GekkoScience BM1384 Project Development Discussion

[TheRealSteve]

Probably having a senior senility (brain fart) moment, but, what hash chip(s) does Innosilicon make that are appropriate to Bitcoin hashing?

They made the A1 Booster, aka Bitmine's CoinCraft A1 - and the A3 is likely to be what the LK Group will be using as their 14nm solution.
(you can read about the drama between those two+ companies, if you're bored; later posts turns into a mess about mostly unrelated things)

[PlanetCrypto]

Probably having a senior senility (brain fart) moment, but, what hash chip(s) does Innosilicon make that are appropriate to Bitcoin hashing?

They made the A1 Booster, aka Bitmine's CoinCraft A1 - and the A3 is likely to be what the LK Group will be using as their 14nm solution.
(you can read about the drama between those two+ companies, if you're bored; later posts turns into a mess about mostly unrelated things)

Where can I find a datasheet/specs (prelim or otherwise) on the A3?

[TheRealSteve]

Where can I find a datasheet/specs (prelim or otherwise) on the A3?

I guess you'd have to ask them - beyond their claim of tape-out and "over 60% power saving over [A1]", I don't recall seeing any solid public info from them.  Presuming it is the chip LK Group would be using, they lay claim to a future 5Th/s, 1kW miner.  BW.com claims 0.26W/Gh. Either would fall under preliminary at best.

[QuintLeo]

Probably having a senior senility (brain fart) moment, but, what hash chip(s) does Innosilicon make that are appropriate to Bitcoin hashing?

They made the A1 Booster, aka Bitmine's CoinCraft A1 - and the A3 is likely to be what the LK Group will be using as their 14nm solution.
(you can read about the drama between those two+ companies, if you're bored; later posts turns into a mess about mostly unrelated things)

 The drama was between BitMine and Innosilicon, Lketc wasn't involved as such.

[TheRealSteve]

The drama was between BitMine and Innosilicon, Lketc wasn't involved as such.

Ah, glitch of an early edit - that was supposed to refer to just Innosilicon and Bitmine (the + referring to AMT)

[sidehack]

I've been meaning to look into what they're doing for chips. The A1 was about twice the power that I'd like to see from a single chip but they were definitely turned into some pretty good gear.

[PlanetCrypto]

Where can I find a datasheet/specs (prelim or otherwise) on the A3?

I guess you'd have to ask them - beyond their claim of tape-out and "over 60% power saving over [A1]", I don't recall seeing any solid public info from them.  Presuming it is the chip LK Group would be using, they lay claim to a future 5Th/s, 1kW miner.  BW.com claims 0.26W/Gh. Either would fall under preliminary at best.

I asked.

sidehack - didja' get my cc?

[PlanetCrypto]

Where can I find a datasheet/specs (prelim or otherwise) on the A3?

I guess you'd have to ask them - beyond their claim of tape-out and "over 60% power saving over [A1]", I don't recall seeing any solid public info from them.  Presuming it is the chip LK Group would be using, they lay claim to a future 5Th/s, 1kW miner.  BW.com claims 0.26W/Gh. Either would fall under preliminary at best.

Thanks Steve.

[sidehack]

Yes, yes I did. Lemme know what you find out.

[PlanetCrypto]

sidehack - Bcc'ed you and PM'ed you.

Heading to bed, am F'ing beat.

[sidehack]

I received both and replied. No offense, the various gentlemen which alert me in this thread to messages in other places, but everythings ends up funneled through email so I see it all at once and it's kindof a waste of time to get three messages about the same thing at the same time from the same person in the same place.

On the BM1384 Project Dev front, I just got an email from Bitmain saying there were no details yet about new chips. Not terribly surprising. There's background chatter with InnoSilicon and Canaan-Creative and hopefully some of that will turn up info. I'm taking care of all the manufacturing tasks at the shop for at least the next week or two to give Novak time to focus on the three different projects he needs to wrap up before month's end, and when I get stock caught back up I'll get back to board dev. Next step is testing some new level-shifting I sketched up a while back, and then getting a barebones 18-chip board functional (off an external supply, probably the hacked TPS53355 circuit). It'll be run off an S5 controller until Novak's got multi-chip code of our own figured out, but that likely won't be for a few weeks. I'll be ironing out some buck designs and working on peripheral controls (fan, temp, voltage etc) for a basic pod miner board, which would translate directly to larger boards.

I also need to codify and post for discussion our ideas for a more-or-less standardized modular rack-scale miner. It'd be something we intend to build boards for, and intend to be flexible enough that others could just as easily build boards for it, and ideally that different boards from different people, even using different chips and topologies, could all be run together with cgminer support being the only real limitation to worry about.

[philipma1957]

I received both and replied. No offense, the various gentlemen which alert me in this thread to messages in other places, but everythings ends up funneled through email so I see it all at once and it's kindof a waste of time to get three messages about the same thing at the same time from the same person in the same place.

On the BM1384 Project Dev front, I just got an email from Bitmain saying there were no details yet about new chips. Not terribly surprising. There's background chatter with InnoSilicon and Canaan-Creative and hopefully some of that will turn up info. I'm taking care of all the manufacturing tasks at the shop for at least the next week or two to give Novak time to focus on the three different projects he needs to wrap up before month's end, and when I get stock caught back up I'll get back to board dev. Next step is testing some new level-shifting I sketched up a while back, and then getting a barebones 18-chip board functional (off an external supply, probably the hacked TPS53355 circuit). It'll be run off an S5 controller until Novak's got multi-chip code of our own figured out, but that likely won't be for a few weeks. I'll be ironing out some buck designs and working on peripheral controls (fan, temp, voltage etc) for a basic pod miner board, which would translate directly to larger boards.

I also need to codify and post for discussion our ideas for a more-or-less standardized modular rack-scale miner. It'd be something we intend to build boards for, and intend to be flexible enough that others could just as easily build boards for it, and ideally that different boards from different people, even using different chips and topologies, could all be run together with cgminer support being the only real limitation to worry about.

sounds like good news.

[sidehack]

Also, I just noticed that one of the four sticks I had running at 275 (15.1GH) pulled a 208M share sometime in the last few days, which puts them, by highest share, well ahead of the S4+ that's been running since Saturday (46M).

[Mikestang]

Also, I just noticed that one of the four sticks I had running at 275 (15.1GH) pulled a 208M share sometime in the last few days, which puts them, by highest share, well ahead of the S4+ that's been running since Saturday (46M).

I've got a few nanofury usb sticks and every now and then they'll pop a share that beats my S3+, but it's pretty rare.

[Meech]

I've held out but need to get my hands on a few for fun.  Flashy lights and all.  Keep it all rollin'.

[chiguireitor]

Also, I just noticed that one of the four sticks I had running at 275 (15.1GH) pulled a 208M share sometime in the last few days, which puts them, by highest share, well ahead of the S4+ that's been running since Saturday (46M).

Interesting would be to have several of these sticks pointed to http://solo.ckpool.org/ and see what happens

[philipma1957]

Also, I just noticed that one of the four sticks I had running at 275 (15.1GH) pulled a 208M share sometime in the last few days, which puts them, by highest share, well ahead of the S4+ that's been running since Saturday (46M).

Interesting would be to have several of these sticks pointed to http://solo.ckpool.org/ and see what happens

two sidehack and 1 u-2

http://solo.ckpool.org/users/1JdC6Xg3ajT3rge3FgPNSYYFpmf53Vbtje


"bestshare": 255,764,049.1084682

[Mikestang]

Interesting would be to have several of these sticks pointed to http://solo.ckpool.org/ and see what happens

That's where mine will be pointed, just as soon as I can find the pennies in my couch to buy a couple.  

[chiguireitor]

...

two sidehack and 1 u-2

http://solo.ckpool.org/users/1JdC6Xg3ajT3rge3FgPNSYYFpmf53Vbtje


"bestshare": 255,764,049.1084682

Nice... be sure to tell us all if you happen to hit a block

[AJRGale]

Also, I just noticed that one of the four sticks I had running at 275 (15.1GH) pulled a 208M share sometime in the last few days, which puts them, by highest share, well ahead of the S4+ that's been running since Saturday (46M).

Interesting would be to have several of these sticks pointed to http://solo.ckpool.org/ and see what happens

two sidehack and 1 u-2

http://solo.ckpool.org/users/1JdC6Xg3ajT3rge3FgPNSYYFpmf53Vbtje


"bestshare": 255,764,049.1084682

jee, and my U3 has only gotten "bestshare": 2171759.8444529232
i remember pulling ~170M on kano.is pool with 2x New-r-boxs