HashFast announces specs for new ASIC: 400GH/s

[arorts]

Oh and I should mention that the PSU that this happend to was a Corsair AX860 which is likely made by Seasonic. I don't know the gauge of the wire they use but if I had to guess I'd say it 18 AWG.

 Strange :| I'm running a full rig @ ~570GHs off a TX850M and she seems cool as a cucumber.

But like I said I bet you did not make the mistake of running BOTH 6pin connectors off of the same main PSU 8pin PCIE feeder (talking about the modular plugs on the PSU itself) Essentially it's a splitter that converts 8pin to TWO 6pin. That is what I was using and that's a big no no. Check your total power usage and feel to see how hot the cables feel at the PSU end. That is where it will most likely melt the plastic around the pins.

I also observed a funny phenomenon where the system slowly overclocks itself. Essentially the resistance on the h-cards at the R02 resistor lowers itself with time. Don't ask me how this happens, but it may be somehow the way I pencil modded them? I dunno, either way I'm hoping that it will remain stable for now.

Anyways, wrong thread for this discussion.

Yes, that's true!! The R02 resistor lowers itself with time. I guess the hot carbon sitting in that resistor becomes more "resisting" over time?? I've consistently seen this effect as well. I used HB pencil. A good experiment is to use H2 or harder pencil to see if it has the same or worse effect.

[1714026327]

1714026327

[1714026327]

1714026327

[1714026327]

1714026327

[1714026327]

1714026327

[aerobatic]

"They’ve started assembling Baby Jets (overnight tonight). They aim by night’s end to have the first round of assembly done for 400 Baby Jets."

very impressive to be able to build almost all the baby jet orders in just one night... yea!  looking good for xmas...

-- Jez

[allinvain]

Oh and I should mention that the PSU that this happend to was a Corsair AX860 which is likely made by Seasonic. I don't know the gauge of the wire they use but if I had to guess I'd say it 18 AWG.

 Strange :| I'm running a full rig @ ~570GHs off a TX850M and she seems cool as a cucumber.

But like I said I bet you did not make the mistake of running BOTH 6pin connectors off of the same main PSU 8pin PCIE feeder (talking about the modular plugs on the PSU itself) Essentially it's a splitter that converts 8pin to TWO 6pin. That is what I was using and that's a big no no. Check your total power usage and feel to see how hot the cables feel at the PSU end. That is where it will most likely melt the plastic around the pins.

I also observed a funny phenomenon where the system slowly overclocks itself. Essentially the resistance on the h-cards at the R02 resistor lowers itself with time. Don't ask me how this happens, but it may be somehow the way I pencil modded them? I dunno, either way I'm hoping that it will remain stable for now.

Anyways, wrong thread for this discussion.

Yes, that's true!! The R02 resistor lowers itself with time. I guess the hot carbon sitting in that resistor becomes more "resisting" over time?? I've consistently seen this effect as well. I used HB pencil. A good experiment is to use H2 or harder pencil to see if it has the same or worse effect.

Thanks for confirming that for me. I thought it was all in my head lol I too am using HB as recommended by a guide here on the forum. I have a wide variety of pencils available so I guess next time I'll try something harder. The problem may be getting a H2 or harder to actually remain or stuck on the resistor.

[itod]

As each day passes HF looks more and more as major failure. It may sound cruel for BTC community, but now I hope for temporary BTC price drop at refund time, so the refunds would have more value when issued. Even if HF cheats with BTC refunds and calculates them in US$ it would have less impact for the buyers.

 You get what you paid. It's even stupid to think that they would pay you in dollars at a new conversion rate since they took BTC when you closed the sale contract with them. Otherwise, expect a massive class action lawsuit since that'd be fraud. They have to return what that took from you. Did they took oranges or peanuts? No. So no need to worry about what's the current conversion rate with any currency. It'd be double stupid to do otherwise considering that the MPP is based on BTC return!! Not dollar returns.

You are calling me stupid for suspecting HF won't refund full BTC amount? Just stop for a second and think about responsibility you are taking upon your self for convincing people something will happen you have no evidence on. On the contrary, there were repeated questions on this forum about this subject that never once were responded by any HF official. More evidence to the contrary side, they've claimed at several places that the official accounting currency for their goods is US$. What would happen when some poor soul request a refund based on your's and similar ramblings, only to find the fraction of original BTC amount in their wallet? You will say "I'm sorry"? You will fund their legal action if they don't have the funds to do that themselves?

[crumbs]

...
I also observed a funny phenomenon where the system slowly overclocks itself. Essentially the resistance on the h-cards at the R02 resistor lowers itself with time. Don't ask me how this happens, but it may be somehow the way I pencil modded them? I dunno, either way I'm hoping that it will remain stable for now.

Anyways, wrong thread for this discussion.

Yes, that's true!! The R02 resistor lowers itself with time. I guess the hot carbon sitting in that resistor becomes more "resisting" over time?? I've consistently seen this effect as well. I used HB pencil. A good experiment is to use H2 or harder pencil to see if it has the same or worse effect.

This is pretty odd, when carbon resistors age, the resistance *always* goes up (as in 1k resistor will become 1.1k).  Weird.

[allinvain]

...
I also observed a funny phenomenon where the system slowly overclocks itself. Essentially the resistance on the h-cards at the R02 resistor lowers itself with time. Don't ask me how this happens, but it may be somehow the way I pencil modded them? I dunno, either way I'm hoping that it will remain stable for now.

Anyways, wrong thread for this discussion.

Yes, that's true!! The R02 resistor lowers itself with time. I guess the hot carbon sitting in that resistor becomes more "resisting" over time?? I've consistently seen this effect as well. I used HB pencil. A good experiment is to use H2 or harder pencil to see if it has the same or worse effect.

This is pretty odd, when carbon resistors age, the resistance *always* goes up (as in 1k resistor will become 1.1k).  Weird.

Yeah it makes no sense to me either. I would expect as some of the carbon falls off essentially what should happen is seeing as how there are fewer carbon chains there the resistance should go UP and thus the cards should begin to slow down in hashing performance as the voltage drops. I guess there must be some other factor that I am not aware of. Maybe the extra heat is causing some of the solder around the resistor to change shape and perhaps causing more contact or some voodoo like that?

Anyways, I just hope that these babyjets don't start melting my PSU cables like the bitfury units.

[fubly]

Will HF deliver us batch one customers also the upgrades at first?

But inside the Babyjet they can only put two pcb, will they deliver a second BJ, and when?

Is there any official statement.

Thx

fubly

[thoughtcourier]

...
I also observed a funny phenomenon where the system slowly overclocks itself. Essentially the resistance on the h-cards at the R02 resistor lowers itself with time. Don't ask me how this happens, but it may be somehow the way I pencil modded them? I dunno, either way I'm hoping that it will remain stable for now.

Anyways, wrong thread for this discussion.

Yes, that's true!! The R02 resistor lowers itself with time. I guess the hot carbon sitting in that resistor becomes more "resisting" over time?? I've consistently seen this effect as well. I used HB pencil. A good experiment is to use H2 or harder pencil to see if it has the same or worse effect.

This is pretty odd, when carbon resistors age, the resistance *always* goes up (as in 1k resistor will become 1.1k).  Weird.

Yeah it makes no sense to me either. I would expect as some of the carbon falls off essentially what should happen is seeing as how there are fewer carbon chains there the resistance should go UP and thus the cards should begin to slow down in hashing performance as the voltage drops. I guess there must be some other factor that I am not aware of. Maybe the extra heat is causing some of the solder around the resistor to change shape and perhaps causing more contact or some voodoo like that?

Anyways, I just hope that these babyjets don't start melting my PSU cables like the bitfury units.

Not to embarrass anyone, but I will have to say that you can't measure the passive resistance of a device that is on with a multimeter. It makes a lot of sense that the multimeter is measuring a slowly decreasing resistance when that happens. (Exceptions apply, but this is probably not the case here)

[DeathAndTaxes]

The point that I'm trying to get a cross here is that over-engineering something is a good thing when you can do it without an unreasonable increase in cost.

This is one thing I don't know why more companies don't do.    Simple way to overengineer .... include a second power connector.   It is like a $0.20 part (probably less in high volume) and it adds maybe $0.02 to asembly cost.   Just attach both connector to the same power plane.  

When running close to the limit it is an easy way to add "insurance".  Given that consumers may be using unknown power supplies from unknown companies it is a way to spread out the load and if the pins in a connector are slightly loose (increased resistance) it may be enough to prevent a fire.   This is one reason why the PCIe standard is so conservative.   Yeah a 6 pin connector "can" handle 300W assuming a tight fit (hot often will a consumer push the connector in but it doesn't seat fully and results in increased resistance), the power supply company didn't cut corners, some knock off OEM didn't cheap out and use 20 AWG wiring instead of 18 AWG or use some chinese knock off just slightly out of spec crimp pins instead of ones by Molex or other major part supplies.  

In the real world lots of things can happen which are unpredictable.  A graphics card which pulls 150W from the PCIe connectors will normally use two 6 pin instead of 1 pin pin.  It is pulling 75W per connector for one which optimally should be able to handle 300W.  That overengineering is cheap insurance.  Say it adds $0.12 in cost to a $120 card ($199 retail).  If it reduces the warranty failure rate by 0.006% you broke even.

[edgie13]

Has there been any mention of when batch 2 is "projected" to be shipped? 

[DeathAndTaxes]

Not to embarrass anyone, but I will have to say that you can't measure the passive resistance of a device that is on with a multimeter. It makes a lot of sense that the multimeter is measuring a slowly decreasing resistance when that happens. (Exceptions apply, but this is probably not the case here)

Not sure but I don't think he means measured while operating (as in resistance is declining in real time).  More likely rig is off resistance is measured.    Rig operates for some times (weeks? months?) user notices hashrate is higher, turns off rig and measures resistance which is now higher.

[DeathAndTaxes]

Has there been any mention of when batch 2 is "projected" to be shipped? 

They "really" haven't even said when Batch 1 will ship.  They have said they will ship Batch 2 right after Batch 1.  There will be no artificial gap or delay.  Yet another way that Batch 1 customers got screwed essentially they are Batch 2 customers who paid double the price for nothing.

[defcon23]

Has there been any mention of when batch 2 is "projected" to be shipped? 

They "really" haven't even said when Batch 1 will ship.  They have said they will ship Batch 2 right after Batch 1.  There will be no artificial gap or delay.  Yet another way that Batch 1 customers got screwed essentially they are Batch 2 customers who paid double the price for nothing.

yes at this point : completly screwed .. really hope to receive mpp with machines.. to limit the crash..

[edgie13]

They "really" haven't even said when Batch 1 will ship.  They have said they will ship Batch 2 right after Batch 1.  There will be no artificial gap or delay.  Yet another way that Batch 1 customers got screwed essentially they are Batch 2 customers who paid double the price for nothing.

Thanks. I feel bad for people that bought into batch 1. 

When I placed my order the were 2 options on their site:

1. November delivery
2. December delivery

Option 1 was $700 more, but I decided to go with it.  Then a few days later they announced that there was going to be a production delay.  It makes me angry that they obviously knew that November delivery was impossible, but they took my extra $700 anyway. I fully understood the risks before I placed my order, but in this case I feel robbed.

[abuelau]

You are not the only one that feels robbed.

[Syke]

Latest word was 5 days ago with "Running two of the four die in the chip." Any chance of seeing all 4 dies working this week?

[Easy2Mine]

in this case I feel robbed.

Welcome to the club.
Lucky you are not alone.
Second time for me to get robbed.
First time for me was BFL.
I will never ROI bitcoins with my BFL singles.
Lucky for me I only bought 1 Sierra.
I don't think it will ROI the bitcoin that was robbed from me.

[RoadStress]

Latest word was 5 days ago with "Running two of the four die in the chip." Any chance of seeing all 4 dies working this week?

I wonder if they will ship units with only 2 working dies only to avoid refunds.

[cedivad]

On a completely hypothetic scenario, what stops them from shipping 2 boards per bj with 2 dies on each board instead of 4? Nothing. No issue there. I keep betting that 99% they will ship by the end of year.

[edgie13]

Latest word was 5 days ago with "Running two of the four die in the chip." Any chance of seeing all 4 dies working this week?

I wonder if they will ship units with only 2 working dies only to avoid refunds.

...good question.  I wouldn't be too surprised if they did.