GekkoScience is now dabbling with 16nm ASICs for new designs

[sidehack]

Klintay here actually is the one who provided the datasheet for that hub chip. Thanks again.

It's not a bad idea to keep a stock of bricks. That option will make things easier for the customer.

[1713528215]

1713528215

[takagari]

Looking forward to these usb hubs.
I only ordered two sticks for supoort. but still need to find a hub for them for when they arrive

[sidehack]

Got a question that's actually TOTALLY related to 16nm development.

One of the reasons I've been thinking from the get-go about using a microcontroller to interface to the Bitfury chips is their communication setup. The chips use a specific internally AC-coupled 2-wire communication to a central level-shifting multiplexer chip. This guy takes in standard 5V SPI through a particular IO port, which shouldn't be difficult. However, there's a specific non-SPI reset signal to initialize all your chips that needs to be fed in this port. So I figured, let's put in a micro to relay data from USB to SPI, add a command to enact that specific signal, and then use the micro to also implement temp sensing and voltage control. I could use a basic USB-SPI converter and scrub the active controls (like the Compac and 2Pac, give it manual voltage and whatnot), but there's no way to implement that reset signal.

Except I may have just figured out a remarkably simple way to do it, which will also be really easy to test and requires only a few components that I already have in bulk. In fact, all of them are already in use on the Compac and 2Pac.

So, here before long I'm going to rig up a hardware test and VH is going to see if he can make it work. If he can, we'll have the approximate guts of a 2-chip BF16 Compac implementing manual voltage control. I'll also wire fixed thermostatic control onto it, which is to say there'll be a temp sensor that turns off the Vcore power if the stick overheats and turns it back on when the stick cools down. I should have done that with the 2Pac, really, but oh well.

So, the plan has been to develop a BF16 Compac with temperature sensing and software voltage control. But firmware development is going to be the part of design that takes the longest and has the most hiccups. Therefore, who would be offended if I tried to bring to market a BF16 Compac with BM1384-Compac-style manual voltage control if it meant we could get it done that much sooner?

[olseh]

Definitely wouldn't be offended. Would love something bigger but I'd end up buying a couple anyway I'm sure

[sidehack]

"Something bigger" wouldn't be too far behind it. Hopefully I'll have a BM1384 form-factor prototype of something U3-scale in a couple weeks to show off which I'd like to make a Bitfury version of as well.

Also I hadn't talked to Punin in a while so I emailed him last night and already got a confirmation that chips are still available for this project, any quantity.

Depending on what space I have available on a manual-voltage BF Compac board, I may try to shift things around a bit to make room for some labeling to the pot so you can get at least an approximation of what the voltage is set at. Since the chips have require almost no support circuitry there should be a decent amount of space even with an extra comm chip.

The Amita pod and TypeZero refit boards would definitely have software voltage control and a full sensor package, don't worry about that.

[philipma1957]

"Something bigger" wouldn't be too far behind it. Hopefully I'll have a BM1384 form-factor prototype of something U3-scale in a couple weeks to show off which I'd like to make a Bitfury version of as well.

Also I hadn't talked to Punin in a while so I emailed him last night and already got a confirmation that chips are still available for this project, any quantity.

Depending on what space I have available on a manual-voltage BF Compac board, I may try to shift things around a bit to make room for some labeling to the pot so you can get at least an approximation of what the voltage is set at. Since the chips have require almost no support circuitry there should be a decent amount of space even with an extra comm chip.

The Amita pod and TypeZero refit boards would definitely have software voltage control and a full sensor package, don't worry about that.

A bit fury one chip stick would be really nice.
A two chip would also be nice.

Questions would a one chip bit fury do 50 gh at 5 watts?
I would
Like a rough estimate of a one chip bit fury stick

I think I would start a new sidehack USB solo mining club if we had bit fury sticks

[sidehack]

There will not be a 1-chip stick. The buck on the Compac has as low a baseline voltage as I can find (most bucks operate 0.6 to 0.8 volts minimum, that one will do 0.5) but the Bitfury's full range is from about 0.45 to 0.55 volts per ASIC. The "expected" power draw from the chip is also below 7 watts. It makes more sense to put two chips in series like the 2Pac, which shifts the buck's range away from impossible/suboptimal to 0.9-1.1 volts (this change increases overall system efficiency in three different ways) and puts the peak operating power at about 15W, which a lot of Compac and 2Pac owners like to approach.

The expected hash/power of a 2-chip BF16 Compac is from approximately 30GH at 2.5W up to approximately 130GH at 15W. These are approximate numbers based on expected performance of the chip gathered from others' test data. Once I have a working prototype the first thing I'll do - well, the first thing I'll do is post pictures and show it off but the second thing I'll do is work up a performance chart.

[philipma1957]

There will not be a 1-chip stick. The buck on the Compac has as low a baseline voltage as I can find (most bucks operate 0.6 to 0.8 volts minimum, that one will do 0.5) but the Bitfury's full range is from about 0.45 to 0.55 volts per ASIC. The "expected" power draw from the chip is also below 7 watts. It makes more sense to put two chips in series like the 2Pac, which shifts the buck's range away from impossible/suboptimal to 0.9-1.1 volts (this change increases overall system efficiency in three different ways) and puts the peak operating power at about 15W, which a lot of Compac and 2Pac owners like to approach.

The expected hash/power of a 2-chip BF16 Compac is from approximately 30GH at 2.5W up to approximately 130GH at 15W. These are approximate numbers based on expected performance of the chip gathered from others' test data. Once I have a working prototype the first thing I'll do - well, the first thing I'll do is post pictures and show it off but the second thing I'll do is work up a performance chart.

they would be good for a solo pool club.

If you get them done I would set the club up.

1 stick minimum to join the club.

I will go back and check rules of the last club,  Been a while.

if we had 50 people using 1 stick = 5000gh   that is  about 1 in 16 chance at a block in the next month.

if we had 50 people using 2 sticks = 10000 gh that is  about a 1 in 8 chance in a block in the next month.

[qctechno]

There will not be a 1-chip stick. The buck on the Compac has as low a baseline voltage as I can find (most bucks operate 0.6 to 0.8 volts minimum, that one will do 0.5) but the Bitfury's full range is from about 0.45 to 0.55 volts per ASIC. The "expected" power draw from the chip is also below 7 watts. It makes more sense to put two chips in series like the 2Pac, which shifts the buck's range away from impossible/suboptimal to 0.9-1.1 volts (this change increases overall system efficiency in three different ways) and puts the peak operating power at about 15W, which a lot of Compac and 2Pac owners like to approach.

The expected hash/power of a 2-chip BF16 Compac is from approximately 30GH at 2.5W up to approximately 130GH at 15W. These are approximate numbers based on expected performance of the chip gathered from others' test data. Once I have a working prototype the first thing I'll do - well, the first thing I'll do is post pictures and show it off but the second thing I'll do is work up a performance chart.

Can't wait for BF sticks 

they would be good for a solo pool club.

If you get them done I would set the club up.

1 stick minimum to join the club.

I will go back and check rules of the last club,  Been a while.

if we had 50 people using 1 stick = 5000gh   that is  about 1 in 16 chance at a block in the next month.

if we had 50 people using 2 sticks = 10000 gh that is  about a 1 in 8 chance in a block in the next month.

Sound like lot's of fun, count me in. 4x 2Pac

[aarons6]

There will not be a 1-chip stick. The buck on the Compac has as low a baseline voltage as I can find (most bucks operate 0.6 to 0.8 volts minimum, that one will do 0.5) but the Bitfury's full range is from about 0.45 to 0.55 volts per ASIC. The "expected" power draw from the chip is also below 7 watts. It makes more sense to put two chips in series like the 2Pac, which shifts the buck's range away from impossible/suboptimal to 0.9-1.1 volts (this change increases overall system efficiency in three different ways) and puts the peak operating power at about 15W, which a lot of Compac and 2Pac owners like to approach.

The expected hash/power of a 2-chip BF16 Compac is from approximately 30GH at 2.5W up to approximately 130GH at 15W. These are approximate numbers based on expected performance of the chip gathered from others' test data. Once I have a working prototype the first thing I'll do - well, the first thing I'll do is post pictures and show it off but the second thing I'll do is work up a performance chart.

depending on how much they cost i definitely want one or two of these bf16 compacs.
i have 5 regular 1st gen compacs now, these bf16 ones should replace all 5 of them and be about the same speed? about 60 ghs with minor oc?

[sidehack]

Very minor. Should see 60GH from about 6-7W, the 16.5GH setpoint of the original Compac.

[philipma1957]

Very minor. Should see 60GH from about 6-7W, the 16.5GH setpoint of the original Compac.

yeah  those effienices  would make a solo pool club worth doing .

Gives an okay chance of roi if we hit a block

[leowonderful]

Very minor. Should see 60GH from about 6-7W, the 16.5GH setpoint of the original Compac.

yeah  those effienices  would make a solo pool club worth doing .

Gives an okay chance of roi if we hit a block

It'd also be insane if we hit a block with a pool made up solely of USBs can't wait for the OCing efficiency/hash rate chart, interested to see how far we can push these things. I'm in for two sticks and one or two pods or more depending on the performance of these once I actually get my hands on em

[philipma1957]

Very minor. Should see 60GH from about 6-7W, the 16.5GH setpoint of the original Compac.

yeah  those effienices  would make a solo pool club worth doing .

Gives an okay chance of roi if we hit a block

It'd also be insane if we hit a block with a pool made up solely of USBs can't wait for the OCing efficiency/hash rate chart, interested to see how far we can push these things. I'm in for two sticks and one or two pods or more depending on the performance of these once I actually get my hands on em

I would want at least 5 of them.

[sidehack]

Just throwing this out there, but a good way to grease the dev wheels is to buy stuff so I can continue to afford dev materials. Right now I'm out of pocket on the USB hub and the pod formfactor I'll show off if I can get it working. I should have everything on hand for a rough proof-of-concept of the BF stick, and almost everything for the pod, but the hub's a lot of parts I've never stocked and a large (therefore expensive) PCB.

2Pac sales are always appreciated (including the factory-seconds black sticks), and I've also got a buttload of DPS1200 breakout boards and cables for anyone needing a solid PSU.

There's also the 2PacLoan program.

[sidehack]

Good news on the dev front. I just sent off for some parts which will start grinding on the second batch of 2Pacs, and also a bunch of stuff for USB hubs and the new pod formfactor. I also got my hub chips in today, as well as stencils for both the hub and new pod.

In better news, I breadboarded a communication setup for BF16 stuff today and it appears to work, which solves one of the delays there. I'm going to try and assemble a BF16 single-chip test board tomorrow and wire it up for VH to see if he can get it hashing. If he can get it going, I might be ordering prototype PCBs of a BF16 Compac (well, 2Pac) as early as the end of next week.

After the end of next week I'm taking a week's vacation so there won't be any updates. Wasn't my idea, I'd rather keep doing stuff but oh well.

[kipper01]

Vacation is good I just got back from Spring training in Goodyear, AZ.

Put me down for 2 hubs!  MY 10 port hub from Klintay is down to 2 ports.  The other 8 went dead for some reason.  Was a good hub while it lasted.

[sidehack]

My 49-port from him is still chugging along, testing 2Pacs full time. Yesterday my hot air station went belly-up though, that wasn't fun. Gonna slow down progress until the replacement gets here. I had a bunch of spares, but the ones issued to various minions over the years were kept by said minions when they left. Oh well, got 4 years of hard labor out of it.

[Morguk]

Excited for any form of pod!

[gt_addict]

Excited for any form of pod!

Same here. Looking forward to a few BF pods to play with.