Apparently my drivers are so popular that dropbox banned the download links due to too much traffic
Just think how bad it will get if you get a LK-1402 stick miner going. 9-)
|
|
|
LiteGuardian.
The only issue I have with them is they never updated their "estimated payout" figure to account for the block halfing last summer - but it's easy enough to just "divide by 2" on that estimale.
I find that the Dodge merge-mined rewards more or less evens out to the pool fee.
They're also one of the VERY few pools that never had a reported issue with Alcheminers (not that I have any of those, but I was looking - they sold out on the final batch too fast).
|
|
|
Is there anything even remotely close to this performance?
Bitfury's chip has demonstrated comparable performance, if they ever get the bloody thing into MINERS available for sale - depending on how hard they push for hash/chip vs efficiency, they could IN THEORY top the S9 efficiency. On the other hand, Bitmain is probably not running THEIR new chip at max efficiency, it was amazing the initial batches of the S7 showed up at the mid-point of the chip's operating range rather than at the top of the voltage range / least efficient point like the later, lower-chip-count batches went to. I suspect that "yield limit" issues are behind the low available numbers (so far) on the S9. I'm sure the numbers will ramp up with each batch - not to mention they probably kept quite a few (probably a majority) of the initial month or two of unit manufacture for internal farm/Hashnest use.
|
|
|
Oh yeah, my speculation.
S11 will continue to be a 3-board design in a form factor similar to the S7/S9.
25.3 TH plus/minus 5% at 1487 watts +7% at the wall using their 1600 watt PS (93% efficienty). Initial shipping date 16 Febuary 2020. Initial batch cost 4.3 Bitcoin, or whatever that equates to in Dollars = or $3200 dollars, whichever is MORE.
The CHIP (BM1389) will be capable of 0.045 J/GH but they won't run it at it's highest efficiency point or anywhere close to that.
I'm giving it 50/50 they'll come up with a BM1388 chip in late 2017, squeezing another 25-35% efficiency from it through improvements in process node tech and more work at optimisation - in which case there will be an S11 much sooner, but it will be more of an S5 to S7 transition on the same node with quite a bit lower percentage improvement on specs.
BTW - anyone else notice that the BM1386 is missing in action? Gotta wonder if they started work on a 14/16nm not-full-optimised chip design and aborted it when they realised it wouldn't be efficient enough over the BM1385 to be worth producing (Hello LK-1401)....
|
|
|
A board that would fit on the old Gridseed "80 blade" heatsink/fans setups would be nice - even a small one that needed 3-5 boards to fill the entire heatsink assembly would be nice.
Tons of those things around, and they were severely UNDERutilised on the old blade miners - probably good for 200 watts dissipation even pulled apart and used individually with any-old 92mm fan (the ones that came on them were pretty good flow for as quiet as they were).
|
|
|
Have you seen anybody achieve the 40 MH/s on the 290 or 390?
My 5 x R9 290 cards right now are managing 24-25 MH/s each at best. 1350 mem clock is optimal for them (Sapphire ref boards) - anything more DROPS the hashrate fast, anything less drops it but slowly. They stock clock at 947 - but when they thermal limit, they go down to about 900 before the hashrate gets impacted, so DEFINITELY memory system limited. Ballpark 300 watts - have not gotten around to undervolting any of them yet, Real Life keeps getting in the way. When Ethereum is no longer profitable, they're gonna ROCK RC5-72 - already did a bit of testing. 200kblocks/day out of ONE machine not too shabby for years-old tech. 290 are NOT mem bound, check with GPU-Z how big is memory controller load 800/950 and 900/1100 are optimal core/mem With good cooling & undervolting one can run 290 at 1000-1050 core, mem at 1200-1300 and get 30MH I have only 1 such card, undervolted to 0.97V it runs 900/1125 - 25MH, underclocked because is in a case Well, the Bios is Stilt-modded for reference 290 with additional undervolting Don't ASSUME Windows. I don't inflict Windows on any system I use unless I plan to game on it - though I DO wish there was a LINUX version of GPU-Z, I LIKE that utility. I CAN'T get mine to clock to 800, much less 900 on core, or underclock memory at all - limitation of the bios on the things and ATICONFIG even when I SET the peak clocks lower they IGNORE the settings. Hopefully that's something a bios mod could fix. I've seen the same issue on other R-series cards but NOT on the older 7xxx series cards, where I set the memory clock and core clock down but the card don't STAY there, seems that the more recent cards with "smarter" BIOS are STUPIDER about ignoring the settings I give them because they think they know better than I do what's good for specific applications and don't care that their "smarter" settings make them OVERHEAT. Only reference I ever found to the Stilt bios mods had BROKEN LINKS, couldn't download any of them to even try undervolting. 8-( Do you have a reference to them that has working download links?
|
|
|
I'd say 30mh is really all we're going to get on the 480, but 2 of those for $400 getting 60mh @ 300W is still a pretty good deal. I'm considering picking 2 up, but want something to hold me over for the next month. I'm going to see if my microcenter has any 380's and if they'd be willing to take them back for the 480 when it launches.
It will be less than 30 MH/s. The memory bus is just 256bit, it is lower than the 512 used by the 390. Memory is also a lot faster - 8 GB/s vs 5GB/s or some such bandwidth figure. The increased speed should make up a lot for the narrower bus, and archetecture upgrades *might* make up the rest - or might make it worse - THAT is the real wild card we can't judge without a card in hand. All indications are that it shoud solidly beat the 280/380 and probably the 280x but not sure vs. the 290 or 390.
|
|
|
It's pretty widely reported and well known that 3 specific counties in Washington State near the Columbia River have "masses and masses of cheap hydropower".
Think "MegaBigPower" and "Zoomhash's farm" (which is nowhere near their CORPORATE headquarters, I'm supprised they haven't consolodated locations by now) and the like.
|
|
|
BTW - that "S7 lite" thing Bitmain came out came fairly close to your demand. Bit more power usage than the S5, but not all that different - I could have swapped it in place one-for-one with my S5s easy enough, though what I'd do with those Enermax PS it comes with I have no clue - I'd have prefered it NOT have a power supply and LOSE the cost and shipping weight of those things.
|
|
|
What would be interesting to me would be a scrypt OR a SHA256 board with a current chip that was designed to bolt onto the HS/fan assembly from the old Gridseed "80 blade" design.
Plenty of heat soak capasity, it was badly misused and UNDERutilised for that gridseed design, and a TON of them around cheaply available.
(no, the horse isn't dead yet.)
|
|
|
The only difference between "14nm" and "16nm" seems to be marketing and minor implimentation details between Global Foundrys and TSMC (and Intel which doesn't sell capasity to anyone else - I could also specify Samsung but they're allied with one of the others using the SAME process) - which is why I've tended to refer to the "14/16nm" generation. It's interesting that NVidia has specified it is qualifying (or has qualified by now, IIRC they stated "2Q 2016" for the designs to be qualified on both) it's 10xx series products for production on BOTH process nodes - apparently with no change in their specifications, indicating the actual performance is pretty close to identical.
There is a major difference between Innosilicon (among others), with a PROVEN TRACK RECORD of delivering their claims and a long corporate history in semiconductor manufacture and design before Bitcoin existed, vs. a company like Black Arrow that never did make good on all of their claims and was very very late on what little they DID eventually deliver and didn't even exist when Bitcoin first came to be.
On a more important point, got a reply to my email inquiry about A4 pricing today.
"
Thanks for your interest! The pre-order price of A4 miner is not determined yet. Please stay tuned.
Best regards,
Chloe Zhang
+86 18040500320
Innosilicon Technology Ltd."
*SIGH*
|
|
|
My prediction is that if there is an S10, it will be a MEGA MONSTER miner - something on the order of A2 Terminator sized, or more likely 5-6U rackmount, with 60+TH and multiple power supplies needed to run it.
And it won't show up before late winter 2017.
|
|
|
Is it really profitable to mine bitcoin and alt coins? if so then why alt coins have small hash rates?
A large factor is that the algorythms used by altcoins were deliberately made TOUGHER as a way to make it more difficult for someone to create an ASIC for those coins. Look at the hashrate out of an Innosilicon A1 vs an Innosilicon A2 for a good comparison - SAME technology used on both chips, essencially identical board-level design, essencially same SIZE chips (identical chip package but the internal chips might vary a hair on area), yet the A1 hashrate is a TON higher 'cause SHA256 is much easier to compute than Scrypt.
|
|
|
I think you should have bought Bitcoin around June-October timeframe last year.
8-)
|
|
|
Then there is the A4 factor.
I suggest reading up on the Innosilicon A4, which should be showing up probably next month (likely late in the month if so), and WILL be a game-changer for Litecoin mining.
I don't anticipate the A2/Alcheminer/Titan becomming unprofitable with cheap-enough electric real fast - but the A4 should fuel some significant hashrate increase over the next year.
There's a REASON Zoomhash started blowing out their A2 farm a few months back....
|
|
|
I had the Antpool ETH running on one of my miners for a few hours - same issue, it was mining and reported at least a couple of shares but nothing showed up in the pool.
Yes, I did notice and use the BTC/ETH toggle - didn't change anything.
I suspect the pool itself isn't actually working yet.
No Scrypt ASIC can mine non-Scypt algorythms - there have been *2* ASIC (Gridseed GC3355 and SFARDS (Gridseed/WiiBox merger company) SF100) built to do both Scrypt and SHA256 (Bitcoin etc) but those were a rare exception and were specifically built to do so.
I have seen one site advertising pre-orders for an Ethereum ASIC but I STRONGLY suspect it's a scam. I just don't see ETH lasting long enough as a PoW coin to make an ASIC viable to design and build.
|
|
|
RX 480 is aimed at the "mainstream" segment.
Specs so far indicate it should fall somewhere inbetween the R9 380 and R9 380x, while eating less power.
If I had to guess based on ALL of the information that has shown up so far, I'd guess 24-30MH/s at 100-120 watts.
I also suspect availability is going to be a BIG issue for a while, though since it's Global Foundry not TSMC they shouldn't be affected by the earthquake a few months back that took the TSMC production offline for some weeks....
The price however, even if the card *IS* in very short supply, should put some serious pressure on the price of EXISTING card models - which is indirectly a bonus for us miners.
|
|
|
Have you seen anybody achieve the 40 MH/s on the 290 or 390?
Why the 390x is the same speed as the 390?
My 5 x R9 290 cards right now are managing 24-25 MH/s each at best. 1350 mem clock is optimal for them (Sapphire ref boards) - anything more DROPS the hashrate fast, anything less drops it but slowly. They stock clock at 947 - but when they thermal limit, they go down to about 900 before the hashrate gets impacted, so DEFINITELY memory system limited. Ballpark 300 watts - have not gotten around to undervolting any of them yet, Real Life keeps getting in the way. When Ethereum is no longer profitable, they're gonna ROCK RC5-72 - already did a bit of testing. 200kblocks/day out of ONE machine not too shabby for years-old tech.
|
|
|
The real issue is they took too long getting to market.
HELLO S9......with very competative stats.
So much for Bitfury getting to be the "first to market gouge on pricing for a while" option.
|
|
|
Nah, game not over for Bitfury - their announced efficiency for their chip is competative.
On the other hand, they DID take too long getting to market, so they're going to HAVE to compete when they do finally get something out there. No "first to market" bonus gouging pricing for THEM!
Bitmain I figure will drop the price quite a bit once someone actually DOES release a competative product - I can and WILL wait, $2100 is too much IMO even with the hashrate and efficiency.
If you assume 3% or so hashrate growth, I suspect you can get that 7-8 month ROI - the halfing is likely to KILL a bunch of older miners due to massive unprofitability even at VERY VERY cheap electric rates, and so will likely see a substantial dip for a while - the S7/A6/BEleven units should stay profitable if your electric is cheap enough, but SP20/S5 generation forget it unless you have FREE electric I'd bet.
I'd NOT assume network hashrate growth in July.
|
|
|
|