HashFast announces specs for new ASIC: 400GH/s - page 387.

jones31

newbie

Activity: 28

Merit: 0

Quote from: HarrisonS on December 12, 2013, 06:28:26 PM

Quote from: SolarSilver on December 12, 2013, 06:24:46 PM

Quote from: defcon23 on December 12, 2013, 06:10:42 PM

HashFast ‏@HashFast 8 min

Golden Nonce board bringup: board on the test bench. http://t.co/RvlUJCHiKG

On top of the fans in this image:

are those cobwebs or is that smoke?

Looks like tape.

Yeah , it's tape , god knows what's doing there.
But , couldn't they come with something more professional looking nevertheless?
This looks like Alf's spacecraft! And that thing crashed.

HarrisonS

newbie

Activity: 28

Merit: 0

They are running the chips at higher a higher clock speed than they were originally going to. Who knows if they will ship this way.

cedivad

legendary

Activity: 1176

Merit: 1001

20% more hashing power vs. almost one order of magnitude increase of the initial difficulty.
Yes, sure, it's amazing.

(btw, if they get more hashing power per chip, they will ship you a lower amount of chips in the MPP, so don't expect any gift there)

iCEBREAKER

legendary

Activity: 2156

Merit: 1072

Crypto is the separation of Power and State.

We are looking at 500GH+ per chip! Cool

ALWAYS BET ON iCE...

xstr8guy

hero member

Activity: 784

Merit: 1004

Glow Stick Dance!

Quote from: testerx on December 12, 2013, 02:54:56 PM

Quote from: DeathAndTaxes on December 12, 2013, 10:21:17 AM

Also to put it into perspective 400 GH/s devliered 1 Nov would have net ~10.8 BTC for the month of November and another ~6.8 BTC for the month of December. Had they actually shipped earlier it would have been an additional 0.5 BTC per day before 1 Nov.

Yes, but this is the world of bitcoin hardware where nobody has ever delivered on time ever (though KnC did come impressively close I will admit) where if you think about all the possible lost BTC you could have theoretically mined you surely go insane...

This is not entirely true. The most recent two ASIC companies to develop their own chips, KNC and Bitfury were no more than a few days late with shipping the bulk of their orders. That's a pretty good trend! And that has now become the new standard to bear... Not the old BFL/Avalon past history repeating itself.

Can we stop making sweeping generalizations and put this tired old cliche to bed now?

HarrisonS

newbie

Activity: 28

Merit: 0

Quote from: SolarSilver on December 12, 2013, 06:24:46 PM

Quote from: defcon23 on December 12, 2013, 06:10:42 PM

HashFast ‏@HashFast 8 min

Golden Nonce board bringup: board on the test bench. http://t.co/RvlUJCHiKG

On top of the fans in this image:

are those cobwebs or is that smoke?

Looks like tape.

ninjarobot

hero member

Activity: 761

Merit: 500

Mine Silent, Mine Deep

Quote from: DeathAndTaxes on December 12, 2013, 05:57:56 PM

The problem w/ HF was never the hardware, they seem to have some talented engineers. The problem was the sleazy and dishonest way they went about selling off stock they knew they couldn't deliver upon.

^ this

I'm happy to see this product come together and I have no doubt it will be high quality so congrats to HF for that. At the same time FU HF for playing dirty marketing games with what could have been your most loyal customer-base. One BFL is more than enough.

SolarSilver

legendary

Activity: 1112

Merit: 1000

Quote from: defcon23 on December 12, 2013, 06:10:42 PM

HashFast ‏@HashFast 8 min

Golden Nonce board bringup: board on the test bench. http://t.co/RvlUJCHiKG

On top of the fans in this image:

are those cobwebs or is that smoke?

defcon23

legendary

Activity: 1120

Merit: 1002

HashFast ‏@HashFast 8 min

Golden Nonce board bringup: board on the test bench. http://t.co/RvlUJCHiKG

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: HarrisonS on December 12, 2013, 04:59:48 PM

Each die has 96 cores, and each clock performs two hashes. Therefore with 2 dies running at 700Mhz should be achieving hashrates of 96*2*2*700=268.8GH/s. This means that either some of the cores are not working properly at all, or they are making errors. At the same error rate, and nominal clock of 550Mhz a hash rate of 194.9GH/s will be achieved, which is lower than the advertised 422.4GH/s. They will need to increase the frequency thus reducing the efficiency of the ASIC.

Well no ASIC has 100% core yield. It just doesn't happen. The 550 Mhz is only "nominal" in the sense of it is the starting point. They are still running under 0.85v which is the nominal voltage for 28nm TSMC HP process so if they can get 700 Mhz out if then there is no real reason to go slower. Ultimately they will have to find a sweet spot between power, error rate, and raw clock and have the chips target that. The problem w/ HF was never the hardware, they seem to have some talented engineers. The problem was the sleazy and dishonest way they went about selling off stock they knew they couldn't deliver upon.

defcon23

legendary

Activity: 1120

Merit: 1002

Quote from: jones31 on December 12, 2013, 03:38:51 PM

Quote from: defcon23 on December 12, 2013, 02:34:32 PM

what a pleasure to see someone receiving what a company promised to deliver in time they promised to deliver... happy christmas ! thank you ~~hashfast~~(slow...) ... thank you knc !

And you're supposed to be a knc shill?

no no

= meme pas ...

SolarSilver

legendary

Activity: 1112

Merit: 1000

Quote from: mgio on December 12, 2013, 02:58:25 PM

Not even batch 2 Avalons! My batch 2 Avalon has only made 70 coins and it isn't likely to make more than 1 or 2 more through the rest of its life.

It's all about the timing... Batch2 Wave1 was more expensive than Batch2 Wave2, and if you got it late you might have lost out some revenue but a Batch2 Wave1 unit should have brought in to date BTC 105 and the Batch2 Wave2 should have raked in BTC 93. How *late* did you get your unit to only have BTC 70?

HarrisonS

newbie

Activity: 28

Merit: 0

Each die has 96 cores, and each clock performs two hashes. Therefore with 2 dies running at 700Mhz should be achieving hashrates of 96*2*2*700=268.8GH/s. This means that either some of the cores are not working properly at all, or they are making errors. At the same error rate, and nominal clock of 550Mhz a hash rate of 194.9GH/s will be achieved, which is lower than the advertised 422.4GH/s. They will need to increase the frequency thus reducing the efficiency of the ASIC.

VolanicEruptor

sr. member

Activity: 392

Merit: 250

are my icedrill shares worth anything yet Undecided

ImI

legendary

Activity: 1946

Merit: 1019

Quote

GN chip bringup: Running two of the four die in the chip. Clocking it at 700mhz and .84v core voltage, it's doing 248Gh/s on only 2 dies!

yes, thats what i want to hear!

mgio

hero member

Activity: 546

Merit: 500

Quote from: Puppet on December 12, 2013, 04:01:25 PM

Quote from: mgio on December 12, 2013, 03:23:18 PM

Quote from: tacotime on December 12, 2013, 03:10:11 PM

I feel like the moral of the story is that no one should buy mining hardware until we hit mining hardware equilibrium where it is readily available for anyone who may want it.

I'm batch two with no MPP...

No, it's the opposite. Once we hit equilibrium, pretty much NO ONE will make money and those who do are the ones who have free electricity and cheap time.

If we'd approach an equilibrium, there should be opportunity to make a small profit for those who do it most efficiently. Cheap electricity will be a key factor, but not the only one. The problem however, is that we will likely overshoot any equilibrium, because people dont seem to mind pre ordering 6+ months ahead while underestimating how much is being ordered and without minding paying (a lot) more for hardware than the bitcoins that hardware can generate cost on the exchanges.

Yes, exactly. Plus there are enough people out there who can't do math and want to jump on the bitcoin bandwagon who will continue to pay too much for miners keeping the price high.

Puppet

legendary

Activity: 980

Merit: 1040

Quote from: mgio on December 12, 2013, 03:23:18 PM

Quote from: tacotime on December 12, 2013, 03:10:11 PM

I feel like the moral of the story is that no one should buy mining hardware until we hit mining hardware equilibrium where it is readily available for anyone who may want it.

I'm batch two with no MPP...

No, it's the opposite. Once we hit equilibrium, pretty much NO ONE will make money and those who do are the ones who have free electricity and cheap time.

If we'd approach an equilibrium, there should be opportunity to make a small profit for those who do it most efficiently. Cheap electricity will be a key factor, but not the only one. The problem however, is that we will likely overshoot any equilibrium, because people dont seem to mind pre ordering 6+ months ahead while underestimating how much is being ordered and without minding paying (a lot) more for hardware than the bitcoins that hardware can generate cost on the exchanges.

Ghrindy

newbie

Activity: 47

Merit: 0

jones31

newbie

Activity: 28

Merit: 0

Quote from: defcon23 on December 12, 2013, 02:34:32 PM

what a pleasure to see someone receiving what a company promised to deliver in time they promised to deliver... happy christmas ! thank you ~~hashfast~~(slow...) ... thank you knc !

And you're supposed to be a knc shill?

jones31

newbie

Activity: 28

Merit: 0

Topic: HashFast announces specs for new ASIC: 400GH/s - page 387. (Read 880461 times)