Topic: [ANN][AUTO-SWITCH] Profit-switch auto-exchange pool: CleverMining.com - page 149. (Read 554384 times)

I paid around that, the current average for such a set would be 6000 USD.

They're annoying to keep running since it's "a new product" and not much has been invented/cristalized though.

Well you have now.  We're done talking about them.

Is he mining scrypt with them?  

I know there are units for Bitcoin, I just have not heard ones that do scrypt. 

Just use google.  Google is your friend.  There is no need to keep making up numbers if you just check with google first.

There you go -- there's an owner of them.  32 gridseeds would run about US$6-7K, right?

Please post it, I would like to actually see one.

Your impression would have been correct up until about a month or two ago - give or take.
I am not going to post links here to any sellers and give them free advertising, but i'm sure you could easily find them by using google.
The hashrate and power utilization numbers I posted were not pulled out of thin air.  They are measured reported rates of actual units.

Currently running 32 gridseeds, total power usage is 270 Watt. 10 Mh/s works fine.

I was under the impression this is only being done with non-scrypt coins.

Only doge and LTC are big enough for giant farms, its an alt coin for a reason.  You cant park your supertanker in the paddling pool sorry.  
Multipools already get complained about and they could be dwarfed by a production line, they will have to mod code to always be high diff I guess

People already own gridseed units and are currently mining with them at these power consumption levels.

That is awesome if that is the case, but, I still am a bit skeptical.   Pre-orders are good way to get to play with people's money.  There is every reason for a company to be less than truthful, or embellish their offering in order to secure that pie.

How much heat does the chip generate at that performance level, how long can it sustain it, how much cooling do you need.   A raw MH/s doesn't mean much, it could be a peak, it could be what they got in a short time span.   What is the life of that chip when it runs that hard, etc...   These are all things AMD and NVIDIA have to deal with, before they shoot their mouths off. 

I also have a feeling these things will be DOA when they are given to the customer, I don't mean they won't function, but I think the manufacturer will weigh the profit/loss ratio of when to give you your miner.   We will see them on the network long before a singe pre-order is filled.  

From what I read recently, 10 gridseed units, each of which contains 5 gridseed chips, are capable of scrypt hashing at 3 - 3.4MH/s at a total power consumption of 100 watts, or 10 watts per gridseed unit, presumably plus whatever it costs to run the device to which they are connected (computer, special purpose controller, whatever).

That comes to just about 35ish watts per 1MH/s, which is a lot less than the 150 watts per 1MH/s figure you suggest for a future generation asic device.  And the gridseeds are available and running now.  This compared to GPUs which consume a bare minimum of approximately 250 watts per 1MH/s at present?

I don't know about the low costs of running things.  Right now 1Mh required 300 watts.  I have heard that power is the barrier in producing these ASIC's that can mine scrypt.  AMD and NVIDIA have been spending millions in R&D to get these chips to be more efficient.  Heard that even at 20nanos these things still suck power.  They would love to omit those power connectors and keep you from needing to upgrade your power supply.   I can't see how some fly by night is going to beat AMD and NVIDIA on R&D.
 

The one offering I am hearing about is the Titan, its clearly stated that the miner can be pre-ordered without a Power Supply.  You should ask yourself, why a $10,000 miner is lacking a powersupply.   Perhaps the requirements are going to be steep.  Lets say 100Mh, is cut down to 150watts per 1Mh, which blows away the big boys like AMD and NVIDIA by 100%,  that still requires a 15,000 watt power supply.   What these preorders don't tell you is that you may need to spend another 10 grand on the powersupply unit, and maybe even require 3 phase power to run your miner.


So far, there is not even a jpeg of one of these units, and I sure have not seen the specs of how much power one of these Titans will suck, and that is as much of a consideration as the 10K pre-order price.

My point is not about ROI about rather about running cost.
These gridseeds have a very low cost to keep running, which means they'll continue to account for a large amount of the network for scrypt coins, even while profitability continues to drop.

And yes, my point is GPUs will either shut down or switch, thus leaving ASICs as only viable scrypt miners.
While ROI is relevant for the purchase of new units, it is not for the large amount of existing units already out there.
Existing units will keep running until the profit drops below the cost of running them.

But I already addressed all of that in my earlier post.

I honestly don't think crap coins are ready for ASIC, this isn't bitcoin, most people are middle mining them for something tangible.   Once ASIC comes on the scene it will just kill the coin.  Bitcoin had really good distribution when ASIC hit the network, the crap coins will die fast when ASIC miners come on the scene...   and I doubt they will be able to recoup their investment, the market will drop fast.   Right now its working because of the distribution, screw that up further and it will be a very dead horse.   

Your pool operator Terk seems to have an excellent grasp of what is going on in this hashpower theft attack.  

If I could add a point or two, if the mitm attack is a dns hijack, in #2 above, if a server host name is configured in miner client, and that server host name is still contained with the operating system's cache of host name to ip address mappings, then the miner should connect back to the legitimate server when a stratum connection breaks.  However, if the cache entry has already expired, then the server host name would need to be resolved again and possibly receive an illegitimate ip address in return.

So in the very short term in order to minimize your attack surface, you could disable external host name to ip address resolution, by hard coding a server ip address into the miner configuration, or by creating a static host name to ip address mapping in your operating system's version of a hosts file.  Either of these attack mitigations must be accompanied by disabling client.reconnect messages in your miner software (for now), in case the mitm attack is affecting more than just dns resolution.

In #3 above, the hashpower thieves are likely using client.reconnect to capture your miners at an illegitimate server if they are unable to steadily maintain the mitm attack.  For without it, you miner would just connect back to the legitimate server if that hijacked stratum connection were to break.

Terk, if you have a minute, I would really like to get your perspective as a pool operator on this stratum mining protocol enhancement suggestion.  Thanks.
https://bitcointalksearch.org/topic/m.5867183

Except, of course, when the new ASIC farms come online in the next 3-6 months, and everyone switches away to nScrypt and other algorithms which these ASICs can't handle.

At the moment the ROI for a Gridseed is in the 9-12 month range (accounting for difficulty), and I have little faith that straight Scrypt based coins will remain profitable for that long.

I think GPU mining will remain the best long term bet for quite some time now.

This is the version I am using.  So far so good.  I did have cgminer hang twice on one rig in the last few days, once where GPUs went idle, and another when they kept running submitting 100% rejects to CM.  I think those instances were unrelated though.

Is there a main thread to discuss this specific issue anywhere?

And what percentage of the posts that you have read in this forum and others like it would you say have actually given you that idea?  There are indeed some very intelligent people around, but at times they seem to be few and far between, or completely silent.

From the one message alone of yours that I have read, I can easily assume that you likely know your tipping point.  Can you really say the same of others?