970s were making more money on ETH+SIA. It's been that way for about a week.
legendary
Activity: 1512
Merit: 1000
quarkchain.io
With 2x 6pins you don't have enough power. Open gpu-z and check the vcap reason. if it shows VREL, try to reduce memclock the lowest you can get, and increase the intensity & core clock. The 1060/1070 kernals are not ready yet. Waiting for the cards to arrive.

Tested many combinations , I didn't see that "vcap reason" My GPU-Z shows PerfCap reason , end the state is Pwr.
I could gain 160 MH max , with power limit 106% (the max you can tweek in AB).
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
2x6 pins deliver 150 watt. so you are drawing 40 watt per card from the motherboard. with 6 cards 240watt from the motherboard. To much load will increase the heat and reduce the speed. Sometimes cause cards to trottle, and motherboards to crash or melt....
..but you have 2* Molex powering the PCI-E bus on the BTC motherboard.

You need to make sure that each molex come from a clean plug(nothing else is connected to the cable but the motherboard) . And that your PSU is single railed. If not you need to calculate the load on each rail. I normally only use one of the Molex on the btc boards. If you use one, you should plug it into the molex that is closest to the pci slots.
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Sp_ please ditch the 0.1 fee, we subscribed a few times for that amount for different miners but you always come up with something else we should subscribe to.

I am curently testing the donation code in the LBRY sp-mod #1.

I plan to release Optimized Pascal lyra2v2,skein,pentablake,x17 etc.. as well. free for the the donators in the past(but with a donation %). The sp-mod ultimate.
legendary
Activity: 1512
Merit: 1000
quarkchain.io
2x6 pins deliver 150 watt. so you are drawing 40 watt per card from the motherboard. with 6 cards 240watt from the motherboard. To much load will increase the heat and reduce the speed. Sometimes cause cards to trottle, and motherboards to crash or melt....

..but you have 2* Molex powering the PCI-E bus on the BTC motherboard.
legendary
Activity: 2002
Merit: 1051
ICO? Not even once.
Even ETH alone is almost 2 times more profitable than LBRY on a 970 given the current difficulty and public miners.

Not to piss in anyone's cereal I'm only saying this to show that an upfront price plus a fee won't attract many miners who do basic math (but it's mostly the upfront price).

Just checked..
Not twice - almost same +\- depends from pools lags and exchange play

Yes, discard my previous post (partly).

I only ever look at current figures of profitability and not averages. LBRY jump a lot up and down.

Anyway, I barely spent any time with mining lately but still, Eth's profitability is pretty poor compared to other alternatives and LBRY - even if it might be slightly better sometimes it probably won't be for long.


Sp_ please ditch the 0.1 fee, we subscribed a few times for that amount for different miners but you always come up with something else we should subscribe to.

Don't get me wrong, I'm not saying that you don't deserve it but it's just not a transparent and reliable system. What if you never make a newer version or decide to opensource it tomorrow? Or what if it doesn't work for us as advertised for whatever reason? And for most miner 0.1 is either too much or too little.

With a fee based version none of that's an issue. (Yeah, I know it introduces a bunch of other issues but a fee based miner is more reliable and user friendly.)
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
2x6 pins deliver 150 watt. so you are drawing 40 watt per card from the motherboard. with 6 cards 240watt from the motherboard. To much load will increase the heat and reduce the speed. Sometimes cause cards to trottle, and motherboards crash or melt....
member
Activity: 92
Merit: 10
With 2x 6pins you don't have enough power. Open gpu-z and check the vcap reason. if it shows VREL, try to reduce memclock the lowest you can get, and increase the intensity & core clock. The 1060/1070 kernals are not ready yet. Waiting for the cards to arrive.

my MSI 970 have 320dtp , but at LBRY they takes 68%, say 180-190W, so it's not a problem for 2x6pin cards, too
member
Activity: 92
Merit: 10
Even ETH alone is almost 2 times more profitable than LBRY on a 970 given the current difficulty and public miners.

Not to piss in anyone's cereal I'm only saying this to show that an upfront price plus a fee won't attract many miners who do basic math (but it's mostly the upfront price).

Just checked..
Not twice - almost same +\- depends from pools lags and exchange play
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
With 2x 6pins you don't have enough power. Open gpu-z and check the vcap reason. if it shows VREL, try to reduce memclock the lowest you can get, and increase the intensity & core clock. The 1060/1070 kernals are not ready yet. Waiting for the cards to arrive.
Jump to: