what i mean is its not that far fetched a story from talks cheep.. I feel with bitmain past record its only a matter of time.. these machines are fire hazards! All of them.. Not just bitmain. if it has a pcie connector its subject to fire. end of story
Best Regards
d57heinz
Oh? Then why do so few PCs with mid-to-high-end video cards burn? It's not the connector, it's morons like KNC that OVERLOADED the connector past specs on certain models (especially the Neptune) that caused melt issues. The SP20E, oddly enough, was one of the very few machines that put limits in how much power it COULD draw into it's firmware to make it very very difficult indeed to overstress it - that "288 watt" limit per connector Spondoolies implimented happens to be the SPEC for those connectors for max power draw at 12V, oddly enough. |
|
|
|
90% of the time its a bad time to invest in mining hardware however people in the past always made money.
Its a gamble that you just gotta take.
SOME people in the past made money mining. Some did NOT. Now is one of the WORST times to consider getting into Bitcoin mining - unless you have FREE electric, then it doesn't really matter. |
|
|
|
"Whether it is practical or theoretical performance of energy efficiency, LK1401's performance was significantly better than other previous generation 28nm chip technology, more advanced processes to bring greater room for improvement. LK1401 as the first version of the currency network, plans 3 - Release low-power optimized version LK1402 4 months, it is expected to reduce the power consumption of each T to 95W. " https://www.bw.com/news/show-70-industry Unfortunately, by the time they actually got the LK1401 to market, the BM1385 was already around and is almost identical efficiency - despite being 28nm. I don't remember seeing specs on the Avalon 6 CHIP, it should be in that same ballpark as well. Apparently they're counting the BM1384 (Antminer S5, R1) and the Rockerbox (SP20E etc) as the "other previous generation 28nm" chips. |
|
|
|
They also seem to think every bitcoin data center is in China and they all run off this mythical cheap power you cannot get anywhere else in the world. Off the top of my head I can name 4-5 states in the US alone where you can get power under $0.04/kwh (No im not talking about residential rates because they dont matter.)
Washington, in 3 counties, but Grant county you have to be an industrial-level user to get into the relevant tariffs - you can get fairly close as a home miner though, 4.something all up as I recall. Where are the other alleged "under 4c/KWH" locations? Are you CERTAIN you are including ALL of the relevant charges? For example, where I live in an Alliant Energy service area, my current base rate at my usage level is just a little over 3 cents/KWH - but once I add in the "Regional Transmission Fee" and the "Fuel Surcharge", my last bill put me a bit OVER 8c/KWH all up - though I've usually this past winter been more like 7.5ish. |
|
|
|
Since AMD and Nvidia also have their own specialties, would it be possible to run a rig that has both brands of cards?
It's possible, but it's a bit of a pain to get set up and working with both type GPUs in the same system at the same time. I've got a couple systems running an AMD APU with NVidia cards, but the APU is set up to do something else (generally RC5-72) while the NVidia cards are doing the heavy lifting on the main thing the machines are doing (ETH mining for now, Folding@Home eventually). ASIC only blow GPUs out of the water on SHA256 (Bitcoin and alts), Scrypt (Litecoin Doge and alts), and starting to on X11 (Dash and alts) but not many in the wild on X11 yet. Other algorythms are either GPU-based or CPU-based, and unlikely to attract ASIC attention due to relatively low market cap or other factors. Ethereum won't be mineable long enough to make the development time AND COST worthwhile for someone to create an ASIC for it (the plan is for it to go Proof of Stake sometime in 2017) - though in theory I could see someone setting up a FPGA to mine Ethereum, I don't think any commonly-available FPGAs have enough RAM to work. NVidia can match AMD very closely on a Hash/W basis for ETH mining - but the cards to do so will be more expensive for the same hash rate, AMD wins hands down on hash/$. Only realistic reason to go NVidia for Ethereum mining is if you have some other usage after ETH becomes unprofitable where NVidia is significantly better than AMD, *OR* you already HAVE the NVidia cards to use. |
|
|
|
|
Found something interesting in NVidia-smi last night - it shows the wattage (AND the TDP limit) as part of it's standard list (-l option) command.
Dual GTX 950s in one box, the one that X is actually running on eating about 85 watts, the other one about 75, running about 10MH/s each with no undervolt and stock clocks (they DO seem to be autoboosting quite nicely, and I DID kick the fan settings up to something SANE).
It's too bad I've not seen something similar on any of the Windows-based utilities to date.
Undervolting CAN reduce your "maximum" hash rate if you take it far enough because you can't clock the card as high - but sometimes the gain in efficiency is worth it.
MINOR undervolting well tend to allow you to maintain hashrate while using a little less power.
ANY undervolting tends to reduce heat, which is a GOOD THING from a longevity viewpoint, as long as you don't go too far and completely brick the card.
|
|
|
|
|
The one common issue folks have with the A2 miners was that the power supplies DID tend to die fairly often.
They're not quite ATX standard - no peripheral connections, just the 24-pin ATX motherboard-type and PCI-E style connectors for the hash boards - but you CAN use a standard ATX-type power supply OF SUFFICIENT CAPASITY to replace them.
I've got *2* of my Mega Terminator units running on Seasonic X1250 units right now.
|
|
|
|
|
Mining on most coins is NOT a high data rate dependent thing.
Ethereum, it might slow the "load the DAG" stage some - but I've not noticed a significant difference on my triple-card rig between the pair of PCI-E v3 8x slot cards and the PCI-E v2 4x slot card, so even there it would be a minor difference at most.
|
|
|
|
|
I'm not - I already dropped out of Bitcoin a few months back 'cause the diff increases were more than enough to make it unprofitable with ANY current miner at my electric cost where I'm at now.
I MIGHT get back into it after I move, and the 14/16nm full-custom generation machines finally show up.
|
|
|
|
|
Nvidia is good at some things (They rock Folding), they are poor at others (they do NOT rock Ethereum mining on a hash/$ basis though they CAN compete on hash/watt), and they absolutely SUCK compared to AMD at RC5-72.
You have to look at SPECIFICS before you can decide which way to go.
TFlops is a totally artificial rating that has very little Real World value as a comparative tool.
|
|
|
|
|
Precision-X (EVGA utility) can do TDP setting under Windows.
I believe it's handled by nvidia-settings or nvidia-smi under Linux.
I suspect there are other options under Windows, but I tried MSI's Afterburner and strongly DISliked it (dark text on dark background VERY HARD TO READ issue as I recall).
|
|
|
|
|
Hint.
MegaBigPower - which at one point has a Bitcoin farm that was mining close to 5% of ALL Bitcoin mining done in the world.
ZoomHash also has a good sized farm in the same area, among others.
There are at least a couple threads talking about "big farms" and where they are at - most of those big farms are in low-cost electric areas.
|
|
|
|
|
Given folks have built their own miners from ASIC from major manufacturers, it could be that Bitfury and Bitmain are planning to leave that "small market" to the "small fry manufacturers" in the future.
|
|
|
|
Yeah I sent them an email just to see if I'd get a reply. They got right back to me and asked 'how many' I said TWO (should have said 20 duh just to get some replies) and they have ignored my other 2 emails since  too small fry ..they are chasing IPO and big miner $$$ without a lot of success imho BUT they make other chips..not just asic's for ltc or btc...THUS the IPO $$$ was for making 14nm chips if they DO get IPO $$ it is a win/win/win for them in 1) investors paid for 14nm equipment needed to make these scrypt miners and such equip can be used to make other 14nm chips by their firm (non-crypto) 2) they can build their data hall first using the above equipment and a much lower cost per chip and cost to mine them as a data hall 3) the 2 things above 'if successful' would be something to point to to get a) more IPO $$ or b) pursue 'bulk sales' of these 14nm chips to other data halls for equip So my 'guess' is the A4 Dominator will be a limited run in that why sell a bunch and water down the $$$ made by your data hall miner operation with a lot of home miner users they will make a few to show stuff exists (like sfards) but the main thrust imho if I am right was the expansion of equipment needed for their biz of 14nm chips and their own data hall mining such (and no one else) till such point they go to bulk chip sales and move on..probably months later I'm betting also the A4 Dominators or the relatively 'few' that will be available for 'pre-order' will run about $4k at the least... we will see how close i am supposedly this A4 Dominator is supposed to hit the 'world' in July 2016....looking a lot like butterfly labs vaporware at this point ....at least to the 'masses' To the best of my knowlage, Innosilicon doesn't HAVE a datahall. They also don't NEED an "IPO" - they have a long-established chip making business that appears to be quite profitible on it's own ALREADY. On the other hand, they probably are pushing for initial sales to large farms and resellers like Zoomhash. There is no real evidence that KNC ever deployed a significant amount of their "Solar" chip in quantity - at most, they managed to deploy enough of it to replace older miners and little if any more, since their block find rate never demonstrated a significant jump. The Alcheminer folks have announced they're working on a "next gen" chip/miner, but no idea how far along they are on it - Innosilicon MIGHT end up with competition again in Scrypt eventually. I sent an email asking about even APPROXIMATE miner pricing, specifying that I couldn't say "how many" without a PRICE. No reply yet, but that might be in part explained by the weekend. |
|
|
|
|
Paper launch is NOT "available" - though NVidia paper-launching the GTX 1080 less than a month before actual availability (presuming they meed the announced May 27 "available for sale" date) was semi-impressive.
|
|
|
|
|
(1) By the time Polaris shows up, ETH could very well have collapsed on price. We're talking MONTHS out before they'll be available for sale.
(2) There is no guarentee the Polaris cards will mine any better or even as well as current cards. Consider what a dud the GTX 1080 is so far on ETH mining.
(3) AMD is notorious for needing quite a bit longer to get drivers optimised for their new cards than NVidia does.
(4) The reports I've seen are all about Polaris initial cards being mid-range - R7 or BOTTOM R9 replacements - NOT mid-upper to high-end. AMD seems to be targeting the "mass sales" nich around where the GTX950/960 are going to have to "hold the fort" for NVidia for the rest of the year more-or-less.
That last point should be a good thing for AMD's bottom line this year, which they need badly after too many years in a row of losses.
|
|
|
|
|
They have some basic specs out on the chip, very little available on the miner itself yet.
Why worry about cgminer, the A2 Terminator units were self-contained and from what I've seen the A4 is also self-contained (possibly might be "add power supply" like the Farm Boys") and appears to be based on the A2 design.
Titans should continue to be profitable for a while - and if you own a Titan and it's not achieved ROI by now, you're in trouble anyway or it was one of the many that had cores die young.
Probability of an Ethereum ASIC is basically ZERO - by the time someone could get one designed and into production, ETH will probably have gone POS or will only have a VERY FEW MONTHS left before it does go POS leaving way too short a time to achieve ROI on an ETH ASIC miner.
|
|
|
|
|
ETH is definitely number 2 for now, but I wouldn't bet on the "solid" part.
S9 gonna have to be a wonder machine - IF it shows up soon - to compete with ETH mining.
If it doesn't show up for a few months, it might have a lot less competition - despite the halfing.
|
|
|
|
|
I don't know about 2, but if you can go with the right size of consumption basis, you can get under 3 in at least 2 and I believe 3 counties in Washington State (as long as they're not hitting their customers with "dry weather" surcharges).
Down side - part of that area is near-desert and kinda warm in the summer, but dry so evap cooling should work REAL well.
|
|
|
|
|
I have seen ZERO about Polaris being available for sale in June. At MOST, it might get a demo/announcement sort of thing per everything I've read SOMETIME in June or early July.
Show sources on your claims, they're NOT matching up to ANYTHING I've seen to date.
Do NOT bother with that "spoof" article the one site did "comparing Polaris to Pascal" on cards that hadn't even paper launched yet - and got some of it's estimated specs wrong anyway.
The earthquake in Taiwan put some serious hurt into TSMC and delayed a lot of stuff a couple months or so - they're back online with that fab now but still playing "catch-up" on scheduals, and there isn't any "space capasity" available yet at that node since it's too new.
Spacing matters to those of us that don't have the OPTION to build open rigs (have you ever seen a cat get into an open rig?).
I might have the option to go open after I move, but where I'm at right now I don't.
It DOES save somewhat when you can go open - no need for high-price high-volume pricey fans, no need for cases, and easy to fit more cards onto one machine with good cooling - though the price of risers mitigates the savings a LITTLE, I'm sure overall it's cheaper.
On the other hand, you definitely get a lot more money spent per watt into bigger power supplies to RUN that many cards at one time - past around 600-700 watts, PS start climbing a LOT on a $/watt basis.
You also have a higher risk of losing a BIG chunk of hash of one part goes down - but that can be mitigated with good design using good parts.
S9 is gonna have to be a wonder machine to compete with ETH mining for at least the next couple months and possibly the rest of this year.
|
|
|
|
|
Show Posts
