Topic: Who was first with 28nm? (Read 2274 times)

Let me get right on that for ya!

Pointing me to unfiltered resources I already know about does not help. Please point me to where in the thread KnC has gone into detail about their clock speed and voltage choices on their ASICs. If they haven't, and no one has the scoop on it, that is also a sufficient answer.

If they have, point me to where KnC has said anything about releasing power saving firmware and tuning. I bet it's not in that thread either because it's not going to be relevant for at least a few months.

My last question is an opinion, so it's definitely not in the thread.

So a quote from VMC saying "we delivered" is enough proof for you?  Hmmm, I would like just a little more.  A single person's hands-on account of their experience with their VMC ASIC would be nice.  Oh maybe they just released a single unit to one very shy guy.  Yep, that must be it.

 

There's a 1200+ page thread discussing KNC in great detail that answers all of questions.  I know it's hard to find since it's been hiding way up at the top of the forum for several months. It's located here...

https://bitcointalksearch.org/topic/swedish-asic-miner-company-kncminercom-170332

A good place to start is page one.  Enjoy!

Tuning is currently only available for Batch1 (October) models. Although its expected soon for the November batch

I am very happy with my KNC hardware, but I do have some questions for those who have been keeping up with KnC so I can properly evaluate their promises:

I assume the extra hashrate over the promise came from overclocking (400 -> 550) and overvolting (550 -> 650). Power consumption likewise had to increase at least linearly, but likely exponentially ( -> 600 -> 850).

So if this is correct, what's the news on underclocking/undervolting firmware or tuning API?

Would you have egg on your face if after some min-maxing, machines can run and say... 800 MH/J? Or rather, at what efficiency would you have egg on your face?

The bulk chips are delayed - they've already delivered a working ASIC. I'm not sure about who was first, KNC has been much more public about it than VMC.

And even if they started shipping as stated in the PR on Nov 30th, it would make them a FULL month later than KNC.  Certainly not first, as tiaguitah stated!

Seriously, what kind of blatant VMC shilling is going on here?  Is someone trying to con newbies into preordering from VMC?  Go ahead and try to find the VMC thread(s) in the Custom Hardware subforum.  Chances are that no one has posted in days.  I give VMC about a 30% of delivering a functioning ASIC.

Pfft!

Absolute bullshit!  Did you even read the release?  If you read the next paragraph it says something like "the chips are going to be late".  I don't see how they could have shipped product if that was true... unless they are selling "other" product, perhaps Bitfury clones, t-shirts, "hang in there kitty" posters.

They announced shipping last week - https://bitcointalksearch.org/topic/m.3782916

I agree KnC can certainly improve the speed and quality of their communication.  However, their support is still light year ahead of Avalon and BFL.  The current state of ASIC hardware companies are all primarily community driven and depend on us within the community to support each other.  I don't see that changing any time soon with the accelerated speed of development, the people with the most knowledge about the products will be those who own and run it since Day 1.  KnC did handled my RMA pretty well, paid for express shipping both ways and 5 days turnaround from the time the bad board going out to receiving the replacement board.


That website made a mathematical error.  They listed the Jupiter as using 600W and Hash at 500,000 Mhash/s, so the Mhash/J should be 500,000/600 = 833 (or 1.2W/Gh), but the Mhash/J figure was incorrectly listed as 400Mhash/J or 2.5W/Gh.  On my old Oct batch Jupiter with firmware 0.96 and all 4 ASIC with 8 VRM, I was hashing at 550Gh/s while using 545 Watts on the wall with a Seasonic 80Plus Platinum PSU, which is around 1 W/Gh.  I understand newly redesigned Nov batch KnC ASIC boards are more power hungry but also hash faster at a cost to the efficiency.  The actual figure is still way less than 2W/Gh.


Product    Mhash/s    Mhash/J    Mhash/s/$    Watts
KnC Jupiter    500,000    400        80        600w

VMC delivered?  No!  Maybe they announced first but as far as I know they are still months away (if ever) from delivery.  I have no idea what tiaguitah is smoking.

Have they done the delivery?

data just taken from a running October Jupiter running latest firmware hashing 562.86 Gh/s at the pool:



that means more or less 530 watts for the 4 asic boards. Add to that a few more for the BBB and the controller board and we could get something like 560-70 at the wall.

so roughly we have ~ 1w/GHs

Although we don't know power specs for VirtualMining devices yet, they are shipping 28nm products designed by a very very reputable firm (eASIC)

55nm BitFury chips are the most power efficient of the chips which state their power consumptions.

Hm? I'm not that negative, as far as I can tell a lot of people are perfectly happy with them. When I communicated with them they couldn't keep a consistent story and it made we wary, but I'm glad other people are happy with them. I'm personally not all that happy with any of the major hardware companies right now, I'm concerned that their business practices have not been doing the ecosystem well, but thats neither here nor there and KNC is certainly not the worst of it.

Go update the mining hardware comparison as it's claiming 2.5w/Gh, if thats wrong then I retract my whining.

virtualminingcorp.com

was the first. Check the details guys.

I am not sure what KnC did or didn't do that caused you to be so negative about them.  I remember you accused them of mining with customer's hardware just before they start shipping their first batch of ASIC system.  But I wonder who is the one who has problem telling a consistent story.  The Antminer S1 are 2W/Gh (at the wall), versus 1W/Gh for KnC (Oct Batch pre-0.98 firmware), and yet you are railing against the "inefficiency" of KnC and the 2x operating cost/thermal load while suggesting the more power-hungry antminer is better because they are low power?

I love how casual you said thank you.. I bet you didn't realize the big debate you caused when you first post your question here

Thanks for the answer to my question

I try not to be rude..I know you are being a purest...   all hell who cares......

are you saying, that as long as your daily cost to mine is less than the btc you produce...but you get it so late vs the diff that you never get your money back you are fine since you will last a few months longer than a less efficient one that came out months ahead and made the power cost difference irrelevant?

that's like saying you make the bestest oil drill but you are so late that all the easy oil is already gone... did you really do better?

 

KnC's inability to tell a consistent story was one of the reasons I happily chose to not to business with them. They've clearly stated both before and after product production that it was a structured asic run, and their power results support it. Ultimately it doesn't matter if they used crackerjack boxes to make their masks, at the end what matters is the specs and they're a mixed story. I mean, sure, feel free to not care.  But a 2x increase in operating cost, and thermal load is not "a few watts", especially for those of us not interested in a high risk gamble involving mining for a few months and then throwing the hardware out.

By all means, be happy with their product— they shipped a working device to many people mostly on time, better than a lot of other vendors, and many of those customers will be happy enough with a product that goes to the landfill before the bitfury and bitmain devices. Though in terms of feature size I don't see a reason to brag about 28nm when it doesn't achieve substantially better hashrate per U or hashrate per watt even close to multiple competing 55nm designs.

Cheapest $ per gh/s chip or miner available right now?

Power efficiency does not matter right now because of the exchange to fiat.  If BTC price was $10, everybody would want to run bitfury and nothing else.
Electricity is a small fraction of the revenue generated by most miners today.

People already forget the times when BTC price was $5 and everybody wanted to get FPGAs because the GPU farms did not scale very well.

Right now, space is an important factor and heat removal can be an issue in some installations.  But cost of electricity is negligible.

Thank you! So many people get fixated on power efficiency when the total monthly electric cost for my "inefficient" KnC Jupiter is paid for within 6 hours of hashing.

Power efficiency might not mean much to the bottom line.  But it does have significance when we're potentially talking about a 3,000 watt miner.  Although I do think its way to early to worry much about that.  We have no idea how efficient KNC's 20nm is going to be.  It's purely conjecture at this moment in time.

Don't you love it when people outside of KNC/OrSoc know even more about the KNC chips than the guys who designed them? We have a number of "experts" here...

Also, agree 100% with you DPoS, if you are in a situation where a few Watts will make or break your endeavor you should know it's time to call it quits.

anyone still talking 'power efficiency' in the ASIC era is on the sidelines looking in.   it is such a small % of the equation and only means something at the end of a gen era...which is when you should be plotting to be first on the next major tech advance

tl;dr  once power efficiency means anything, you are running obsolete miners

I was with Sam from KnC a few days ago at the bitcoin expo in london and he confirmed once and for all that the knc asic is not an eAsic nor any kind of shortcut design.  it is a standard cell asic, the same as everyone else's.  the only difference is that theirs shipped in early october so whatever decisions they took in the design to get it out so fast, were very effective and valuable...

and as for power efficiency, its not that bad.  its 1W/GH at the wall.  thats better than most bitfury boards ..  most of those are > 1W/GH at the wall.

even in a datacenter— Three antminer S1 are is faster than a KNC jupiter, and I believe they take up less space (if not less, it's close— I don't have the dimensions of the KNC handy). yea, sure they involve more chips... but they are low power so they can reach reasonably high chip density in a single unit.

Although you're right about the power efficiency part, I'd say that for GH/s per m² will be in the favour of the 28nm process geometry, whatever shortcuts in design methodology are used. It's hard to see how you could design a chip layout that was only as space efficient as something twice the feature size.

But this will only make a difference to people that wish to pack hashing power into a datacenter, really. Hobbyists should still be thinking along the lines of "28nm of what?".

I dunno that fixating on the process node matters.  KNC's product is a structured asic, just one step up from a FPGA hardcopy... and it shows it— the power efficiency is half that being achieved in shipping products by others (bitfury, bitmain) on 55nm, and much lower than the 28nm products in preorder are claiming.

I believe KncMiner is the only company that has delivered a 28 nm bitcoin mining device thus far.

I have been looking around at various mining hardware sites and noticed many of them claim to be the first companies to offer 28nm ASIC's.

Maybe a silly question but which company or group was the first to have the 28nm ASIC?