Pages:
Author

Topic: SILENTARMY v5: Zcash miner, 115 sol/s on R9 Nano, 70 sol/s on GTX 1070 - page 13. (Read 209334 times)

sr. member
Activity: 588
Merit: 251
The amount of RAM needed isn't prohibitive - but the algorythm's "lots of memory access limiting the speed" issue might limit the gains an ASIC can manage to the point it's not cost-effective enough to bother.
 I'm still a bit amazed that not just one, but *3* different ASIC showed up pretty close together for X11 a while back - that was a MUCH smaller market by the time the first one arrived than any other ASIC had ever been built for.

FPGA Altera Intel Stratix 10 MX have HBM2 onboard with 1 TB/s
So it's not a big problem.

Sure it will not become a mass product like GPU, but ...
And the other question - cost.

The Stratix 10 MX is targeted at high-end switches and routers.  We're not talking about a $5 Lattice iCE...
legendary
Activity: 3248
Merit: 1070
Well...I'm already hashing 1kS with 5 cards(4xRX480+1 R9 390) heavily undervolted to save power. Power consumption (wall) is only 880W...This is with Claymore and early era of zcash mining...Until limits are reached noone knows how much this hardware can push. As I see these are MXM cards in custom PCB. It's not pure hardware dedicated ASIC. If you have $3300 consider building NV 1070 RIG. 7 Cards cost ~$3k + $300 for mbd,mem,cpu and additionally some decent PSU...total should not exceed $3500 and you can for sure hash with more than  1kS(+50% or even more) with this piece of hardware.

nvidia hash at better efficiency, i can hash at 1200 sol with only 680w, but it's more expensive

it should be better to go with the 1060 if the ratio with the 1070 is the same
This is just preliminary calculations Smiley
Until final optimizations are done, you shouln't make conclusions Smiley

and nvidia is not even done yet with coding it can be improved in speed, so even better ration is expected, i doubt amd can catch up with the efficiency
Currently, some AMD cards can hash at the same efficiency as this, at a fraction of the cost. Efficiency is only part of the game, if you are 15% more efficient, but cost 2-3x as much, you don't have an advantage.  Sure you could save the money in power in like 8 years (assuming all your cards last that long), but by then you are obsolete, and missed a huge opportunity to be mining twice as fast for 8 years, so still a loss. Also, what are specs that will mine  Nvidia at this rate, consuming that much at the wall?
Nvida is just too expensive you can now buy a 480 4gb for $179 and prices are just going lower
with vega comming in december , i have some 1070/ got them to play around with i love them but
i only get 120 sols or so and the cost me $400 each i can now get two 480s
for that that hash at 195 sols each at 130 watt

a 1070 can do 220 sol but it's locked on nicehash, still around the same profit as mining zcash

if you want something cheap, the 1060 3gb can do 140 sol and it's $200, wattage should be around 60w(not sure i'm asking around about this as i don't have one)
legendary
Activity: 2408
Merit: 1102
Leading Crypto Sports Betting & Casino Platform
Well...I'm already hashing 1kS with 5 cards(4xRX480+1 R9 390) heavily undervolted to save power. Power consumption (wall) is only 880W...This is with Claymore and early era of zcash mining...Until limits are reached noone knows how much this hardware can push. As I see these are MXM cards in custom PCB. It's not pure hardware dedicated ASIC. If you have $3300 consider building NV 1070 RIG. 7 Cards cost ~$3k + $300 for mbd,mem,cpu and additionally some decent PSU...total should not exceed $3500 and you can for sure hash with more than  1kS(+50% or even more) with this piece of hardware.

nvidia hash at better efficiency, i can hash at 1200 sol with only 680w, but it's more expensive

it should be better to go with the 1060 if the ratio with the 1070 is the same
This is just preliminary calculations Smiley
Until final optimizations are done, you shouln't make conclusions Smiley

and nvidia is not even done yet with coding it can be improved in speed, so even better ration is expected, i doubt amd can catch up with the efficiency
Currently, some AMD cards can hash at the same efficiency as this, at a fraction of the cost. Efficiency is only part of the game, if you are 15% more efficient, but cost 2-3x as much, you don't have an advantage.  Sure you could save the money in power in like 8 years (assuming all your cards last that long), but by then you are obsolete, and missed a huge opportunity to be mining twice as fast for 8 years, so still a loss. Also, what are specs that will mine  Nvidia at this rate, consuming that much at the wall?
Nvida is just too expensive you can now buy a 480 4gb for $179 and prices are just going lower
with vega comming in december , i have some 1070/ got them to play around with i love them but
i only get 120 sols or so and the cost me $400 each i can now get two 480s
for that that hash at 195 sols each at 130 watt
legendary
Activity: 3248
Merit: 1070
Well...I'm already hashing 1kS with 5 cards(4xRX480+1 R9 390) heavily undervolted to save power. Power consumption (wall) is only 880W...This is with Claymore and early era of zcash mining...Until limits are reached noone knows how much this hardware can push. As I see these are MXM cards in custom PCB. It's not pure hardware dedicated ASIC. If you have $3300 consider building NV 1070 RIG. 7 Cards cost ~$3k + $300 for mbd,mem,cpu and additionally some decent PSU...total should not exceed $3500 and you can for sure hash with more than  1kS(+50% or even more) with this piece of hardware.

nvidia hash at better efficiency, i can hash at 1200 sol with only 680w, but it's more expensive

it should be better to go with the 1060 if the ratio with the 1070 is the same
This is just preliminary calculations Smiley
Until final optimizations are done, you shouln't make conclusions Smiley

and nvidia is not even done yet with coding it can be improved in speed, so even better ration is expected, i doubt amd can catch up with the efficiency
Currently, some AMD cards can hash at the same efficiency as this, at a fraction of the cost. Efficiency is only part of the game, if you are 15% more efficient, but cost 2-3x as much, you don't have an advantage.  Sure you could save the money in power in like 8 years (assuming all your cards last that long), but by then you are obsolete, and missed a huge opportunity to be mining twice as fast for 8 years, so still a loss. Also, what are specs that will mine  Nvidia at this rate, consuming that much at the wall?

you have forgot the resale value which is huge for nvidia, a 1070 can be sold again at 80% of its retail price, so i'm not actually paying more in the end

also there is the 1060 which is cheaper and has a 140 sol/60watt ratio which is better than amd and price is there same as a 480 4gb
sr. member
Activity: 449
Merit: 251
Well...I'm already hashing 1kS with 5 cards(4xRX480+1 R9 390) heavily undervolted to save power. Power consumption (wall) is only 880W...This is with Claymore and early era of zcash mining...Until limits are reached noone knows how much this hardware can push. As I see these are MXM cards in custom PCB. It's not pure hardware dedicated ASIC. If you have $3300 consider building NV 1070 RIG. 7 Cards cost ~$3k + $300 for mbd,mem,cpu and additionally some decent PSU...total should not exceed $3500 and you can for sure hash with more than  1kS(+50% or even more) with this piece of hardware.

nvidia hash at better efficiency, i can hash at 1200 sol with only 680w, but it's more expensive

it should be better to go with the 1060 if the ratio with the 1070 is the same
This is just preliminary calculations Smiley
Until final optimizations are done, you shouln't make conclusions Smiley

and nvidia is not even done yet with coding it can be improved in speed, so even better ration is expected, i doubt amd can catch up with the efficiency
Currently, some AMD cards can hash at the same efficiency as this, at a fraction of the cost. Efficiency is only part of the game, if you are 15% more efficient, but cost 2-3x as much, you don't have an advantage.  Sure you could save the money in power in like 8 years (assuming all your cards last that long), but by then you are obsolete, and missed a huge opportunity to be mining twice as fast for 8 years, so still a loss. Also, what are specs that will mine  Nvidia at this rate, consuming that much at the wall?
sr. member
Activity: 410
Merit: 250

http://www.ufominers.com/zcash-equinox

Although one user pointed out that ufominers are known scammers could this actually be the start of asics for zcash, they also apparently do asics for eth as well according to thier site


 Known scammers that do NOT in fact have an ASIC for ETH.

 On the other hand, ZCash doesn't have a "pre-planned obsolete date" like ETH does so if it stays profitable enough it COULD attract the interest of one or more LEGITIMATE ASCI designers.

 The amount of RAM needed isn't prohibitive - but the algorythm's "lots of memory access limiting the speed" issue might limit the gains an ASIC can manage to the point it's not cost-effective enough to bother.


 I'm still a bit amazed that not just one, but *3* different ASIC showed up pretty close together for X11 a while back - that was a MUCH smaller market by the time the first one arrived than any other ASIC had ever been built for.






I suppose it all comes down to the old adage of where the is a market for a product there is some to exploit it (just like the old snake old salesmen from back in the day).

I guess they hope those with little knowledge of mining will buy their wonder product to help to supposedly make them rich but in fact all they are doing is handing over large sums of money for something that isnt worth it in the long run.
member
Activity: 91
Merit: 10
The amount of RAM needed isn't prohibitive - but the algorythm's "lots of memory access limiting the speed" issue might limit the gains an ASIC can manage to the point it's not cost-effective enough to bother.
 I'm still a bit amazed that not just one, but *3* different ASIC showed up pretty close together for X11 a while back - that was a MUCH smaller market by the time the first one arrived than any other ASIC had ever been built for.

FPGA Altera Intel Stratix 10 MX have HBM2 onboard with 1 TB/s
So it's not a big problem.

Sure it will not become a mass product like GPU, but ...
And the other question - cost.
legendary
Activity: 1498
Merit: 1030

http://www.ufominers.com/zcash-equinox

Although one user pointed out that ufominers are known scammers could this actually be the start of asics for zcash, they also apparently do asics for eth as well according to thier site


 Known scammers that do NOT in fact have an ASIC for ETH.

 On the other hand, ZCash doesn't have a "pre-planned obsolete date" like ETH does so if it stays profitable enough it COULD attract the interest of one or more LEGITIMATE ASIC designers.

 The amount of RAM needed isn't prohibitive - but the algorythm's "lots of memory access limiting the speed" issue might limit the gains an ASIC can manage to the point it's not cost-effective enough to bother.


 I'm still a bit amazed that not just one, but *3* different ASIC showed up pretty close together for X11 a while back - that was a MUCH smaller market by the time the first one arrived than any other ASIC had ever been built for.



sr. member
Activity: 378
Merit: 250
If looking into your own cup and your profit makes you happy - then you are happy. This is simply narrowing your point of view and not caring about cosmic problems(like what if black hole opens in our solar system).

Since you want to use space metaphors: the black holes are created in the other galaxies. The smaller miners get to watch the black hole's beauty through their telescopes.

Anyways, we're drifting off topic. It was a legitimate question to the dev since his work benefits lots of smaller miners.
You don't have to have understand (or even agree with the question) in order for him to give his opinion.
sr. member
Activity: 652
Merit: 266
Quote

That's where we disagree. If such effort is being put into a miner that wants to remain "free" for people to use, the net effect of that means that if it is kept "closed and still free", it means they get to stay profitable for longer.
Isn't just that what I meant? Smiley
If looking into your own cup and your profit makes you happy - then you are happy. This is simply narrowing your point of view and not caring about cosmic problems(like what if black hole opens in our solar system).
 
sr. member
Activity: 378
Merit: 250
Ok...glad you agree with me at some points. But I have to disagree on "Trusted source" note you've mentioned.
Do you have contract with Claymore? Do you by any chance know him in person?
Your definition of "trusted source" is legally absurd. 

Legally, sure. However people do in fact have "trust" by the community that is built up with years of interaction.
Does that mean it will prevent anything malicious from occurring? No.
Does that mean it is less likely from anything malicious occurring? Yes.

Finally if you squeeze your point of view to only a few positions it doesn't matter at all since "small miner" is happy with that he/she gets.

That's where we disagree. If such effort is being put into a miner that wants to remain "free" for people to use, the net effect of that means that if it is kept "closed and still free", it means they get to stay profitable for longer.
sr. member
Activity: 652
Merit: 266
Ok...glad you agree with me at some points. But I have to disagree on "Trusted source" note you've mentioned.
Do you have contract with Claymore? Do you by any chance know him in person?
Your definition of "trusted source" is legally absurd.  Finally if you squeeze your point of view to only a few positions it doesn't matter at all since "small miner" is happy with what he/she gets for what he/she gives.
sr. member
Activity: 378
Merit: 250
Cons of CS miner.
1. Not having open source miner leaves small miners narrowed to use only closed source miners and what they offer.

Having it closed source isn't going to defer the functionality as the devs are adding it anyways.

2. Not knowing what's inside that closed source miner may put you and your system in a risk which depending on the person(company,farm) that created this miner may take advantage of your rig(small farm).

As it's coming from a trusted source, that should be enough when combined with hashes to verify its the correct binary.
Look at Claymore's miners for example.

Pros of CS miner.
1. Big farms will hash exactly with your speed multiplied times your rig is capable of using same miner disassembled and probably optimized for certain ASIC/FPGA.

My view is: new opensource releases will be adopted faster by larger farms in an automated fashion, thereby putting smaller miners at a disadvantage.

Cons of FOSS miner.
1. Big farms can use the code to optimize it for their systems/rigs and probably ASIC/FPGA.
Pros of FOSS miner.
1. Code is open and anyone with decent coding knowledge could see if there is maliscious stuff in it.
2. If you are brilliant coder you can port the code for your individual use as you like on any operating system you like.
3. COMMUNITY! Quantity + Quality. Or in other words "If you can see my big balls it really means they are big!"

I'm all for opensource (it's part of my profession) so I agree with your points on the positive aspects of open source.
sr. member
Activity: 652
Merit: 266
How exactly do small miners suffer?
Are you really asking me that?

Yes, obviously you are seeing something I do not and unlike most people I don't assume I'm correct or call people idiots/crazy for having their own opinion.
So go ahead..
As you wish.
Cons of CS miner.
1. Not having open source miner leaves small miners narrowed to use only closed source miners and what they offer.
2. Not knowing what's inside that closed source miner may put you and your system in a risk which depending on the person(company,farm) that created this miner may take advantage of your rig(small farm).
3. How about previous 2?
Pros of CS miner.
1. Big farms will hash exactly with your speed multiplied times your rig is capable of using same miner disassembled and probably optimized for certain ASIC/FPGA.

Cons of FOSS miner.
1. Big farms can use the code to optimize it for their systems/rigs and probably ASIC/FPGA.
Pros of FOSS miner.
1. Code is open and anyone with decent coding knowledge could see if there is maliscious stuff in it.
2. If you are brilliant coder you can port the code for your individual use as you like on any operating system you like.
3. COMMUNITY! Quantity + Quality. Or in other words "If you can see my big balls it really means they are big!"
4. How about previous 3?

And last but not least I am not mentioning DevFee anywhere because it's not important at all.
When you use someone's piece of work you should pay for it. Really doesn't matter the exact payment method (DevFee, One time purchase, Donation, etc...)
I think I covered most of what you're asking...
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Alright guys, I'm over the hump as far as debugging is concerned.
No worries again, it's a really good news for open-source folks, and the ASIC resistance of Equihash is not compromised.
The authors of the original Equihash paper may be professionally embarrassed by it, though.

How fast is it? When do you plan to publish your work?
sr. member
Activity: 378
Merit: 250
How exactly do small miners suffer?
Are you really asking me that?

Yes, obviously you are seeing something I do not and unlike most people I don't assume I'm correct or call people idiots/crazy for having their own opinion.
So go ahead..
sr. member
Activity: 652
Merit: 266
Alright guys, I'm over the hump as far as debugging is concerned.

Would you be opposed to make it closed source (and have no fee attached)?
Again and again...making closed source miners only brings hell on author's head. Private farms have the resources and coders to disassemble anything they want. From closed source miners the only who suffers is small miners...there is no point to release binary versions of the miner.

How exactly do small miners suffer?
Are you really asking me that?
sr. member
Activity: 378
Merit: 250
Alright guys, I'm over the hump as far as debugging is concerned.

Would you be opposed to make it closed source (and have no fee attached)?
Again and again...making closed source miners only brings hell on author's head. Private farms have the resources and coders to disassemble anything they want. From closed source miners the only who suffers is small miners...there is no point to release binary versions of the miner.

How exactly do small miners suffer?
sr. member
Activity: 574
Merit: 250
Fighting mob law and inquisition in this forum
These guys aren't getting what they say..the just have $$$ in mind and profit .soon there will be no miner anymore thank to those cashing guys.
sr. member
Activity: 652
Merit: 266
Alright guys, I'm over the hump as far as debugging is concerned.

Would you be opposed to make it closed source (and have no fee attached)?
Again and again...making closed source miners only brings hell on author's head. Private farms have the resources and coders to disassemble anything they want. From closed source miners the only who suffers is small miners...there is no point to release binary versions of the miner.
Pages:
Jump to: