About the same error rate per hour, actually. STILL way high on the error rate. I'm starting to think MarkAz has a point on the long cable lengths.

 They can put up to 20 BM1384 on one string per the chip specs - which works out to a lot better efficiency, abet quite a bit lower hashrate. IIRC I posted a calculation in another thread where a 20-string was somewhere in the .30-.35 watt/GH (or J/GH, a Joule (sp) is 1 watt/sec so it should be J/GHs).



 I don't think ANYONE was expecting this particular unit or form factor. I still think it should have been called "S6" not "S5+" since it is not even CLOSE to the S5 form factor but is semi-close to a rack-mount form factor. Or perhaps "S5Trio+" might have been even better yet.


 At the current Yuan price on the Bitmain chinese website, the $ price will probably be around $2750 + shipping at ballpark $100-$150 when they make it available on their main website, unless Bitcoin makes another significant jump/drop before Friday.




 3337 HW errors in 8 hours? Something wrong with at least one of those blades - the 67C on the one blade don't look real good either.
 Nice to see this CAN be done though in case of a controller failure at some point. Just use some cable ties to hang one unit upside down from the next shelf up (with the bottom closed up with cardboard or something for cooling) above the other to keep the cable length short.

I figure there will be a jump in the month before the halfing - based on speculation.

 I dunno if the price will STAY higher a couple months or more past that though. Some of the jump might stick, but I'd hate to bet on all of it doing so.

 Longterm, I believe bitcoin will slowly increase in price - possibly even beating inflation.



 There is of course the possibility of someone with a large number doing a "dump and pump" or "pump and dumb" and bouncing the price at ANY point, or multiple points.

Lktec probably in December per a few comments I've seen. Might be as early as this month but I suspect their August 20 announcement is more likely to be "more exact specs" and possibly "you can now preorder" their unit(s). Not sure if they're going to be SFF though, their previous units weren't real small.

Possibly Innosilicon with a direct unit of it's own around the same timeframe, but their previous unit was a rack-mount size design not SFF.

In theory SFards should have the SF100 available in retail quantities by then, but I want to see how reliable those are going to be at the BOARD level - Gridseed tended to be GARBAGE as a board-level designer and so far what little feedback I've seen on the SF100 is VERY much "more of the same bad design work" or just bad manufacuring/shipping causing too many dead units.

In theory the Avalon project could have something out soon - they've been VERY quiet for quite a while, have had plenty of time to get a good start on a "next gen" product past the 4.1

 There is an already existing LARGE supply of Bitcoin though, the block reward just adds a very small amount to the supply.


 The pair of used S5s I got from Bitmain were clean, and in generally good shape. The shipping packaging wasn't great, but the only damage was a couple of fins on the bottom of each heatsink bent up a little, trivial to "fix" and would probably have worked OK without "fixing" them.

But unless you're on free electric, or something totally insane like 1-2 cents/KWH, .9GH/watt is losing money making those S1s a liability. Well, no longer an asset anyway.


 (edit) looks like a dropped a decimal point, that would actually be profitable up to 8-9 cents/KWH at current bitcoin diff and price. Which is still pretty low cost electric though, most of the USA pays at least 10 cents/KWH and I think the average is pushing 13 as of last year.


 I'd hate to try to ROI if the unit wasn't already mostly paid off though. 8-O

 Varies some, I've seen a few high-end Gold units lose closer to 2% between those points.


 Miners are not liabilities as long as they can mine at a profit. I'd class them as short-term high depreciation assets, vs, power supplies where a good power supply can be a very long-term asset.

 Given that I anticipate the "in process of showing up" generation of miners to have a high probability of being viable for 4+ years, I doubt that power supplies will outlast 3 generations of miner past the "current showing up" generation on a high-probability bases. Miners are catching up to "state of the art" in the current generation, after a few years of each ASIC generation being well behind the current state of the art for semiconductor manufacture.

 50% to 80% load on MOST gold power supplies is usually less than a 1% drop in efficiency. Seems like the optimal point from a ALL costs basis is around 70-80% ballpark - which is also a good point to be at from a LONG TERM reliability basis.


 In my S5 testing, the efficiency was pretty close to flat vs. clock rate from 300 Mhz to 380ish. The varience seemed to be mostly "measurement tolerance" and a very small contribution from the small efficiency difference in the power supply with varying load level.
 I don't see dropping to 275 being a significant efficiency change.

 POWER usage on the other hand did change with the clock changes, downclocking WOULD drop the total power used appx. in proportion to the clock rate change.

 2TH at 700 watts wouldn't be practical out of an S5 form factor unit, you'd need more chips than you could fit on 3 boards. 1TH at around 350 though I could see happening if Bitmain wanted to do it, 2 strings of 20 chips per board with 3 boards - I'd have to go look up one of my posts in the Gekko BM1384 thread to find the figures I worked out.

 When they claim their English is good but the english they use is poor enough to make them VERY hard to understand for a native English speaker, it does.


 I can handle folks where English is a second language and they're poor at it - but folks that lie STUPIDLY about how good their English is when it's obvious that they don't understand it well are irritating, like any OTHER liar.

 They'd have to go 20-string to even get close to the numbers Lketc has already posted for it's upcomming miner at the miner level. I don't think they have enough space on the board to put 3x20 strings on it, though with the power reduction per chip cooling would be NO issue with either an 18-string or a 20-string.

 
 I don't see any chance the S7 will be BM1384 based, given previous comments out of Bitmain. I can see a VERY FAINT possibility of them putting out a "S5++" with more chips per string, but it would have to show up Real Soon Now and would have to be priced lower per TH than the current S5+ to have a prayer of ROI.

Interesting, folks have previously said that the S5 couldn't be reliably unvervolted past about 11 volts....

The original S5 fan appears to be the PWM version of the Delta FFB1212EHE-F00 - 2 amp, 4000 rpm, 190 CFM, 120x120x38mm. I can't find a link to the exact model though.

 They are DEFINITLY a delta design, as that "flow director" setup is patented as far as I can recall....

 Delta makes good fans, but they aren't the only folks that do so - I've had very good luck with NMB and some mixed but usually good luck with YS Tech, but THE longest lasting fan I have is an ancient Rotron metal-frame 115VAC that's been running for literally decades. Mostly fan duability comes down to being a ball-bearing design - most of these recent "rifle/fluid dynamic/etc" just are just fancy sleeve bearing stuff that does NOT last nearly as well as a well-done ball bearing.



 Whynot - there is a "for sale" area of the forum to try to sell stuff in.

 Probably a good setup for X11 or such, but it's gonna LOSE money hand-over-fist for Scrypt or SHA256.

 I do agree that the Maxwell NVidia cards are the way to go for cryptomining on GPU-viable coins now.

 

PCI-E specs the 6 pin version at 75 watts because they are being VERY VERY conservative. It was the 8 pin version (2 extra ground leads, NO extra +12 leads, DUMB choice IMO) that specs 150 watts.

 These are PCI-E specs, NOT connector specs.

 Power supplies with PCI-E connectors are a LOT more common, I see no point in reverting to primitive screw-type terminals and many reasons NOT to do so in the wattage range these miners run.


 P.S. - worked at Qualidyne for a while, where 100 amps was a LOW current output on many of the PS I tested for use in mainframes, commonly at +5v. This was a long time back though, you needed most of a square foot to fit a 1KW power supply into back then, switchers had a lot lower frequency capability due to the parts being many many generations older tech....



 One plus side (for me) on using standard ATX power supplies - eventually the living ones can get switched into real computers, if Bitcoin mining ever becomes completely uneconomical and I get to switch back to working on my d.net/GIMPS/Folding@Home planned farm....




 Not a real good idea, but if you use the heavy-duty A/C type cords it would work. I prefer to just wire additional outlets close to the mining area(s).

I have run my Gridseeds on a windows system, and a Linux system. The Pi is commonly used to run a small Linux system, but I already had existing LINUX machines around so I just used one of those.

They went to 16 chips per string, dropping the chip voltage from .8 to .75 - the BM1384 has a LOT of room to be more efficient via undervolting, sadly nothing out there except for one stick miner ever did anything with that ability to a significant degree (I don't count the S5+ change as significant).

 S5 doesn't seem to change efficiency much at all just via underclocking. Small enough to be measurement error rather than any actual change.

 I was thinking about a move to the low-electric-cost part of Washington State at least a year before I ever started Bitcoin mining, the mining just gives me more reason to go once I get a viable amount of cash saved up for the move. Summer rates here in particular are insane - there's a REASON I turned off pretty much all of my d.net and gimps machines in mid-June. If I had 60TH of mining power I'd already be GONE.

 BTW - based on my research 18-gauge wire in a common PCI-E cabling setup should be good for 7-8 amps at 30degrees C (NEC code doesn't appear to go below 14 gauge, so I had to extrapolate from wire cross-circular area) - so yeah, it would be definitely marginal vs. the capasity of the connector. Still plenty to handle the under-200 watts per connector of a S5+ on a 2-connector setup though unless you get the thing to overclock a LOT. On the other hand, cool running cables are a Good Thing so the third connector doesn't hurt anything (except perhaps the price by a few bucks or so).
 Keep in mind that the "3400 watts" is the usual Bitmain "AT THE WALL" spec, the actual draw on the connectors would be more like 3100 watts total or ballpark 14.5 amps per connector on a 2-connector setup per hash board, since they claim to use a Gold-rated power supply for that spec (but never really state which one they use, some are a little MORE efficient than others).

Sfards has barely shipped anything. Not really worth being responded to, especially if the ONE report I've seen about "in the wild" ones is a good representation of the (UN)reliability level of the SF100.


 I'd already decided I wanted to see (1) a major price DROP, AND (2) quite a few reports on the SF100 before I was willing to risk anything on their design - Gridseed's JUNK board-level design work on both the orbs AND the blades was not what I'm interested in putting up with on a miner ever again.

I'm guessing Bitmain will sell it at very close to the $US equivilent to what their chinese site quites it for in yuan - and then translate that to bitcoin.

 Given that this S5+ has *6* fans vs. the single on a S5, and guessing they're probably the same fan (the pics aren't clear on that point), it should run about 7-8db louder than a single S5. Possibly a little more as the air path is a lot more crowded than on the S5, but possibly a bit less as it's fully enclosed.

It's an expression meaning something along the lines of "only a dumb bird would think this is a good idea".

 "Flying into plate glass doors is for the birds" for a particularly illustrative example.

 In my case, Naproxin (by prescription) and Ibuprofin are the only thing that have allowed me to do the job I've had the last few years.

 Depends on the situation and the drug, don't assume they ALWAYS are misused and "ruin" a life.

 9-)