Pages:
Author

Topic: HashFast announces specs for new ASIC: 400GH/s - page 28. (Read 880520 times)

sr. member
Activity: 434
Merit: 250
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.

The amount of GH per Wafer has a direct affect on the GH per chip which goes into the price of a chip and eventually into the price of a unit.

If you have 1 wafer that can make 100 chips and each chip is 10gh or 1 wafer for 100 chips and each chip at lets say 50gh, the second chip you have a better shot at turning a profit on.

In your comparison all that matters is that one wafer has 1000 gh/s and one has 5000 gh/s.

Using big chips is only detrimental because it requires $150 worth of watercooling for each chip compared to small chips which require ~$20 worth of extruded aluminum per KW for cooling.

You don't need watercooling if you are using AM's immersion cooling.  Big hot chips like HF's are perfect for Novec.

Many small chips require many more components, raising the total device cost and chance of failure.

If AM doesn't buy HF's 16nm design and stick 100,000 of them in datatanks ASAP, they will be crushed by KnC, BitFury, and Cointerra before the next block reward halving.

i think a helluva lot of due diligence is going to have to be done regarding those chips.
I doubt it's as clear cut as you think it might be. interesting to know who does finally buy the rights tho... interesting times indeed.
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.

The amount of GH per Wafer has a direct affect on the GH per chip which goes into the price of a chip and eventually into the price of a unit.

If you have 1 wafer that can make 100 chips and each chip is 10gh or 1 wafer for 100 chips and each chip at lets say 50gh, the second chip you have a better shot at turning a profit on.

In your comparison all that matters is that one wafer has 1000 gh/s and one has 5000 gh/s.

Using big chips is only detrimental because it requires $150 worth of watercooling for each chip compared to small chips which require ~$20 worth of extruded aluminum per KW for cooling.

You don't need watercooling if you are using AM's immersion cooling.  Big hot chips like HF's are perfect for Novec.

Many small chips require many more components, raising the total device cost and chance of failure.

If AM doesn't buy HF's 16nm design and stick 100,000 of them in datatanks ASAP, they will be crushed by KnC, BitFury, and Cointerra before the next block reward halving.
legendary
Activity: 3878
Merit: 1193
Let's hope HF's market-leading GH per wafer IP fetches a great price at the auction

I hope you have facts to back up that claim.
hero member
Activity: 770
Merit: 509
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.

The amount of GH per Wafer has a direct affect on the GH per chip which goes into the price of a chip and eventually into the price of a unit.

If you have 1 wafer that can make 100 chips and each chip is 10gh or 1 wafer for 100 chips and each chip at lets say 50gh, the second chip you have a better shot at turning a profit on.

In your comparison all that matters is that one wafer has 1000 gh/s and one has 5000 gh/s.

Using large/high powered chips is only detrimental because it requires $150 worth of watercooling for each chip compared to small chips which require ~$20 worth of extruded aluminum per KW for cooling.
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.

The amount of GH per Wafer has a direct affect on the GH per chip which goes into the price of a chip and eventually into the price of a unit.

If you have 1 wafer that can make 100 chips and each chip is 10gh or 1 wafer for 100 chips and each chip at lets say 50gh, the second chip you have a better shot at turning a profit on.

It's no use trying to explain things to Syke rationally.  He is in an emotional snit, suffering great butthurt, and in no mood to concede one nm to your mean old facts and logic.

Let's hope HF's market-leading GH per wafer IP fetches a great price at the auction, so Skye will get his refund and stop being so grumpy and obtuse.   Cool
legendary
Activity: 1630
Merit: 1000
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.

The amount of GH per Wafer has a direct affect on the GH per chip which goes into the price of a chip and eventually into the price of a unit.

If you have 1 wafer that can make 100 chips and each chip is 10gh or 1 wafer for 100 chips and each chip at lets say 50gh, the second chip you have a better shot at turning a profit on.
legendary
Activity: 3878
Merit: 1193
Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.

You're confusing size with density. "GH per chip" is size. "GH per mm^2" is density. Density means you get more hashrate per wafer. Size of the chip is meaningless.
legendary
Activity: 1630
Merit: 1000
I wouldnt say worthless. Hashfast still has the biggest hashrate per chip out there. The closest competitors come is Butterflylabs or cointerra at 400-600gh

"hashrate per chip" is a useless metric. In fact, it's worse than useless. The "big" chips are the least successful chips. The "small" chips are what has been successful, for good reason. I'll let you figure out why small is better.

Yes small cooler chips are more efficient for sure, but bigger chips that get alot more power per chip are cheaper.
legendary
Activity: 3878
Merit: 1193
I wouldnt say worthless. Hashfast still has the biggest hashrate per chip out there. The closest competitors come is Butterflylabs or cointerra at 400-600gh

"hashrate per chip" is a useless metric. In fact, it's worse than useless. The "big" chips are the least successful chips. The "small" chips are what has been successful, for good reason. I'll let you figure out why small is better.
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
I do see a fatal flaw in yoir argument though.
The already established h/w manufacturers were once little miners like ourselves, and we all take great pride (especially so - the honourable Chinese) in our creations.

Why breed a mongrel when your farm is full of pegidree dogs?

HF's IP is a race-winning thoroughbred horse, not a mongrel.

And thanks to hybrid vigor, mutts have healthier immune systems plus longer life spans.

Inbred pedigrees have all kinds of issues.  Poor little pugs' eyeballs sometimes pop out of their sockets!   Undecided

The "pride" issue is the major obstacle to AM using HF's 16nm IP, IMO.

There is no other reason for AM to pay extra to create its own 16nm IP from scratch when another one is available at a discount.

They should take what the hotshots at DXCore, Sandgate, and Uniquify have done for HF, and modify it with/graft on their own ideas for a final design.
sr. member
Activity: 434
Merit: 250
Good point. Also bear in mind that the chip can be evolved once the rights to it are secured. However....

Does this get any of us our money back?

That's true!  The 16nm GN2 is a work in progress and could easily be changed from a big hot chip into an AM style small cool one.

Our best chance to get our money back is to find some deep pockets for HF's IP and start a bidding war.

KnC might bid on it just to keep it out of the hands of their competition.

I've posted my suggestion on the AM thread.  Who else needs a strong headstart in the 16nm BTC ASIC space?

I'm sure 21e6 already know about the auction, but many Bitmain could use a heads-up...


I do see a fatal flaw in yoir argument though.
The already established h/w manufacturers were once little miners like ourselves, and we all take great pride (especially so - the honourable Chinese) in our creations.

Why breed a mongrel when your farm is full of pegidree dogs?
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
Good point. Also bear in mind that the chip can be evolved once the rights to it are secured. However....

Does this get any of us our money back?

That's true!  The 16nm GN2 is a work in progress and could easily be changed from a big hot chip into an AM style small cool one.

Our best chance to get our money back is to find some deep pockets for HF's IP and start a bidding war.

KnC might bid on it just to keep it out of the hands of their competition.

I've posted my suggestion on the AM thread.  Who else needs a strong headstart in the 16nm BTC ASIC space?

I'm sure 21e6 already know about the auction, but many Bitmain could use a heads-up...
sr. member
Activity: 434
Merit: 250
ASICMINER should try to outbid everyone else for HF's 28/16nm ASIC IP, but they'd probably rather waste money on another one of their own failed designs than simply buy a proven one and its bleeding edge successors.

not sure what youre spouting nonsense like this for - Asicminer's BE200 is a 40nm chip that is capable of 0.7w/GH. They already have a 28nm BE300 taped out that is expected to be around 0.3w/GH at the chip level

hashfast has effectively made thier hardware and IP worthless at this point

"Worthless?"  When you protest too much like that, it signals to everyone that you would rather make an ignorant ass of yourself than present a reasonable argument.

ASICMINER has missed their spec by a mile on every generation of their chips except the first.  The BE300 won't get 0.3W/GH, although as a shareholder I wish it would.

Did you miss the bit about HF's 16nm finFET GN2?  Buying that would give AM a chance to catch up with KnC and Cointerra.

Good point. Also bear in mind that the chip can be evolved once the rights to it are secured. However....

Does this get any of us our money back?

Does it fuck.
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
ASICMINER should try to outbid everyone else for HF's 28/16nm ASIC IP, but they'd probably rather waste money on another one of their own failed designs than simply buy a proven one and its bleeding edge successors.

not sure what youre spouting nonsense like this for - Asicminer's BE200 is a 40nm chip that is capable of 0.7w/GH. They already have a 28nm BE300 taped out that is expected to be around 0.3w/GH at the chip level

hashfast has effectively made thier hardware and IP worthless at this point

"Worthless?"  When you protest too much like that, it signals to everyone that you would rather make an ignorant ass of yourself than present a reasonable argument.

ASICMINER has missed their spec by a mile on every generation of their chips except the first.  The BE300 won't get 0.3W/GH, although as a shareholder I wish it would.

Did you miss the bit about HF's 16nm finFET GN2?  Buying that would give AM a chance to catch up with KnC and Cointerra.
legendary
Activity: 1630
Merit: 1000
So I guess this all means that us regular creditors are screwed.

All of us creditors have been screwed since the bankruptcy motion was filed months ago.

The lawyers going through the motions and doing their expensive little dances just makes the screwing official.

ASICMINER should try to outbid everyone else for HF's 28/16nm ASIC IP, but they'd probably rather waste money on another one of their own failed designs than simply buy a proven one and its bleeding edge successors.

not sure what youre spouting nonsense like this for - Asicminer's BE200 is a 40nm chip that is capable of 0.7w/GH. They already have a 28nm BE300 taped out that is expected to be around 0.3w/GH at the chip level

hashfast has effectively made thier hardware and IP worthless at this point

I wouldnt say worthless. Hashfast still has the biggest hashrate per chip out there. The closest competitors come is Butterflylabs or cointerra at 400-600gh
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
So I guess this all means that us regular creditors are screwed.

All of us creditors have been screwed since the bankruptcy motion was filed months ago.

The lawyers going through the motions and doing their expensive little dances just makes the screwing official.

ASICMINER should try to outbid everyone else for HF's 28/16nm ASIC IP, but they'd probably rather waste money on another one of their own failed designs than simply buy a proven one and its bleeding edge successors.

not sure what youre spouting nonsense like this for - Asicminer's BE200 is a 40nm chip that is capable of 0.7w/GH. They already have a 28nm BE300 taped out that is expected to be around 0.3w/GH at the chip level

hashfast has effectively made thier hardware and IP worthless at this point
legendary
Activity: 2156
Merit: 1072
Crypto is the separation of Power and State.
So I guess this all means that us regular creditors are screwed.

All of us creditors have been screwed since the bankruptcy motion was filed months ago.

The lawyers going through the motions and doing their expensive little dances just makes the screwing official.

ASICMINER should try to outbid everyone else for HF's 28/16nm ASIC IP, but they'd probably rather waste money on another one of their own failed designs than simply buy a proven one and its bleeding edge successors.
legendary
Activity: 3080
Merit: 1083
So I guess this all means that us regular creditors are screwed.
full member
Activity: 127
Merit: 100
Whatever funds are recovered from the two of them ONLY goes to the bidder....
Only Barber's claims are being auctioned off.
Mr Cavion is right, i am slightly disgusted by this though.

Meaning you don't think they should be auctioned or you think that there should be more auctioned?
thats a good question.
I don't think they should be auctioned off, but if they are going to, they should have more people on the docket.
its disgusting how easily they all seem to be getting off.
sr. member
Activity: 479
Merit: 250
Whatever funds are recovered from the two of them ONLY goes to the bidder....
Only Barber's claims are being auctioned off.
Mr Cavion is right, i am slightly disgusted by this though.

Meaning you don't think they should be auctioned or you think that there should be more auctioned?
Pages:
Jump to: