HashFast announces specs for new ASIC: 400GH/s

[cypherdoc]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

the good news is there's roof right above the ceiling and a sliding glass door out to a balcony for some flexibility.

[aerobatic]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

the good news is there's roof right above the ceiling and a sliding glass door out to a balcony for some flexibility.

since the life of the systems doesn't have to be that long, you may not have to worry that much about humidity so give some thought to using large amounts of ambient air as an intake and exhaust the hot air out the roof.  if you can get enough airflow from the outside...  you may only need the a/c to supplement the cooling in the warmer months...

and the life of these systems will start in the middle of winter, and by the time the height of summer rolls around in 8-9 months, their value may not be so much that you wouldn't consider selling them and buying the next generation gear anyway...

-- Jez

[aerobatic]

damn.  clarification on the airflow direction from the psu's for the Sierras.

D&T is correct, the Seasonics intake air from the sides and exhaust out the back.  i swear Simon told me the opposite.  oh well.

cypherdoc -

its also possible that hashfast has special ordered versions of the seasonic psus that have the airflow reversed...

-- Jez

no, i just heard back from him right now.  but i swear i heard the opposite from him earlier.  no matter.

If you wanted to max overclock the hardware you could probably put a bigger radiator on each one, e.g. the H100i (240mm rad, with two or four fans) - or the extreme cooling of the H110 (280mm radiator!!) on each chip... and leave the rads sticking out the back (even less chance of winning the `Concourse D'Elegance'...   while you're at it, you could remove the psus and have them running outboard, so their airflow doesn't contribute to the internal heat.

http://www.corsair.com/us/cpu-cooling-kits/hydro-series-water-cooling-cpu-cooler/hydro-series-h110-280mm-extreme-performance-liquid-cpu-cooler.html

[cypherdoc]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

the good news is there's roof right above the ceiling and a sliding glass door out to a balcony for some flexibility.

since the life of the systems doesn't have to be that long, you may not have to worry that much about humidity so give some thought to using large amounts of ambient air as an intake and exhaust the hot air out the roof.  if you can get enough airflow from the outside...  you may only need the a/c to supplement the cooling in the warmer months...

and the life of these systems will start in the middle of winter, and by the time the height of summer rolls around in 8-9 months, their value may not be so much that you wouldn't consider selling them and buying the next generation gear anyway...

-- Jez

that's the way i'm intuitively leaning as well.

[dexX7]

Here is a question about MPP v2:

A November Baby Jet costs $ 2760 without MPP and $ 4560 with MPP. This means the MPP costs $ 1800.

If the Bitcoin network hashrate increases so that your Baby Jet doesn’t generate more Bitcoins in ninety days than you paid for it, HashFast will give you additional ASICs.

What amount is covered by the MPP? $ 2760 for the BJ or $ 4560 for the BJ + MPP?

[DeathAndTaxes]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

What is your peak summer time temp.

Assumming you can get enough forced airflow it is ALWAYS going to be cheaper than AC.  AC at best is going to be 3:1 COP in real world.  That means to remove 3KW of heat is going to require another 1KW of AC power.   Using forced airflow a COP of 6:1, 8:1, even 10:1 is possible.   

However we are talking some serious airflow.  Do you have the ability to cut holes in walls, preferably a large intake hole and exhaust hole on opposite walls with the ASIC racks in between?

[perezoso]

What amount is covered by the MPP? $ 2760 for the BJ or $ 4560 for the BJ + MPP?

Of course the latter.  An $1800 insurance policy on a $2760 investment?

----

Anyway, with it totally obvious that everybody on the MPP is going to be getting several boards under the MPP  - even the guys that ordered one BJ and no upgrade - I think HashFast should just face up to the music and eat the BJ hardware cost, at least that hardware bought to fill purchases under the MPP.

Keep it simple and put your clients in a position to easily and quickly get the MPP boards hashing.  Send everything on the MPP program out in Sierra cases from day 1. So fully equipped Sierra cases initially go out with 1, 2, or 3 ASICs, depending on the orders of the specific customer, with open slots for MPP boards or purchased upgrades.

Hashfast wants to sell miniboards, after all.

That's without prejudice to the other discussions about when the MPP should kick in...

[crumbs]

What amount is covered by the MPP? $ 2760 for the BJ or $ 4560 for the BJ + MPP?

Of course the former.  An $1800 insurance policy on a $2760 investment?

----

Anyway, with it totally obvious that everybody on the MPP is going to be getting several boards under the MPP  - even the guys that ordered one BJ and no upgrade - I think HashFast should just face up to the music and eat the BJ hardware cost, at least that hardware bought to fill purchases under the MPP.
...

Hashfast has made its position clear from the start, stating that delivery dates were approximate & not binding in the first draft, and tightening up the wording to "by December 31" in the final version of the Terms of Sale.

This is not news to anyone who has bothered to read this thread.  Nor to anyone who bothered to read the Terms of Sale.  So it's not Hashfast who needs to face the music here. 
 

[cypherdoc]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

What is your peak summer time temp.

Assumming you can get enough forced airflow it is ALWAYS going to be cheaper than AC.  AC at best is going to be 3:1 COP in real world.  That means to remove 3KW of heat is going to require another 1KW of AC power.   Using forced airflow a COP of 6:1, 8:1, even 10:1 is possible.   

However we are talking some serious airflow.  Do you have the ability to cut holes in walls, preferably a large intake hole and exhaust hole on opposite walls with the ASIC racks in between?

the door has an ordinary vent screen maybe 18x12". 

i'm thinking of installing a large attic fan in the ceiling out to the roof.

[cypherdoc]

what's the best way to cool down a 10x9 room with all these hashing units in it?

ac, fan enabled high airflow, or a combination of both?  don't forget these chips respond to cool air by self oc'ing which is what i want.

i've priced out the specs doing it by ac alone and it is expensive.  the amount of kW of heat is enormous on calculation.  there's gotta be an easier way.  i live in an inclement area with pretty standard 72 degrees most of the year with some heat during summer.

What is your peak summer time temp.

Assumming you can get enough forced airflow it is ALWAYS going to be cheaper than AC.  AC at best is going to be 3:1 COP in real world.  That means to remove 3KW of heat is going to require another 1KW of AC power.   Using forced airflow a COP of 6:1, 8:1, even 10:1 is possible.   

However we are talking some serious airflow.  Do you have the ability to cut holes in walls, preferably a large intake hole and exhaust hole on opposite walls with the ASIC racks in between?

avg August high is 74

[Wesly]

I will ask here before opening a poll. For the first batch customers, what your prefer:

a) BJ in mid november and stay with the MPP starting the day shipped
b) BJ in mid november and the MPP starting 1st november
c) Turn your order into a sierra that ships in late november and refuse to the MPP

Um who would choose a over b?

If you select b) MPP will start calculating bitcoin generated at the lower difficulty starting from Nov 1st, which from my estimate meant 50% of the bitcoins 90 days = only 2x extra hashing, but you are getting 800GH 15 days sooner than option a)

If you select option a), you will likely ended up with the max 400% MPP, but you are receiving 1.6TH 15 days later than option b), which depend on how fast you believe the network will grow in that 15 days, might or might not be a better option than getting half the extra hashing chips 15 days sooner.

[itod]

I will ask here before opening a poll. For the first batch customers, what your prefer:

a) BJ in mid november and stay with the MPP starting the day shipped
b) BJ in mid november and the MPP starting 1st november
c) Turn your order into a sierra that ships in late november and refuse to the MPP

Um who would choose a over b?

If you select b) MPP will start calculating bitcoin generated at the lower difficulty starting from Nov 1st, which from my estimate meant 50% of the bitcoins 90 days = only 2x extra hashing, but you are getting 800GH 15 days sooner than option a)

If you select option a), you will likely ended up with the max 400% MPP, but you are receiving 1.6TH 15 days later than option b), which depend on how fast you believe the network will grow in that 15 days, might or might not be a better option than getting half the extra hashing chips 15 days sooner.

I think it is the other way around, if you select b) and mine for 2-2.5 months instead of 3 you will mine less BTC, so MPP compensation should be more boards, not less. But even that is irrelevant since looking how things are right now all MPP will be maximum amount of boards, the loss of a current delay will be that big. The main point is you get MPP compensation 2-3 weeks earlier. That is the key, delivery time in 3 months may be as important as it is now or more.

[Tornate]

They added a discount of $900 for December batch (compared with November one), thats only a 12% discount! It is unreasonable low discount due to 120% monthly difficulty increase.

[Ytterbium]

Are you ready to invest so much for your needs? Don't you need a considerable amount of money in order to set up this kind of thing?

Depends on what you mean by considerable.  My estimate for the prototype immersion tank is $1,500 to $2,000 (excluding ASICs) and it should be capable of cooling 4TH/s to 8TH/s of hardware.  The actual capacity will depend on a lot of factors primary ASIC board size.

As for "needs"?  Sometimes people just do things to see if they can.  In theory with immersion cooling your ASICs could provide year round hot water and heating in the winter "for free".  What heat you can't use can be cheaply dumped outside.   The fact that is can be done in a smaller space, and with noise is an added advantage.   To take full advantage of the capabilities I will need one or more vendors to sell "naked modules" (no cooling, power, cases, heatsinks, etc) at a reasonable cost.  The initial purchases are just for evaluation.

Have you received your knc yet?

No but it was a Nov order so I don't expect it before Nov 15th.

Oil immersion warming will destroy HF chips. (or cause them to run at a lower hashrate to avoid burning themselves)

Do you have any tangible proof that hashfast will delivery anything before the end of the year?
Because i don't. Like a pic of a PCB. Do we have a pic of the PCB we should have received already?

Do you have any tangible proof that hashfast WONT deliver anything before the end of the year?

I seriously think you should stay out of this business.

Their customers are fucked if they don't deliver before the end of october like they promised. They've already said they'll be late by one or two diff adjustments.

And if you mean the "business" of buying miners from HashFast, everyone should have stayed out of that.

I think it's quite clear, as stated they have one component which has let them down and this is the reason for them being late.

All this talk of HF never delivering is inflammatory given the PR.

Sure thing, paid HF sockpuppet.  Srsly.  8 posts, all in the HF thread, except 1 in newb jail.

They don't have to "never deliver" for their customers to get screwed, they only need to be a month or so late.  And they're already two weeks late.

[dopamine]

So hashfast any update? Or did batch 1 owners get screwed?

[Wesly]

I will ask here before opening a poll. For the first batch customers, what your prefer:

a) BJ in mid november and stay with the MPP starting the day shipped
b) BJ in mid november and the MPP starting 1st november
c) Turn your order into a sierra that ships in late november and refuse to the MPP

Um who would choose a over b?

If you select b) MPP will start calculating bitcoin generated at the lower difficulty starting from Nov 1st, which from my estimate meant 50% of the bitcoins 90 days = only 2x extra hashing, but you are getting 800GH 15 days sooner than option a)

If you select option a), you will likely ended up with the max 400% MPP, but you are receiving 1.6TH 15 days later than option b), which depend on how fast you believe the network will grow in that 15 days, might or might not be a better option than getting half the extra hashing chips 15 days sooner.

I think it is the other way around, if you select b) and mine for 2-2.5 months instead of 3 you will mine less BTC, so MPP compensation should be more boards, not less. But even that is irrelevant since looking how things are right now all MPP will be maximum amount of boards, the loss of a current delay will be that big. The main point is you get MPP compensation 2-3 weeks earlier. That is the key, delivery time in 3 months may be as important as it is now or more.

MPP measures a 90 days (~ 3 months) period, so you are also asking to change the length of the MPP and not just the starting date of MPP?  If that's the case, this should have been specified in option b), or put it simply option b) should be stated as BJ in mid november and MPP started the day 1st BJ shipped until end of January.

[greenbtc]

So hashfast any update? Or did batch 1 owners get screwed?

Looking this way

[DeathAndTaxes]

Oil immersion warming will destroy HF chips. (or cause them to run at a lower hashrate to avoid burning themselves)

I won't be using oil.   Fluorinert.

[MinorMiner]

to hashfast:

Q: What will be done with regard to the October batch miner protection program? Will you immediately send out the upgraded hashing power since the difficulty will be 2 to 4 times higher by the time we receive our units?


Hashfast's answer:

"Thank you for your inquiry.

No, we will not being sending out the MPP modules with the purchased units. MPP is enacted 90 days after the original miners have been delivered.

Kindly,

Erin"


Hmm.


--
HashFast Support
support@hashfast.com"

[Ytterbium]

Oil immersion warming will destroy HF chips. (or cause them to run at a lower hashrate to avoid burning themselves)

I won't be using oil.   Fluorinert.

Seems like a lot more trouble then it's worth, especially since you can just heat water directly with the chips using water blocks. Why go through the intermediate step of boiling fluorinert (I assume you mean Perfluorohexane) and then trying to re-condense it somehow?

Hashfast's answer:

"Thank you for your inquiry.

No, we will not being sending out the MPP modules with the purchased units. MPP is enacted 90 days after the original miners have been delivered.

Kindly,

Erin"

So their delay on shipping the miners also means an equal delay shipping the MPP modules? 

That's rather dickish.