[ANN] Sfards:SF100, the first 28nm Dual-Mode Miner is accepting pre-order now

[dmwardjr]

....
 if there was some way to make a modular miner and just buy chips that would be the best solution I think.

Working on it....

sweeeeeet

Only if ASIC's manufacturers will want to sell their precious ASIC's @ reasonable costs. Not 10x manufacture costs...

Understandable...

Not for everyone. I think that with my design and 20$/SFARD_chip this miner could cost 849$ (fast calculations). With easy (== cheap) upgrade in future,only one PCB to replace with screwdriver....

Sounds reasonable...  Especially, if the hash rate is similar to that of the SF100.  I like the sound of cheap and easy upgrade as well.  That's what attracts me to the Miner's Edge rigs out of Austin, Texas.  And the fact they are water cooled; except for the PSU.

[1714244617]

1714244617

[Dexter770221]

....
 if there was some way to make a modular miner and just buy chips that would be the best solution I think.

Working on it....

sweeeeeet

Only if ASIC's manufacturers will want to sell their precious ASIC's @ reasonable costs. Not 10x manufacture costs...

Understandable...

Not for everyone. I think that with my design and 20$/SFARD_chip this miner could cost 849$ (fast calculations). With easy (== cheap) upgrade in future,only one PCB to replace with screwdriver....

Sounds reasonable...  Especially, if the hash rate is similar to that of the SF100.  I like the sound of cheap and easy upgrade as well.  That's what attracts me to the Miner's Edge rigs out of Austin, Texas.  And the fact they are water cooled; except for the PSU.

My idea is to design and published as open source two universal heatsinks. One air cooled and 2nd water cooled. This heatsinks will accept universal DC/DC converter and universal controller board. Only one variable is ASIC's board. Why open source heatsinks? Becuse it's heaviest in this whole mess and transport costs are huge. You can order heatsinks in your area, boards from me, and cost are considerably lower. In future you only order one PCB and download new firmware...

[Dexter770221]

....
Definitely a lot better than being at the constant mercy of the big Asic manufacturers, if you haven't already but you should make a thread about your findings.

I'm going to, but I've lost my job few days ago and right now I'm organizing my life to survive that 2-3 months of designing. Prototype realese will be painful

[philipma1957]

....
Definitely a lot better than being at the constant mercy of the big Asic manufacturers, if you haven't already but you should make a thread about your findings.

I'm going to, but I've lost my job few days ago and right now I'm organizing my life to survive that 2-3 months of designing. Prototype realese will be painful

Sorry to hear this.  I am pulling for you  job markets tough.

I hope you can build a piece of gear worth buying from you.

@ op is this 1129 number sure?

[CartmanSPC]

I wonder if it will be possible to tune Scrypt and SHA256 separately?  I'm guessing that it's just based on the frequency of the whole chip, so probably not, but it would be neat if we could crank up the hashrate on the Scrypt side while keeping SHA256 side more power efficient if Scrypt were to be paying better...

I've never even tried to dual mine with my original five chip gridseed, so this is hearsay at best, but it was my understanding that the two modes are completely separate and can be tuned as such. Can anyone confirm or clarify this? (tempted to try)

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

[dmwardjr]

I wonder if it will be possible to tune Scrypt and SHA256 separately?  I'm guessing that it's just based on the frequency of the whole chip, so probably not, but it would be neat if we could crank up the hashrate on the Scrypt side while keeping SHA256 side more power efficient if Scrypt were to be paying better...

I've never even tried to dual mine with my original five chip gridseed, so this is hearsay at best, but it was my understanding that the two modes are completely separate and can be tuned as such. Can anyone confirm or clarify this? (tempted to try)

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

Thanks for sharing your experience with them.  As I have no experience with them.  I will probably give them a try with group buy for 2 to 4 rigs if they will approve a group buy for someone in the US.  My intent is still to have a mining operation in Washington State.  Was planning on the 4th Quarter of this year of 1st Quarter of 2016.  Now it's looking like 1st or 2nd Quarter of 2016.  Who knows, I may end up with most of my equipment being SFards.

Thanks again for sharing.

[philipma1957]

I wonder if it will be possible to tune Scrypt and SHA256 separately?  I'm guessing that it's just based on the frequency of the whole chip, so probably not, but it would be neat if we could crank up the hashrate on the Scrypt side while keeping SHA256 side more power efficient if Scrypt were to be paying better...

I've never even tried to dual mine with my original five chip gridseed, so this is hearsay at best, but it was my understanding that the two modes are completely separate and can be tuned as such. Can anyone confirm or clarify this? (tempted to try)

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

Thanks for sharing your experience with them.  As I have no experience with them.  I will probably give them a try with group buy for 2 to 4 rigs if they will approve a group buy for someone in the US.  My intent is still to have a mining operation in Washington State.  Was planning on the 4th Quarter of this year of 1st Quarter of 2016.  Now it's looking like 1st or 2nd Quarter of 2016.  Who knows, I may end up with most of my equipment being SFards.

Thanks again for sharing.

well for scrypt the blades were good gear.  starting price was 3k and they dropped to 500usd pretty fast.

plus the blades sucked so bad at 256 mining no one every did  mine them that way.

If this gear is like that people will not be happy. and no one has seen them mine 256. or scrypt in a you tube video.  be careful don't be quick.

or jump in and get a deal.  No one knows.

I tell you if J4obberwocky gets his development kit running maybe he will let us know what the chips can do.

@J4obberwocky any luck with the board?

[Rabinovitch]

Just think though. In October, they can have a sale! Just in time for Christmas......

$999!!!

Pure speculation on my part. I wonder who will be buying at $1100+?

No one. 600-700 USD is our desireable price.

[Liquid71]

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

well for scrypt the blades were good gear.  starting price was 3k and they dropped to 500usd pretty fast.

plus the blades sucked so bad at 256 mining no one every did  mine them that way.

If this gear is like that people will not be happy. and no one has seen them mine 256. or scrypt in a you tube video.  be careful don't be quick.

or jump in and get a deal.  No one knows.

I tell you if J4obberwocky gets his development kit running maybe he will let us know what the chips can do.

@J4obberwocky any luck with the board?

Yes looks like it's they are going to use the same playbook as they did for the gridseed. unfortunate but can't blame them since there will be people willing to give them their money and mine at a loss  

[J4bberwock]

I wonder if it will be possible to tune Scrypt and SHA256 separately?  I'm guessing that it's just based on the frequency of the whole chip, so probably not, but it would be neat if we could crank up the hashrate on the Scrypt side while keeping SHA256 side more power efficient if Scrypt were to be paying better...

I've never even tried to dual mine with my original five chip gridseed, so this is hearsay at best, but it was my understanding that the two modes are completely separate and can be tuned as such. Can anyone confirm or clarify this? (tempted to try)

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

Thanks for sharing your experience with them.  As I have no experience with them.  I will probably give them a try with group buy for 2 to 4 rigs if they will approve a group buy for someone in the US.  My intent is still to have a mining operation in Washington State.  Was planning on the 4th Quarter of this year of 1st Quarter of 2016.  Now it's looking like 1st or 2nd Quarter of 2016.  Who knows, I may end up with most of my equipment being SFards.

Thanks again for sharing.

well for scrypt the blades were good gear.  starting price was 3k and they dropped to 500usd pretty fast.

plus the blades sucked so bad at 256 mining no one every did  mine them that way.

If this gear is like that people will not be happy. and no one has seen them mine 256. or scrypt in a you tube video.  be careful don't be quick.

or jump in and get a deal.  No one knows.

I tell you if J4obberwocky gets his development kit running maybe he will let us know what the chips can do.

@J4obberwocky any luck with the board?

Hi, the first board is almost finished soldering, just got some trouble with a footprint for the 1.8v converter. On Mouser, it's written sot23-5 (5 pins), and when I received it, it was SOT23 A (3 pins, not the same order, not the same spacing).
I'm waiting for a few samples from the manufacturer in the good package to fit my boards, but I managed to solder the 3 pins one temporarily.

The BGA soldering wasn't as hard as expected, my oven preset temperature was just a little too low for the balls to melt and stick correctly the first time, or maybe my flux is not active enough, but everything seems ok now.
For those willing to do similar BGA soldering, positionning the chips is fairly easy. And with so many balls, self centering will probably allow at least 0.3 or 0.4mm misalignement.

I'll update with some pictures later today.

I hope to have the board finished this weekend and if there were no other mistakes in the design, or only minor ones, I'll have some numbers next week.
Dropbox will be updated today or tomorrow with the correct reference for the 1.8v regulator and 1 ground connection that was missing on another chip (ESD protection on the USB line, so not mandatory anyway, but I used a jumper wire on my board).

I've also been working on something else than the dev board. 2 or 3 users already got some info, I may give some more details in a week or two.
Anyway, no chips available in bulk quantity until late July/early August.

[alh]

Maybe I haven't been paying attention, but do you have a price on the chips yet? It would seem they might view the chips as equally precious as the miners.

[Searing]

I wonder if it will be possible to tune Scrypt and SHA256 separately?  I'm guessing that it's just based on the frequency of the whole chip, so probably not, but it would be neat if we could crank up the hashrate on the Scrypt side while keeping SHA256 side more power efficient if Scrypt were to be paying better...

I've never even tried to dual mine with my original five chip gridseed, so this is hearsay at best, but it was my understanding that the two modes are completely separate and can be tuned as such. Can anyone confirm or clarify this? (tempted to try)

I have a bit of experience dual mining with Greedseeds 5 chip product. Each side can be "tuned" individually but they do NOT work reliably in "dual" mode. Miners will slowly start dying over time requiring a power cycle to get them going again. Since I mine in a remote datacenter without the ability to easily power cycle equipment I resulted to only using the scrypt side. On scrypt they are very stable running for months without fail. I have had mine running non-stop since they first came out.

Having just read the first 40 pages of this thread I am reminded of what Gridseed did with the release of their 5 chip product. They will absolutely try to maximize profits. Expect these new miners to initially be priced at a premium. There will be a limited number available and people will buy them. Once they ramp up production pricing will start to come down quickly.

The original 5 chip Gridseed was brought to market at around $300 and quickly declined from there. I am not at all surprised at the $1,129 price being discussed here. If they do not sell enough of their initial limited quantity it is very easy for them to adjust down.

It would be interesting to try and determine their cost including development and production to get an idea how low they can go before it stops making sense to produce them.

Edit: These would have to be priced at $183.41 for them to compete with the $999 ALCHEMIST 256 Mh/s miner since power cost is not a concern for me...strictly thinking about scrypt of course.  

Thanks for sharing your experience with them.  As I have no experience with them.  I will probably give them a try with group buy for 2 to 4 rigs if they will approve a group buy for someone in the US.  My intent is still to have a mining operation in Washington State.  Was planning on the 4th Quarter of this year of 1st Quarter of 2016.  Now it's looking like 1st or 2nd Quarter of 2016.  Who knows, I may end up with most of my equipment being SFards.

Thanks again for sharing.

well for scrypt the blades were good gear.  starting price was 3k and they dropped to 500usd pretty fast.

plus the blades sucked so bad at 256 mining no one every did  mine them that way.

If this gear is like that people will not be happy. and no one has seen them mine 256. or scrypt in a you tube video.  be careful don't be quick.

or jump in and get a deal.  No one knows.

I tell you if J4obberwocky gets his development kit running maybe he will let us know what the chips can do.

@J4obberwocky any luck with the board?

Hi, the first board is almost finished soldering, just got some trouble with a footprint for the 1.8v converter. On Mouser, it's written sot23-5 (5 pins), and when I received it, it was SOT23 A (3 pins, not the same order, not the same spacing).
I'm waiting for a few samples from the manufacturer in the good package to fit my boards, but I managed to solder the 3 pins one temporarily.

The BGA soldering wasn't as hard as expected, my oven preset temperature was just a little too low for the balls to melt and stick correctly the first time, or maybe my flux is not active enough, but everything seems ok now.
For those willing to do similar BGA soldering, positionning the chips is fairly easy. And with so many balls, self centering will probably allow at least 0.3 or 0.4mm misalignement.

I'll update with some pictures later today.

I hope to have the board finished this weekend and if there were no other mistakes in the design, or only minor ones, I'll have some numbers next week.
Dropbox will be updated today or tomorrow with the correct reference for the 1.8v regulator and 1 ground connection that was missing on another chip (ESD protection on the USB line, so not mandatory anyway, but I used a jumper wire on my board).

I've also been working on something else than the dev board. 2 or 3 users already got some info, I may give some more details in a week or two.
Anyway, no chips available in bulk quantity until late July/early August.

you have  a thread on this I can follow...or are you just 'kibitzing' here and not ready for such yet?

[dogie]

4) your pool is constantly at 100% luck for the day (BIG IF).

Out of all the variables, that should be the one which remains the most stable.

[J4bberwock]

you have  a thread on this I can follow...or are you just 'kibitzing' here and not ready for such yet?

No thread at the moment, just hijacking this one since it's still under development.

[MCHouston]

Anyone using this chip for development thinking of a larger miner?  32 chips or higher per board?  I am very interested.

[philipma1957]

I can't believe they still haven't clarified it at all, anyone receive any emails who inquired? It does seem they're just going to be mining for themselves and don't really care.

It is not that they don't care.  They are most likely 10x maybe 50x smaller then bitmaintech.

So if they price this gear at a competitive  number. (600 usd shipped to the USA)

Bitmaintech would most likely crush them by bringing out the s-6  or the s-7.

  Just like bitmaintech put a hurt on sp-20 sales by pricing the s-5 at 310 usd.

This  new price war would not really help mining. Last price war ended up driving price of coins under 200 usd and coins have not done a good recovery.

It is far safer to do exactly what sfards is doing.

  But I tell you what send them an email and offer 15k in btc   for 10 machines.  I bet you could get them.

We have to play the waiting game due to one really big seller (bitmaintech)  That is just the way it is.

[dogie]

Bitmaintech would most likely crush them by bringing out the s-6  or the s-7.

S6 was skipped / replaced by S4+, next is S7.

[dmwardjr]

4) your pool is constantly at 100% luck for the day (BIG IF).

Out of all the variables, that should be the one which remains the most stable.

Can you explain why BTCGuild has fallen below 82% for a 30 day period and Slush has fallen below 90% for a 30 day period?

BTCGuild is actually below 92% for the year and below 82% for the month.  So, ANYTIME I look at bitcoinwisdom calculator to determine potential earnings for the day, week and year; I always deduct at least 8% from the estimate given by bitcoinwisdom to come up with a more conservative estimate of earnings.  EVERYONE should do this.

I switched over to BTCChina, who remains over 93% for the year.  However, they always showed 1 TH/s less than what I actually have.  THEN, I switched over to AntPool.  They actually remain consistent with my actual hash rate AND remain above 92% majority of the time.  Here lately, it has been rare for a pool to be 100% or higher for the month; much less an entire year.

So, in my opinion, one should be careful about assuming a pool will remain 100% or higher for an entire year.  A few pools MIGHT achieve this but not very many.  This means one should always deduct approximately 7% to 8% of their projected earnings for a rig [To be conservative] when determining ROI.  My honest opinion...

[philipma1957]

Bitmaintech would most likely crush them by bringing out the s-6  or the s-7.

S6 was skipped / replaced by S4+, next is S7.

yeah i think you are correct about that. 

well maybe the s-7 drops late July.

And maybe my next  two miners are an sfards and an s-7  time will tell.

[Gladimor]

Any clue as to how much the miner will cost?