Batch 3 is up again on bitmaintech
and it's gone...or at least I don't see it up now. Hi Guys,
Just setup my S7 and at full operation running after a few minutes, there a whistling sound - is this normal?
TMT
not sure what you are referring to. Usually there is a pretty solid sound. I had one S7 where push fan's grille was overtightened and it was producing a screeching noise that disappeared after I eased one or two fan screws a quarter or half turn. |
|
|
|
Check this out (from pcfli, who might have the inside track): https://bitcointalk.org/index.php?topic=934581.msg12769055#msg12769055New batch is apparently in 10days and will be deprecated to 3.5TH-probably in lieu of the competition that will have a ~3.3Th machine for a bulk price of what was posted here already (very low indeed if it is true). I sincerely hope they would seriously reduce the S7 "lite" price even if this would kill the aftermarket price of S7 "large". Or, maybe they already have a 14/16nm solution? I think its something like the batch 2 being the rejects from Batch 1 except even more unstable at higher freqs than the batch 2's... therefore they have to really drop the clockrate to get the 3.5TH/s and stabilityThose units that do not make the advertising claims of 4.8TH/s are money thrown out the window if they don't make them work somehow. So it would make sense to throw them through a rework process, if they still don't preform, then lower the clock rate and look for the stability sweet spot and sell them as a de-rated S7. The S7 B2's were de-rated to 575Mhz clock, maybe the S7 B5's are the real flunkers and are clocked for stability at 500Mhz clockrate. Before tossing the ones that don't meet spec into the hopper and take the loss, I would re-work them and look for a workaround to recover some cash from the QC fail units. very unlikely, in my opinion interestingly, if software provides option for 4.45TH miner, I wonder what is it? |
|
|
|
Expect 600 by new year and a new ATH before halving!!!
impossible, bitcoin always overheats, then crashes (temporarily). One thing I would agree on: it would be difficult to either mine or cheaply purchase bitcoin after halving. Mining: 14/16nm chips are coming this winter/spring and they will bump the network speed at least 2-4 times over 2016 making it more difficult to mine. In addition, with 50% less bitcoins provided for sale by miners, even with the same demand we might get market imbalance until the new price equilibrium is reached. |
|
|
|
Check this out (from pcfli, who might have the inside track): https://bitcointalk.org/index.php?topic=934581.msg12769055#msg12769055New batch is apparently in 10days and will be deprecated to 3.5TH-probably in lieu of the competition that will have a ~3.3Th machine for a bulk price of what was posted here already (very low indeed if it is true). I sincerely hope they would seriously reduce the S7 "lite" price even if this would kill the aftermarket price of S7 "large". Or, maybe they already have a 14/16nm solution? |
|
|
|
Bitmain is all sold out of all S7s Batch 1, 2, and 3.
SPECULATION (without any facts)
Is there an introduction of a new product? Maybe with the new chip.
Are they releasing the new S7+
Are they now trying to catch up?
Are they freezing sales to avoid loss with the increased BTC value?
Just Saying
Getting caught up, IMHO. IF an S7+ were to come out, that would not stop them from making the S7, IMHO. I do not see a new chip (14nm or 16nm) coming out until possibly this time next year, IMHO. I personally do not feel they are freezing sales. If the price of bitcoin continues to rise, they could simply continue sales while increasing the price of the S7's in proportion to the rise in bitcoin price and difficulty to continue making profits. just a weekend there, we will probably see new batches on Monday or even sunday pm. If bw will really produce something in winter time, BMT will be pressed to accelerate the 14/16nm rollout. |
|
|
|
If I was to give a suggestion to bitmain engineers it would be to force a higher initial fan speed on reboot, like SP20 was doing.
The way it is currently setup, upon reboot there is an abrupt change from high speed fan to initially low speed, which I think is causing temporary chip overheating. Board sensors are not highly relevant here because sensors are probably removed from chips. I think that this putative chip overheating might be causing the initial higher error rate upon reboot. Once fan reaches high speeds, the system equilibrates and errors decrease.
Yes, whatever the reason it is something that Bitmain ought to deal with. It may be thermal related as you suggest, but it may also involve details of how a chip disables cores that are creating too many hardware errors. The restart problem doesn't seem to me to be thermally related, because I've seen roughly the same situation with room temperatures more that 30 degrees C different. (If the problem is an algorithm hardwired into their chips, I doubt we'll see any improvements.) yeah, most likely the actual chip function as it is unlikely to be temperature driven if 30C ambient difference shows the same effect. maybe abrupt shutdown also messes up with chips as well. i had killawatt that died and S7 was connected through it. Once I restarted the miner , it was showing lots of errors for the first couple of minutes, which mostly disappeared upon soft reboot, but it too good 1-2 hr until error rate came back to usual 0.0064% and stayed there. |
|
|
|
Sometimes the error rate can stay high after a system reboot. (Sometimes even after several system reboots.) In that case, I've found it useful to power cycle. Let the machine run for a few minutes and then system reboot. This seems to get things back to a low hw error rate with my unit. Then I can increase the clock rate, hopefully this time to a value that won't screw things up. Presently my early batch 1 unit is averaging 4.87 TH, 618M, 57c, fan 3000 rpm, 1d21h. I plan to leave it alone unless something changes for some reason or other, such as a power interruption.
If I was to give a suggestion to bitmain engineers it would be to force a higher initial fan speed on reboot, like SP20 was doing. The way it is currently setup, upon reboot there is an abrupt change from high speed fan to initially low speed, which I think is causing temporary chip overheating. Board sensors are not highly relevant here because sensors are probably removed from chips. I think that this putative chip overheating might be causing the initial higher error rate upon reboot. Once fan reaches high speeds, the system equilibrates and errors decrease. |
|
|
|
It appears that my Sep 29 B3 order has been shipped per UPS 😀
they are hopefully going by the dates ordered/paid, not by the batch # |
|
|
|
Maybe they start to ship s-7's from batches 2 & 3
Quick question on batch 1 does anyone not have their order shipped?
i sneaked in another batch 2 order yesterday on a btc spike; was deciding whether to order batch 2 or 3 and went with batch 2 this time based on intuition (I could be wrong there). my prior b2 order has not been shipped yet. |
|
|
|
I do not think reward halving will have impact on S7 price. Tooo much time until that...
I would have to agree with you on that one to an extent. That's the only thing I could come up with for the price reduction between batch 1 and batch 3. I think the key is it will not yet. But once we hit say 6 month's, then 3 months etc no doubt it will effect it. With less reward people would not spend as much. But we really don't know a lot of what will happen at halving we need to know to determine miners price. BTC price at halving will affect miners price greatly, and no one can tell us what that will be currently.. Agreed, we don't know what will occur with difficulty and btc price. I plan on buying more every month. I'm not saying how many each month but it will be quite a bit. I plan on buying ten of their 1600 PSU's as well. The way my PDU's are set up, I still have more watts available on each PSU for more S7's. I would rather do this with BITMAIN's PSU's than use an IBM 2880W PSU to power only one S7 when it can easily power two. It would be a waste to do this with an IBM 2880W. Hence, the need for BITMAIN's 1600W PSU. I've been up all night. It is 5:39am at my location now. It's past time to get some sleep. Good night everyone! I must be missing something, so power 2 with an IBM 2880W PSU if that is what you can do, heck power 5 per 2 IBM 2880W PSUs. You have lower cost, and better efficiency than Bitmain PSUs, I am not understanding the logic since you make the point you can power two per PSU. I wanted to ask who sells them, but then figured it out  Quick question: for IBM 2880W PSU, 8 PCIe cords with 8 extenders-how many connectors total: 8, 16 or a different number. Will it work on 208V instead of 240? |
|
|
|
you can't jerk these things around too much as you don't know when bad luck will change to super good luck (I wish i could, but it is random).
it takes a while sometimes. I often see petahashes sloshing around chasing luck.
There have been 17ph jumping in and out of Antpool the last few blocks  Ufo it's like daytrading in stocks; it won't work long term, but you can get lucky a few times. |
|
|
|
My luck on Antpool the last week right before the difficulty adjustment and good coin prices!!!!!
24 Hours 3 Day 1 Week 2 Weeks 1 Month 3 Months All
76.00% 83.00% 94.00% 103.00% 101.00% 100.00% 101.00%
Yes, I have moved half of them to f2pool after noticing how little I made today.
Ufo
you can't jerk these things around too much as you don't know when bad luck will change to super good luck (I wish i could, but it is random). it takes a while sometimes. I often see petahashes sloshing around chasing luck. |
|
|
|
These Batch 2 and 3 miners probably won't get shipped until,
* Price of BTC goes to $300-$315 and reverses back down. I think they are holding them right now and mining with them so they can dump.
*Or will ship them on November 5th as its the last delay day.
Mining is very profitable at the moment.
to buy bitcoin a few weeks back, then ordering batch 2,3 now would be profitable too. b2 is going for less than 6BTC, shipped. |
|
|
|
The s-7 is better then the s-5
And depending on your setup there are mufflers that work.
I disagree. To me 1 x S7 is much louder than 3 x S5's. Even with fan speed reduction to 30%. I say everyone has a different pair of ears. The pitch of the s-7 may sound worse to you. But the s-7 is far easier to manage sound wise then the s-5. In any case, it might be unhealthy to be close to multiple S7 (he had 7 running) without any kind of hearing protection. |
|
|
|
So I just had a slightly disturbing issue with one of my S7's - I just happened to be near it when it started beeping (the same thing as if the Internet was down). I went in to check it out, none of the other machines near it were having problems, so I power cycled it. It didn't renew it's DHCP lease, nor did it try to get another address, but it powered up as if everything was ok... More troubling is that I was watching my power monitoring, and the power jumped as if it were hashing full speed - I went back to the unit, the fan was spinning but on a very low setting, and the air out of the back was HOT. Having flashbacks to the S5 burnup problem, I unplugged it again, let it sit for a couple minutes, and plugged it back in. Same issue. This time I pressed and held in the small hidden factory reset button, after a couple seconds it gave the long solid tone that normally means it was successful, but once again - no DHCP request, and it looks like it's hashing even though it doesn't have any internet connection.
REALLY not wanting to see a repeat of the S5 issue - loosing a couple hundred bucks a pop when it happened was bad enough, but loosing thousands would be brutal.
I'm going to swap the BB with another S7 to see if that might be the issue, and I'll update as I do some more tests and see. They've been running fine for several weeks, I power them off of a IBM 2880W PSU (one PSU for each one, since I have a bunch), so power shouldn't be an issue...
hashing where to? in any case, maybe turning it off for 20 min instead of just 2 would be a better idea to make sure things equilibrate thermally. I would put a fan (like Vornado) blowing at it as well. |
|
|
|
Feels like 2012 again, actually, with this flat prices and quiet move upwards. I really believe that those dudes who sold at ATH are now accumulating for the next run. The whole year is, I think, the greatest time to buy more seeing that it just stayed below 300 most of the time. I see the graphs and read the explanation, and I love it. Realistically speaking, there is a slim chance of this to happen, but hey! Everything is possible. IMO by 2030 there is a slim chance this won't happen. Question is.. are your balls made of steel or balasa wood? Winter is coming John snow. Mine are made of steel, that I'm sure of. If in case it does happen within the next 5 years, I don't see a reason why would I sell anything right now. I'm a believer yes, but not delusional as others may seem to be. Realistically, we wouldn't be able to reach that $13000 in just a year. We can though, if hyperinflation comes to play. All it will take is for people to have faith in BTC as a store of value and move some from gold to BTC and boom. yeah, there is Silbert's video interview out there where he posits that bitcoin price will go up 100 fold if just 5% of current gold value (7tril) transfers to bitcoin, nothing else required. https://www.youtube.com/watch?v=LOtgNhmNRm4 |
|
|
|
Overly optimistic in my view. We cant' even stay at 270 right now. The past is no indication in bitcoin. Remember that the past 24 month long bear period has broken new grounds. Too many people bought in at $1000+. If and when we reach $1000, they will dump their bitcoins, we will go right back down to 200.
No. When the price reach $1000, most of the people would think that their long awaited target reached and start to sell that may pull back some 10% to 20%. Usually we say that profit booking. A profit booking never drag back to the origin. But the new money in flow will make it to pass $1000 again. I can't imagine how awful it's going to be for the responsible ones who sell a significant portion or all of their coins at $1000 or even $2000. Who knows where the top of the next movement is? I was a bitcoin GPU miner back in 2011. I watched as the price hit an all time high of ~$31 that summer, then collapsed. I sold everything when bitcoin was in the low 30s in early 2013, thinking it couldn't possibly get much higher than the previous ATH. Within a matter of months, of course, the price was ~$1,200. I'm not making the same mistake again. up to $40-50K, then? I admit that it is developing slower than I initially thought. The true breakout will happen if we get a wind of some major government actively sourcing miners or purchasing btc. Could be China, Russia or US-I wouldn't be entirely surprised by any of these three. In fact, I am surprised that a few consultants did not rec this already. After all, expenses would still be meager at this point ($400 mil for 10% of btc) and having 10% of BTC might come handy later on. |
|
|
|
LMAO,there will be NO prices for us "home miners",unless someone leaks them  Move on to companies who WILL sell to the general public,SP-tech is not going to be one of them  That's because these new miners are not meant for home miners, they're being made probably mostly for internal use, with a few shipments going out to some big data center folk. The SP50 was never intended for individual use or sales, probably one of the reasons they're so hush hush about pricing and such, because ultimately it doesn't matter unless you are in their inner circle. It's like Ferrari where many of their models are only available for purchase by Ferrari owners that the company personally invites to buy one of their cars. In my opinion making big ass miners is a mistake as it locks you into technology that is prone to be bypassed in a relatively short order. making the boards replaceable only partly solves it since what are you going to do with older boards, unless you have a Sp40 frame where you can stick them in and potentially sell to the public as used goods. I wonder if the chinese company that announced a 14/16 nm miner is for real or not. I am quite skeptical, but if it is real, SPT/BMT will have their hands full competing. LMAO,there will be NO prices for us "home miners",unless someone leaks them Move on to companies who WILL sell to the general public,SP-tech is not going to be one of them That's because these new miners are not meant for home miners, they're being made probably mostly for internal use, with a few shipments going out to some big data center folk. The SP50 was never intended for individual use or sales, probably one of the reasons they're so hush hush about pricing and such, because ultimately it doesn't matter unless you are in their inner circle. It's like Ferrari where many of their models are only available for purchase by Ferrari owners that the company personally invites to buy one of their cars. yes, but this cannot work with a bitcoin miner, no matter how efficient it is, because it is not a relatively unique item like Ferrari or Lamborghini. If you make it an exclusive, the rejected party could just buy bitcoin instead of messing around with miners. No such thing with F/L. |
|
|
|
Last thing to keep in mind, if your PSU is pulling 1200watts at the wall, its not running 300watts per pci-e, since you need to deduct the watts wasted by the PSU conversion and also the fan's usage.
Partly correct. The fan usage DOES come out of one or more of those PCI-E connectors (I think it's probably split between 2 of them, but haven't pulled my SP20 apart long enough to actually check it). DON'T count on 1.7 TH - you have to have a very cold room to even have a prayer of achieving that. 1.3-1.4 TH range is achieveable easily even in somewhat warm rooms though, and is quite a bit more efficient to achieve. Do NOT under any circumstances try to run 2 connectors from one cable, unless it is a VERY custom cable with something on the order of 12AWG wiring. SP20 does not have wireless. Standard twisted-pair Ethernet only. The official spec on the MOLEX connectors that the PCI-E power specification uses works out to 288 watts per connector (the pins are rated more, but have to derate for use in that particular connector style). Exceeding this tends to lead to issues like the KnC Neptune and some Titans had with MELTED connectors and resultant fire and damage-to-unit risk. Most power supplies use 18AWG on their PCI-E connections - this is plenty for a SINGLE connector at 225 watts (per the PCI-E spec), it gets a bit marginal at 288 watts. I wonder, what is the highlighted based on? I run Sp20E very downclocked at ~1200Gh using ~640-650W at the wall from a single Corsair CX750M, which has 2 split PCIE cords providing 4 required connectors. The cords or connectors never exceeded 45C and I have a lazer temperature probe. I grant you that you have to first connect a more powerful PSU (like EVGA 1300) or two CX750, then change settings on the machine to downclock, check power usage, achieve stable hash and then optionally changeover to a single CX750, but it is doable. I run a machine like this for almost 9 mo with no problems whatsoever. |
|
|
|
I wonder how they distinguish between batch one,two,three?
In regards to shipping or in regards to testing each board for its quality? Batch 1 & 3 are the same in specifications with different price and shipping dates. I got a batch one clocked at 575. Is there a label on it to specify? If/when I sell how will the new owner know? It's quite easy... All 44 of my S3's, S4's, SP20's and S5's were sold on eBay. I posted screenshots of the "Miner Status" page for each rig AND the "Network Settings" page for each rig. This way, they new by the MAC IP ADDRESS assigned to each rig by bitmain on the assembly line is the actual rig they will receive. It's like an SSN for each rig to identify it. They also see several days of the rig mining consistently at 4.8+ TH/s at 600 MHz with low HWE's. I also let them know since this rig is running super as proved by screenshots, that also shows the date the screenshot was taken, there will be no returns for the rig(s) they purchase from me. I did all 44 of my various models of mining rigs this way without a hiccup. call me old fashion, but I most likely would have never bought a miner which MAC address was publicly posted on ebay. You don't think that this could make remote hacking of it easier? i am not a network specialist, so maybe this concern is misplaced. |
|
|
|
|
Show Posts
