It means that the entire logical and physical layouts are done from scratch without using pre-made IP blocks. Most semiconductor fabs have built up libraries of various logical functions to use in copy-paste design.

The problem with using pre-made or 3rd-party IP blocks is that they are optimized to perform their specific function but not necessarily optimal when it come to working with other sections. Even worse is that they are stand alone islands that must be connected to each other and that leads to signal routing problems and performance hits from excessively long connections between the blocks.

With full custom everything is done from scratch using pre-made IP as perhaps a guide to what needs to be done in each block but that is it. Given Bitmain's experience with their previous chips they know what the logical structures need to be, now it was just a matter of translating those structures into optimized pathways on silicon. All physical layout of the chips is done with point-1 being shortest possible connections inside of and between the various bits like memory, math processors, coms, etc. Shorten the internal connections and switching losses drop like a stone allowing faster speeds and/or lower power.

Get over it.
By and large, the '14nm' tag has become a marketing tool much like IoT. There are few devices that can benefit from going there (mostly phones, tablets, etc.) and there are devices that just-don't-need-it. As I and others here have repeatedly said, node size is NOT a major factor for mining ASICs. Bitmains new chip just proved that existing mature and well understood node sizes such as 28 and even 20nm have lots of room for performance improvement vs using 16-14nm with its inherent risks, mask costs and poor chip yields.

Of course you can. Just wire the switch(s) in series with the PSU's control pins and your external on/off switch. If you are using the GekkoScience breakouts for IBM 2 or 2.8kw PSU's the control switch will select either local (just the switch) or external control. Use the external setting and wire as needed to toggle the on/off.

Since the GekkoScience look for a (+5) voltage source to turn on you would the thermal switch between the +5 output and the remote on/off pin. Now ya just have to make sure you always have the breakout board switch set to External.

On other breakouts for the HP1200w PSU/s wire the thermal switch in series with the connection to pin-4 that turns on the supply.

Agreed ^^
While glad to see someone who does realize that equipment (of any sort) can need care and at least monitoring, miners are pretty much hands off until trouble. Aside from below a PM schedule is not needed. Hashing issues seem to be pretty random and almost always solved by a hard reboot. I have miners that have never needed a reboot in over a year to some that seem to prefer a reboot every month.

 The only thing that should be checked from time to time is that the fans are in good shape (bearings known to corrode/fail) and the heatsinks are reasonably clean. Same applies to the PSU's used.

Please folks -- ears on rack mount chassis are NOT the only thing you use to mount them! Or better not be.
All heavy rack mount gear is supposed to be sitting on a support shelf/brackets. The ears are mainly only to keep them in position.

If you mean as an active cooling device to pull heat from the chips or heatsinks -- sure they would work. But - all you are doing moving the heat problem from the miner assembly to the thermoelectric coolers. Now you have to keep their heatsinks clean and reasonably cool. Not to mention you now take an efficiency hit from having to power them in addition to the miners.

Since you said you connected a cable direct miner to PC, you need a crossover cable for them to be able to talk to each other. Or run with a switch/router in between. Problem is that the Tx and Rx wires need to be switched so one is sending to the others receiving wires.

No idea why some folks have having trouble with the latest FW release... I've so far done 15 of my S5's with zero problems. Hash board revs range from early Batch-1 up through last new Batch.

The one problem I * have * ran into is several of the latest S5's even with the older as-shipped version of firmware is that M's Miner Monitor can't reboot them via ssh: "incorrect password'. Ok -- so what is the new one the miners ship with?

So far I have 12 out of 15 S5's upgraded to the latest firmware and zero issues.

On the fans speed % I have a hunch it is the percent of the fans rated rpm range - not 0 to max rpm. Yes, PWM fans DO recognize a zero speed command but next notch up is usually around 1/2 their max rpm rating and up from there.

Dinna know that it was part of OpenWRT. Figured it was open-source bit they tacked on.

As for bandwidth... was thinking more of getting hard data for if I move my 30TH farm at work to using a 4G HotSpot. Being in an industrial park the available internet solutions are limited mainly to Multiple T1 lines and they are pricey as well as shared with VOIP for the phones. That farm puts a sizable dent in the b/w we have at work...

Extrapolating from the S3 data I have I puts total monthly usage at a ballpark value of 8GB/month. How well does bandwidth usage scale with TH/s from the S3's across the S4 & s5 miners?

The one request I have for Bitmain is put the network usage graphing back in like the S1 and S3's had! Since it is pretty much part and parcel with the Pi's and BeagleBone OS/s anyway please bring it back. Would be nice to see what the bw of these will be (mainly for planning G3/G4 WiFi Hotspot usage).

Tasty... Me likey

It just screams using a pair of the IBM 2kw supplies in load share which is easy-peasy using the breakouts from GekkoScience. Since I still just happen to still a couple  of each sitting idle and a ton of leads I'll be all set for one.

Only problem is will have to retire some S3's though as I'm at my free power limit of 15kw at work

Oops, silly me missed the date on that...

mid April? as in next year? ? ? ? oh yeah that sounds like a winner...  

Most states and large metros area have several, just check the phone book or do a search for electronics recycling and the state/city. Definitely needs to be close to you to avoid shipping costs!

Wish I could but I do not have the time for it nor unless seriously underclocked would they be viable to run anymore unless ya have free or exceedingly cheap electric <cough> If I had that time I'd first undervolt/underclock all my S1's and bring them back online

To get any scrap value, take them to a place that specializes in scrapping electronics. They may not offer much but it should be better than a general scrap company would give.

re the testbed: the one card that survived more than a couple weeks continued to run at 170 GH/s for almost 6 months before giving up the ghost.

The miners finally sent weren't from Ltek? I could swear the GUI has their name on it...

As for Android - no it is not Linux. AFAIK it is more closely related to JAVA. Nor is Android open source. Yes it's OS code is freely available but if used in a commercial device there are (minimal to zero cost) license fees to be paid and you are not allowed to make unauthorized changes to the OS.

Again, good to hear you giving a more honest sounding appraisal of what went down. +1 for that

EDIT: By Jove ja I was wrong - the company name on the AMT miner GUI is Stetc not Ltek.

Ja, ^^ is a perfect example of how to not cool power chips.

ya <munch munch>
Yes they were and despite trying to help them gratis, the Ltek A1 miner they finally sent was also sans PSU though they did send one for it a week later. I do have to say that the miner has been running 24x7 for almost a year now with zero issues aside from being hungry. Shows what a proper design can do (and I have free power)

Not exactly supporting Josh about his post on his and AMT's role in things but good to finally hear a bit more honest explanation/tale trickling out over what transpired between AMT and Bitmine. As I had been saying since getting into the A1 mess by buying a miner from AMT and then digging into all I could find about the A1 and Bitmine and knowing that AMT's miners were re-branded Bitmine ones the fact that Bitmain was not fulfilling their own general-public customers did not bode well for those of us who were AMT customers...

I get that their major customers - including themselves (Bitmine) via their partnership in mines - get the bulk of production in a sort of 1st-come-1st served BUT if they had a viable and stable design at least a trickle of Consumer rigs would have made it out the door to placate the clamoring hoards. At least the peta-farms they were supplying had to eat the dog food they made

Even my few attempts to get some specifics from Bitmine via Josh on some basic questions I had were -- not promising -- with Bitmine not exactly responding. In retrospect Josh's post is just backup to what I was seeing.

Folks, a reference design PCB & circuit is just that - for reference to test how a chip should work. It is NOT a guide to making a real-world product. That is up to the end-user of the chip to figure out using the reference to guide you in how your final design should operate. Sometimes a reference board/circuit is optimized to show Best Practice for the final design, most often they are not. Now if Inno was supposed to also supply a working Production system or hash board module design, that's another story and is between Bitmine and Innosilicon to work out.

Part of Bitmine's problem was that they copied the PDN (power distribution network) from the 2-chip board and just figured, 'lets make it bigger to fit more chips per-chain". Doesn't work that way as an ASIC has a very 'spikey' current draw and those spikes happen at the clock freq - 10's to 100's of mHz. The PDN has to be designed like a radio circuit or all sorts of havoc ensues. Make a major change as Bitmine did and well... That goes in line with Inno's comment about the DCDC margins as well.

In this case the A1 Dev thread here covered many aspects of what killed the Bitmine design and yet Bitmine ignored them. Possibly because they and the WASP project were pushing for open-source DIY miner designs?

Yes, I am. Tried to post a ticket on Bitmain's support site - no go. Ya hit the Create button and nada. Also posted here a page or two back. Problem is that one of the 4 new s5's I got a week or so ago only starts one card.

Too bad support seems to still be on Holiday as having ran Ants from the S1's on up of my many many Ants this is the first that has a problem from the git-go...

Got the US warranty support email addy for me to try?

I recently purchased 4 more s5 Antminers on order  00120150623154707742q5Yib6T90686.
Last week I began adding them to my farm and one miner has only 1 has board running. The GUI, pool stats and temperature of the heatsinks (1 is cold) confirm this.

As with most of my miners the s5 is fed from a HP 1200w server PSU. I tried swapping the power leads to the miner and even a different PSU with same results: Only the one same board runs.

Any further checks I need to do before requesting a warranty service RMA?
Cheers!