HashFast announces specs for new ASIC: 400GH/s

[cypherdoc]

Could you please take this discussion between you two through PM ?

I dont see the reason to fill almost 2 pages with a no end discussion and how that could be interesting for someone that enters the thread looking for news/updates about HF

if you look closely, this started with me posting a very useful and helpful post about how i was approaching my own configuration process in multiples of 3 and how multiple Sierra's can be driven by one Raspi so you guys can benefit from my insider knowledge.

then crumbs comes in and attacks me just like everyone else.  what am i supposed to do?

[1714894818]

1714894818

[1714894818]

1714894818

[crumbs]

Claiming "I want to handle everything in increments of 3 [or some such]" is inconsistent with trying to sell more than one, and suddenly willing to make space for a whole server room (original 9 + the truckload of miner protection gear)  You don't have room for a bit moar?

the problem with your brain is that it's static.  like mud.

you fail to consider that ppl can change or figure out solutions to a problems as they go along.  i decided to go with being Order #1, maximize the mining while the going is good and deal with the truckload later.  you're right in that i'll probably have to find a solution to that later.  but it's a good problem to have.

how about that bet, Mr. Never Mined a BTC in My Life?

Lordy, you've talked about my yeller button, you offered to prove that you have 9 pre-orders, you accused me of not being a miner.
You've talked about so many ugly things.  The only thing you didn't address is the lie you've been caught in.  
Though i like where you're going with this.  Level up, practice makes perfect

still confused.  what lie are you talking about?

Can i get you to commit to a single story?  Why did you go through all the mutually-contradictory rationalizations before ending up with "people can change their minds"?  Tell me one last time why you wanted to unload a bunch of pre-orders from a company you were shilling for representing at the time?
?

how many times do i need to say it?  b/c of space and electrical capacity considerations.

when i first ordered BJ's, 4 TH/s was supposed to come in 4 machines.  b/c they changed their plans, it went to 10 machines.  then, thru their own generosity, the MPP is going to take it to 50 machines.  i have a small room for these.  50 machines seemed untenable and b/c i ordered first, i could sell my pre-orders for an immediate profit; still can.  but the intent was never to sell them, it was to mine.  the repackaging of these into Sierra's solves the entire issue for now.  period.

still want to bet 1000 BTC?

Why would i want to bet 1000 BTC that you don't have pre-orders?  Of course you do.  Hashfast can't have their paid shill unloading all of his pre-orders -- how good would *that* look. hmm?
But i'll let you get back to telling your stories & trolling KNC threads.

[DeathAndTaxes]

You don't need a 1U control unit. 

Each Sierra is self contained (cooling, power, 3x ASIC modules, case, and a controller/host).  The only advantage of the 1U control unit is that it would allow a single host to control up to 10 Sierras (30 hashing modules).  Also HF has indicated that the modules are connected to the "host" by USB.  This means anything which can run cgminer can be a host.  You can use the included rPi host in the Sierra and/or BabyJet or you could connect multiple hashing modules to a dedicated computer for a custom solution.  Although HF hasn't provided the specs my guess is the 1U "control unit" is just an entry level 1U server running some form of linux and cgminer.  Instead of configuring a farm through the 10 Sierra rPi hosts all 30 modules are connected by USB to the single server for ease of management.

Still if you want to go w/ a BabyJet instead, HF has indicated they intend to sell just modules in the future although no date or price has been provided.  A single sierra is likely a better deal though. as to upgrade a Babyjet to handle a second module will require a larger power supply and a second cooling loop (or larger radiator).  The sierra comes out of the box ready from day 1 for 1.2 TH/s.

I thought so too but it turns out that the BabyJet is self contained, it has the Rapberry Pi, but the Sierra does not:

We don't want to risk overloading the limited resources of a Raspberry Pi by running more than two mini-boards (formerly 'modules') off one of them.

So we recommend something more powerful for one or more Sierras.

so you will have to provide some kind of system running cgminer with USB to hook up the Sierra to.

The Sierra is not self contained

Additionally, directly from their website under Product Description:

System Specifications

Contains 3 HashFast Golden Nonce (GN) ASICs
Performance: 1,200 Ghash/s at nominal clock speed
Chassis: 2U rack mounted
Cooling System: High performance liquid cooling on each of 3 chips, dissipates heat via back mounted radiator
Requires a 1U control unit

I stand corrected. 

[-Redacted-]

Could you please take this discussion between you two through PM ?

I dont see the reason to fill almost 2 pages with a no end discussion and how that could be interesting for someone that enters the thread looking for news/updates about HF

if you look closely, this started with me posting a very useful and helpful post about how i was approaching my own configuration process in multiples of 3 and how multiple Sierra's can be driven by one Raspi so you guys can benefit from my insider knowledge.

then crumbs comes in and attacks me just like everyone else.  what am i supposed to do?

The ignore button works very well for crumbs, regardless of what thread he is currently trolling in....

[RoadStress]

Claiming "I want to handle everything in increments of 3 [or some such]" is inconsistent with trying to sell more than one, and suddenly willing to make space for a whole server room (original 9 + the truckload of miner protection gear)  You don't have room for a bit moar?

the problem with your brain is that it's static.  like mud.

you fail to consider that ppl can change or figure out solutions to a problems as they go along.  i decided to go with being Order #1, maximize the mining while the going is good and deal with the truckload later.  you're right in that i'll probably have to find a solution to that later.  but it's a good problem to have.

how about that bet, Mr. Never Mined a BTC in My Life?

Lordy, you've talked about my yeller button, you offered to prove that you have 9 pre-orders, you accused me of not being a miner.
You've talked about so many ugly things.  The only thing you didn't address is the lie you've been caught in.  
Though i like where you're going with this.  Level up, practice makes perfect

still confused.  what lie are you talking about?

Can i get you to commit to a single story?  Why did you go through all the mutually-contradictory rationalizations before ending up with "people can change their minds"?  Tell me one last time why you wanted to unload a bunch of pre-orders from a company you were shilling for representing at the time?
?

how many times do i need to say it?  b/c of space and electrical capacity considerations.

when i first ordered BJ's, 4 TH/s was supposed to come in 4 machines.  b/c they changed their plans, it went to 10 machines.  then, thru their own generosity, the MPP is going to take it to 50 machines.  i have a small room for these.  50 machines seemed untenable and b/c i ordered first, i could sell my pre-orders for an immediate profit; still can.  but the intent was never to sell them, it was to mine.  the repackaging of these into Sierra's solves the entire issue for now.  period.

still want to bet 1000 BTC?

After you admitting that HashFast screwed you and also their numerous customers by an activated MPP i just realized why are you so angry and why did you went full retard like Inaba. If you have that problem i imagine a lot of their customers didn't think through very well of that MPP and blindly ordered. Let's see how this works our for everyone.

[cypherdoc]

After you admitting that HashFast screwed you and also their numerous customers by an activated MPP i just realized why are you so angry and why did you went full retard like Inaba. If you have that problem i imagine a lot of their customers didn't think through very well of that MPP and blindly ordered. Let's see how this works our for everyone.

i never said they screwed me.  in fact, i think the MPP is incredibly generous.  and who said i'm angry?  i'm happy as hell that it appears HF is going to come out on top of these Asic Wars, imho.  

it's just that logistically, i have to think about how i can handle all this extra FREE capacity.  IT'S NOT A BAD THING.

which is why i posted my multiple of 3 strategy and personal conversion from BJ's to Sierra's all done with OEM parts.  it turns out this is going to be incredibly simple if you read carefully what i wrote.  the only ones that are going to get screwed are HF competitors b/c HF can afford to do the MPP (chips only) b/c of their smaller, more efficient chip design:

https://bitcointalk.org/index.php?topic=270384.msg2975145#msg2975145

[willphase]

which is why i posted my multiple of 3 strategy and personal conversion from BJ's to Sierra's all done with OEM parts.  it turns out this is going to be incredibly simple if you read carefully what i wrote.  the only ones that are going to get screwed are HF competitors b/c HF can afford to do the MPP (chips only) b/c of their smaller, more efficient chip design:

https://bitcointalk.org/index.php?topic=270384.msg2975145#msg2975145

hmm I hadn't seen that post before, it's quite interesting:

Let's use those two numbers as bounds for the size of the silicon under the lid. If the die(s) take up just 30x30mm of the space under the lid, then:
     30x30mm = 900mm^2
     100 GHash / 900 mm^2 = 0.11 GHash/mm^2

    Or if the die takes up a bit more of the space under the lid,
      36x36mm = 1296mm^2
      100 GHash / 1296mm^2 = 0.077 GHash/mm^2

HashFast's Golden Nonce chip: I don't have to estimate the size because I work at HashFast. Smiley
   One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

What's interesting is that compare this to cointerra's chip specifications which give (minimum of) 504 Gh in 300 mm^2 which is 1.68 Ghash/mm^2 which is even more efficient than the HashFast chip.

Will

[crumbs]

After you admitting that HashFast screwed you and also their numerous customers by an activated MPP i just realized why are you so angry and why did you went full retard like Inaba. If you have that problem i imagine a lot of their customers didn't think through very well of that MPP and blindly ordered. Let's see how this works our for everyone.

i never said they screwed me.  in fact, i think the MPP is incredibly generous.  and who said i'm angry?  i'm happy as hell that it appears HF is going to come out on top of these Asic Wars, imho.  

it's just that logistically, i have to think about how i can handle all this extra FREE capacity.  IT'S NOT A BAD THING.

which is why i posted my multiple of 3 strategy and personal conversion from BJ's to Sierra's all done with OEM parts.  it turns out this is going to be incredibly simple if you read carefully what i wrote.  the only ones that are going to get screwed are HF competitors b/c HF can afford to do the MPP (chips only) b/c of their smaller, more efficient chip design:

https://bitcointalk.org/index.php?topic=270384.msg2975145#msg2975145

Cypher, you're batch 1, right?

According to this HashFast release, you're just getting raw chips, not miners (watch for bold font ):

Hey Guys,

This statement needs to be clarified and was released prematurely. We are currently in the final stages of an internal discussion about how to further protect miners moving forward for customers that opt into purchasing it.

We have preliminarily decided to further protect our miners that opt in for the miner protection at a higher price point, and provide full modules for orders placed in Batch 2 and 3.

We’re sorry this went a bit early - we’ll be back with full details as soon as we finalize and get the documentation together.

Cheers,

Team HashFast

[cypherdoc]

which is why i posted my multiple of 3 strategy and personal conversion from BJ's to Sierra's all done with OEM parts.  it turns out this is going to be incredibly simple if you read carefully what i wrote.  the only ones that are going to get screwed are HF competitors b/c HF can afford to do the MPP (chips only) b/c of their smaller, more efficient chip design:

https://bitcointalk.org/index.php?topic=270384.msg2975145#msg2975145

hmm I hadn't seen that post before, it's quite interesting:

Let's use those two numbers as bounds for the size of the silicon under the lid. If the die(s) take up just 30x30mm of the space under the lid, then:
     30x30mm = 900mm^2
     100 GHash / 900 mm^2 = 0.11 GHash/mm^2

    Or if the die takes up a bit more of the space under the lid,
      36x36mm = 1296mm^2
      100 GHash / 1296mm^2 = 0.077 GHash/mm^2

HashFast's Golden Nonce chip: I don't have to estimate the size because I work at HashFast. Smiley
   One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

What's interesting is that compare this to cointerra's chip specifications which give (minimum of) 504 Gh in 300 mm^2 which is 1.68 Ghash/mm^2 which is even more efficient than the HashFast chip.

Will

except that Cointerra hasn't even taped out yet, seems to be late at doing so, and most importantly seems to have disappeared.

[erk]

except that Cointerra hasn't even taped out yet, seems to be late at doing so, and most importantly seems to have disappeared.

Are you sure they haven't taped out? If that's the case, then there is no way they will be shipping in December, remember factories wind down a couple of weeks before Xmas.

[roy7]

except that Cointerra hasn't even taped out yet, seems to be late at doing so, and most importantly seems to have disappeared.

Are you sure they haven't taped out? If that's the case, then there is no way they will be shipping in December, remember factories wind down a couple of weeks before Xmas.

I haven't seen Cointerra announce they haven't taped out, so I don't know the origin of that report...

[pajak666]

I haven't seen Cointerra announce they haven't taped out, so I don't know the origin of that report...

https://picostocks.com/businessplan/31
https://picostocks.com/stocks/view/31
because they had not done it yet

[dropt]

https://picostocks.com/businessplan/31
https://picostocks.com/stocks/view/31
because they had not done it yet

I'm been anxiouslty waiting for some good news on this front 

[HashFast_CL]

HashFast will be providing all MPP customers (Batches 1, 2, and 3) with additional ASICs in the form of complete mini-boards.

This will be confirmed shortly by the Official HashFast account.

[cypherdoc]

HashFast will be providing all MPP customers (Batches 1, 2, and 3) with additional ASICs in the form of complete mini-boards.

This will be confirmed shortly by the Official HashFast account.

i love you HashFast

[HashFast]

Hey Guys,

We realized that a bit of confusion was raised yesterday in regard to the MPP. So this is the official post concerning the MPP:

HashFast will be providing all MPP customers (Batches 1, 2, and 3) with additional ASICs in the form of complete mini-boards.

Cheers,

Team HashFast

[regular]

Good to know.This at least makes break even more of sure thing if not positive ROI.

[cypherdoc]

so just to clarify everyone's strategy.

if you bought a BJ, when the MPP hits, install 1 module into your current BJ.  then simply buy OEM parts: a 2U or 4U rack along with PSU, cooler, fans and plunk in the other 3 modules.  you'll need an old laptop, pc, or Raspi to install integrated HF driver/cgminer and you're ready to go.

a cheap 5x power increase.  extraordinary.

[joshv06]

so just to clarify everyone's strategy.

if you bought a BJ, when the MPP hits, install 1 module into your current BJ.  then simply buy OEM parts: a 2U or 4U rack along with PSU, cooler, fans and plunk in the other 3 modules.  you'll need an old laptop, pc, or Raspi to install integrated HF driver/cgminer and you're ready to go.

a cheap 5x power increase.  extraordinary.

Thanks for clarifying. I have a noob question: If I want to host the rigs on laptops, would I just connect:

BJ->Laptop

or

BJ->Rpi->Laptop

Thanks

[cypherdoc]

so just to clarify everyone's strategy.

if you bought a BJ, when the MPP hits, install 1 module into your current BJ.  then simply buy OEM parts: a 2U or 4U rack along with PSU, cooler, fans and plunk in the other 3 modules.  you'll need an old laptop, pc, or Raspi to install integrated HF driver/cgminer and you're ready to go.

a cheap 5x power increase.  extraordinary.

Thanks for clarifying. I have a noob question: If I want to host the rigs on laptops, would I just connect:

BJ->Laptop

or

BJ->Rpi->Laptop

Thanks

what do you mean "host the rigs on laptops"?  you mean solo mine?

BJ's come with a Raspi installed so you can just plug that into your router and point at pool.  if you want to solo mine you'll have to install stratum with bitcoind on a server and point the BJ at it.