In the README for cgminer it mentions the "--usb" flag.

100KW datacenter on 208V. Works just fine.

I've got a bunch of pulled fans from S5 and S7 if you're looking for a stock replacement. If you're trying to make a miner quieter, good luck.

Okay well it's recognizing it then. Do you have a means of gauging how much power it draws from the USB port when it's plugged in? Should pull 1.8W to start; if it's only drawing ~30mA there's an issue with the main regulator, and ~50mA it's probably a string reset issue - which has a variety of potential causes.

The only time I see "failed usb_init" is before I put the product string and serial number on the CP2102, or if a stick is plugged in at the wrong time in the detection cycle (but it's then properly detected on the next one).

No, because then I'd have to spend a couple days testing each one to see how well it works all the way up, and that's nonsensical for a $33 stick. And I don't really care, all I guarantee is 100MHz stable and overclocking is at your own risk. It's always been like that.


Check in device manager when you plug in the stick. What's it read as? Make sure the voltage isn't cranked way up. Turn the pot so the flat side is at about 2 o'clock position. Unless you have a way to measure it, that would be best; set it to 1.26V.

Yeah but the more people who ask the same stupid questions about a device this thread in no way covers, the deeper the quote nest gets. And that lets people see just how many times the same stupid questions have been asked, and maybe make them consider not doing it again and maybe just maybe following the advice instead.

Honestly though, answering the guy's question properly requires first a discussion of how command-line operation actually works (an education today's touchscreen kids are sadly lacking), then a bit of time on cgminer fundamentals, and then he'll be almost ready to actually use the answer to his specific question.

But yeah the first place to start, if you have a question about 2Pacs, is in the 2Pac thread. For the hundredth time. Dangit.

Whole lot of "if" in there though. A company that claims to be a couple years old with a brand-new website, and asking for several metric buttloads of preorder money to build a machine they so far have exactly one of - in a black-box room somewhere it can't be photographed - with their first chip being better than what guys who've been doing it for five years have been able to generate - and everything stuffed into a nice easily-photoshoppable case of illogical form-factor and using plenty of attractive buzzwords in the description. Just saying, there's a lot of "if". Way more "if" than really merits the attention these guys are getting.

Could be a good home miner, if it wasn't prohibitively expensive by merit of requiring an entire motherboard to attach to. Which depending on what that motherboard is doing, could require substantial engineering overhead.

Changing your 12V won't do anything, until you get your 12V rail below the output voltage of the main regulator. At that point it'll either start to freak out, shut down, or best case hit 100% duty cycle and push your rail straight through. But it's very likely you don't have a power supply both that adjustable and that powerful.

The circuit looks more like what's on an S9, which I haven't spent enough time around to know how to mess with. By merit of they're still effing expensive. If I had a chip-dead donor board with a good regulator I could at least try to play with it, but I'd also have to have time. In addition to manufacturing and my own miner R&D and other projects I just got volunteered to help a friend build a custom 3D printer for his lab work, so free hours are scarce.

It's likely possible to hack the PIC and make the thing volt-adjustable like an S7, but the circuit's not the same and the mechanism won't be the same either.

(except that designs based around a single CPU-like chip have almost universally failed due to reliability issues resulting from inefficient high-current low-voltage regulators and exotic cooling requirements needed to clear dense power quickly; see Spondoolies, Hashfast, BlackArrow, Cointerra and I don't know how many others for reference)

A lot of other companies killed themselves one way or another - unsustainable complexities, being outright scams from the start or turning into one.

Also there's no real evidence that Syan is anything other than another scam.

Also Avalon's been around longer than Bitmain. The Avalon team put up the second functional bitcoin mining ASIC, and the first one available to the general public - if I'm remembering right, ASICMiner's BE100 predated it but were only deployed in their own mines for some months. Butterfly Labs announced development first, but didn't release a product until almost a year behind schedule. Avalon kinda pulled the dine-and-dash with a lot of people's money sometime around their third-generation chip release, but from what I understand, the engineers behind the actual work rebooted the company as Canaan-Creative once the right bastards in charge had disappeared, and carried on the Avalon product name.

And let's not forget Bitfury, who doesn't make small-scale miners but has released historically some of the best chips available at any given time. They had sub-1J/GH ASICs when Bitmain's 2J/GH BM1380 was rolling out on the S1 and obsoleting everyone's 6-10J/GH gear. They put up ~0.2J/GH chips when the S5's 0.4J/GH were being called top-class, and they've had 0.1J/GH 16nm for about as long as the S9 has been a thing.

And let's not forget BW, which I believe succeeded LKETC who grew strong around the A1 Dragon machines, 1J/GH competitors contemporary with the S2 and historically one of the most versatile and reliable large-scale miners yet made. Those guys deployed 14nm chips I believe before Bitmain announced the S9, though they kept them to themselves and other Chinese mines. Heck they were on their third revision of 14nm design a full year ago.

And let's not forget EBang, who's been selling an Avalon741-range miner on the open market for most of this year using their own ASIC design.

Bitmain may want to rule the world, and right now they do have efficiency in their corner. They also have a whole lot of visibility because they have wide open low-volume sales. But they also have crap customer service, crap warranties, crap machine-level engineering and a whole lot of profiteering shenanigans that folks are starting to realize aren't exactly worth the trouble.

Whether or not autotune miners can be changed depends on how the controller code would react to failing to set the board-level voltage. Failing to get any response out of the PIC at all.

Honestly I have no idea about the D3. I've never seen one, and I'm certainly not going to buy one just to mess with. I know this S7 hack doesn't work on the S9, as its regulator control system is completely different.

Consider using the search function. I'm not sure if it's possible to do exactly what you want, but you can limit the number of devices a particular instance will enumerate. I think how to do that has been explained in the last week or so, and also with relative frequency over the life of this thread.

Yeah but if you buy S9s you'll have a couple months to get your building ready before they arrive.

Running 20 off one controller, if your controller bricks you've lost your whole mine. If you run 10 each off two controllers, if one bricks you can put them all on the working controller and keep hashing while fetching a replacement. Something something "eggs in one basket".

I have some sticks where the ASIC performance is unbalanced enough I need to install a parallel resistor around one chip to get enough idle current for the node voltages to balance out and allow a stable startup when the reset RC goes high. A lot of those end up being "factory seconds" though - the chips are unbalanced enough that one'll choke out the other and go "zombie" at what should be a stable running voltage, so to work properly the whole thing has to be turned up past stock specs.

That's one of the issues with a single-wide string. The more chips you have in parallel on a single node, the more stable your string will be due to averaging. Node voltages will tend to average themselves out as weaker chips are randomly paired with stronger chips.

There's no mechanism for zombie recovery on the 2Pac. The string reset is a hardware RC delay that waits a fixed amount of time after the stick is plugged in and then enables the ASICs. The only way to re-reset is to unplug and replug, unless you know exactly what parts to short around on the stick. The original Compac had a reset testpad on the back which could be used for this, but the 2Pac is a bit denser and also has a heatsink on the back so I did away with those testpads.

The new pod miner in development has auto-recovery hardware that'll detect a string lockup (zombie condition) and power-cycle the string without manual intervention. The software to make this work properly is still being ironed out. It's a bit much for a stickminer though, but we'll see. Maybe on the next one.

(Nobody's answering those questions because this is a year-dead thread about something else entirely)

Yeah, Avalon was selling chips to third-party integrators for a while. In 2013 and 2014. If you can get ASICs, integration isn't terribly difficult. But getting the ASICs is.

Yeah but he's asking about building a modern miner, and mentioned bitcoins, and using "scrap computer parts" for bitcoin mining kinda stopped being "modern" approximately five years ago.

Because local power generation is a long-term investment and bitcoin mining farms are operated by short-term-profits people. They're set up in places where low electric rates already exist.