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

[brontosaurus]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

For those of you unfamiliar with scrypt physical implementation please read this excellent article by Alpha Technology:

https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCEQFjAA&url=https%3A%2F%2Falpha-t.net%2Fwp-content%2Fuploads%2F2013%2F11%2FAlpha-Technology-Scrypt-Analysis-on-FPGA-proof-of-concept.pdf&ei=ljMXVarbNJHaaIH-gZAB&usg=AFQjCNG9nKGtWnMgUeKH2mHZclPgyOxKIw&bvm=bv.89381419,d.d2s

It was for an FPGA implementation but it shows very clearly the arithmetic, logic and memory processes. In particular, have a look at the last part that shows hashrate against frequency for a single engine. Factor in the power requirement on a 28nm implementation and you'll see exactly why it's a bad idea right now.

Yes, I know Alpha has it's problems but I always respect any company that freely publishes knowledge for the benefit of others.

As for SFARDS I wish them good luck with their chip, especially with routing it.

[sirslayer]

dual mining did work on the gridseed even though it was a forked version of cgminer and minerd.   i was able to dual mined with the previous hardware as long heat was low and 12v 5 amps power was used..

[klondike_bar]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

[notlist3d]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

I am holding lot of hope for them.  They really did a great product last time with the small ones then the blades.  Price I can't blame them on all scrypt did it.

I would be happy if they made a miner that made scrypt profitable.   I would love to mine it again.   

The BTC side we can hope for some efficiency, as it seems right now were kinda in between generations. 

[dogie]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

I believe the sentiment this time around is that it allows one set of hardware, one set of investments and one setup to swap algorithms as profitability shifts. With a flatter difficulty, each generation *should* be installed longer. Last time it didn't really work because almost immediately the sha256 side wasn't competitive / wasn't worth the cooling hassle compared to the profitability of the scrypt side.

Its hard for us to comment this time around without knowing the exact specs of the 3 modes and the arrangement of the hardware.

[Gleb Gamow]

Question: Who at Sfards should I submit my invoice to for the hours I put in trying to discredit this outfit to no avail? 

Consider this post an endorsement from Bruno Kucinskas.

Later, guys. I'm done here, but PM me if anything changes.

[brontosaurus]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

I believe the sentiment this time around is that it allows one set of hardware, one set of investments and one setup to swap algorithms as profitability shifts. With a flatter difficulty, each generation *should* be installed longer. Last time it didn't really work because almost immediately the sha256 side wasn't competitive / wasn't worth the cooling hassle compared to the profitability of the scrypt side.

Its hard for us to comment this time around without knowing the exact specs of the 3 modes and the arrangement of the hardware.

Us? Do you think you're Queen Victoria?

Seriously, do the maths and you'll see exactly why this configuration is a very bad idea. If you can make any such arrangement work, then there's something seriously wrong in your calculations.

If you need any help, just post here.

[dogie]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

I believe the sentiment this time around is that it allows one set of hardware, one set of investments and one setup to swap algorithms as profitability shifts. With a flatter difficulty, each generation *should* be installed longer. Last time it didn't really work because almost immediately the sha256 side wasn't competitive / wasn't worth the cooling hassle compared to the profitability of the scrypt side.

Its hard for us to comment this time around without knowing the exact specs of the 3 modes and the arrangement of the hardware.

Us? Do you think you're Queen Victoria?

Seriously, do the maths and you'll see exactly why this configuration is a very bad idea. If you can make any such arrangement work, then there's something seriously wrong in your calculations.

If you need any help, just post here.

How can you determine that without knowing the specifications of scypt mode only, sha256 mode only and combined?

[brontosaurus]

Dual mining technologies on a single chip is not a good idea, there are almost no common parts between the two hashing engines, the silicon for the scrypt engine is a waste of space unless LTC price goes up considerably. Also, the scrypt engine will be relatively power hungry.

...As for SFARDS I wish them good luck with their chip, especially with routing it.

i agree. There's very little real benefit to this compared to simply having seperate sha256 and scrypt mining devices. pretty sure whatever gridseeds are still in use are 100% dedicated to BTC at this point.

I believe the sentiment this time around is that it allows one set of hardware, one set of investments and one setup to swap algorithms as profitability shifts. With a flatter difficulty, each generation *should* be installed longer. Last time it didn't really work because almost immediately the sha256 side wasn't competitive / wasn't worth the cooling hassle compared to the profitability of the scrypt side.

Its hard for us to comment this time around without knowing the exact specs of the 3 modes and the arrangement of the hardware.

Us? Do you think you're Queen Victoria?

Seriously, do the maths and you'll see exactly why this configuration is a very bad idea. If you can make any such arrangement work, then there's something seriously wrong in your calculations.

If you need any help, just post here.

How can you determine that without knowing the specifications of scypt mode only, sha256 mode only and combined?

Both are fixed algorithms. If you are familiar with silicon design you can accurately determine how many times each gate/transistor in an engine stage switches in one arithmetic cycle. Knowing roughly how much power that consumes based on the characteristics of the process node - and looks from the initial power figures quoted that it's still a good 'ol fashioned 28nm straight gate based design - then it's very easy to get a good approximation of the power consumed. It's also very easy to calculate the power of the embedded DRAM blocks. Try it and see.

[pekatete]

How can you determine that without knowing the specifications of scypt mode only, sha256 mode only and combined?

Both are fixed algorithms. If you are familiar with silicon design you can accurately determine how many times each gate/transistor in an engine stage switches in one arithmetic cycle. Knowing roughly how much power that consumes based on the characteristics of the process node - and looks from the initial power figures quoted that it's still a good 'ol fashioned 28nm straight gate based design - then it's very easy to get a good approximation of the power consumed. It's also very easy to calculate the power of the embedded DRAM blocks. Try it and see.

Good info there ..... certainly above my paygrade and have no intention of pretending otherwise, unlike the usual suspects, but interesting nevertheless.

[brontosaurus]

How can you determine that without knowing the specifications of scypt mode only, sha256 mode only and combined?

Both are fixed algorithms. If you are familiar with silicon design you can accurately determine how many times each gate/transistor in an engine stage switches in one arithmetic cycle. Knowing roughly how much power that consumes based on the characteristics of the process node - and looks from the initial power figures quoted that it's still a good 'ol fashioned 28nm straight gate based design - then it's very easy to get a good approximation of the power consumed. It's also very easy to calculate the power of the embedded DRAM blocks. Try it and see.

Good info there ..... certainly above my paygrade and have no intention of pretending otherwise, unlike the usual suspects, but interesting nevertheless.

You don't give yourself enough credit, the best technologists are the ones that realise they don't have a piece of knowledge instead of bullshitting that they are an expert. Then they go and learn about it. Silicon design and power estimation are defined by fairly simple mathematics and physics, and for anyone with an inquiring mind, a desire to learn and a bit of patience, it's easy to learn the basics.

Particle physics? That's a whole different ballgame!

[vegasguy]

Does anyone know the total scrypt MHs speed of these, and also the wattage draw. Also what is the total sha-256 speed? Any pictures anywhere?

Vegas

[TheRealSteve]

tl;drt

No, <2.0, no and no.
( Based on the staggering 7 pages of thread, site, twitter, weibo and other public sources.  But mostly the staggering 7 pages of thread. )

[philipma1957]

this looks like a DOA item.

  It will do poorly if it does not show up like April 15th.  I wish it would  but I have very little hope for this item.

[Phoenix1969]

still nothing? Nothing on their other channels either...  no pics of the elusive test product.
SVARDS...
If you had nothing to actually show us, why bother posting the announcement?

[TheRealSteve]

still nothing?

We are expecting to release our first miner in April 2015.

It's only been April for part of 1 day

[notlist3d]

still nothing?

We are expecting to release our first miner in April 2015.

It's only been April for part of 1 day

Only thing is most of us I think expected to see updates or something.  They started off great showing off offices making it look like a good operation.  Then it went dark.  Updates just make people stay interested.  Going quiet really just allows for lots of speculation.

[dropt]

still nothing?

We are expecting to release our first miner in April 2015.

It's only been April for part of 1 day

Only thing is most of us I think expected to see updates or something.  They started off great showing off offices making it look like a good operation.  Then it went dark.  Updates just make people stay interested.  Going quiet really just allows for lots of speculation.

Guys, they only taped out a couple weeks ago.  They likely don't have any chips back from the foundry/packager, give it a rest.

Birdman, what are you doing out of the KnC threads? Also, you should know better by now.

[TheRealSteve]

Updates just make people stay interested.  Going quiet really just allows for lots of speculation.

Some people will speculate one way or the other.  While I, too, would prefer more frequent updates from developers, I'm just not much of a squeaky wheel type.  This wasn't really speculation, though, just a question that could have been self-answered with a few minutes of time.

[Marvell1]

this looks like a DOA item.

  It will do poorly if it does not show up like April 15th.  I wish it would  but I have very little hope for this item.

The main problem is they have given zero specs , even if you were planning a may launch you would have preliminary specs and a data sheet.

I say again this is probably vaporware for this qurter, maybe they will release something 4th q who knows.

I think this inital post was a PR stunt to make people stop buying other hardware (s5s) and hold out for something better/new.