Topic: Algorithmically placed FPGA miner: 255MH/s/chip, supports all known boards - page 35. (Read 119440 times)

Sometimes we are not able to decide the course of things, sometimes they just go a certain way and we have no control over it, sometimes the hand that feeds you gets bit regardless of how badly you dont want to destroy it.

and if you think the Bitcoin core devs make any big decisions based on what the "community deems necessary" you are dreaming and its time to wake up.

As far as the bet goes we need a date, which realistically this could happen in the next 1-2 years kind of long to make a bet dont you think?


I would welcome ASICs to the Bitcoin mining world if I thought that someone who wasnt motivated by greed was going to release one any time soon.

The thing with ASICs are this, realistically once you get passed the design and prototyping stage and you have something that works its very inexpensive to make these, we are talking about a couple dollars, a REALISTIC price for an ASIC chip - resale value is like $5-$10

Now lets just say that BFL's trade in program is completely legit, theres no small print.. if they are willing to give you a FULL refund value on a $600 item that has very little resale value for something that costs a few bucks to make. How much do you think they are going to be selling these ASICs for?

The answer is simple there.

Now what about wealthy private developers who have the money to design, prototype and manufacture their own ASICs?

51% of the Network could quite literally be purchased for probably less than what Gigavps's house is worth.


We TRULY need an OpenASIC initiative, something real and competitive. Otherwise it has the potential to get way out of control.

I dont mean to beat a dead horse here, I just want people to think before they sink everything into overpriced ASIC chips that have the potential to be rendered usesless later on down the line. ASICs can not adapt, we can change the software on our GPU/FPGA miners , but ASICs cannot, think about that before you spend a grand on something that cost a few bucks to make.

1. The core bitcoin developer team would only change the protocol if the entire community deemed it necessary.
2. If ASIC is brought out in a way to appeal to the masses, I do not see this happening.

I would be willing to take this bet any day of the week. No ASIC developer is going to want to destroy the hand that feeds them.

lol ok ok

Personally I would like to see a repeat of last summer, Lets hit $30 a coin just in time for something catastrophic to happen and knock it down to $2

LOL
No I was thinking something like "Bitcoin difficulty will be over x by this date x"

I'm willing to bet 100 BTC that ASICs will be the first mining hardware to be rendered useless by Bitcoin developers.

I think this is the perfect subject for a bet. Does anyone have good ideas for a bulletproof statement that we can gamble on for this?

Personally I'm hoping/speculating that that won't be for atleast another year yet.

One of the next upcoming changes is a huge difficulty spike and FPGAa becoming irrelevant because of ASICs. I'm pretty sure that eldentyrell is going to make LESS from this scheme than if he sold it outright. But we will see.

I'm only trying to point out consequences of following scenario if, and some big if's I agree, the following conditions are satisfied.
This is supported by some very simple math and, maybe, an optimistic attitude towards the future health of bitcoin.

Please consider what would happen if

1.  GPU miners were displaced by FPGA's.  GPU's being utterly obsoleted will happen.  (BFL's impact can screw with this in a big way, of course.)
2.  EldenTyrell is everpresent with commission based bitstreams for future FPGA processes.  Setting a precedent for bitstream performance / pricing.  Which FPGA miner would not sign up?  Not many IMHO.
3.  FPGA's represent an ever increasing percentage of total network hash rate.  What is the upperbound on this?  Any forum member with deeper experience of FPGA penetration is invited to comment.
4.  BTC/USD is healthy whether a steady incline or violent increases.

A 5% hash share commission on current and future FPGA processes could equate to substantial amount of cash.

If the situation is entirely optimistic towards BTC, FPGA's and EldenTyrell this could turn in to a situation where the amount compensated greatly outweighs the skill, experience and time committed to develop the bitstream(s).  At that point we would be paying for the privilege of mining with EldenTyrell's bitstream rather than a fair compensation for his work.

It's folly to not negotiate, and renegotiate, commission based sales.  Why?  Things change.

No, they don't use Spartan 6s.

No, the singles don't use XC6SLX150 chips, so it isn't compatible.

So just out of curiosity: Is it possible to run this stream on the BFL Singles and get more hashes out of them?
(And if so, who will be the first to report the succes? )

Well it is essentially "free"... And don't forget tricone-mining has to maintain the signing servers as well.

What would you prefer? A monthly license for 0.25BTC? I know I'd pay that. Because that's the equivilant you're paying for using the bitstream at the moment.

He/she/they have obviously put a lot of work into this and no-one is forcing you to use it.

I sense greed in you nedbert9, if anyone. (no offense intended)

while i wouldn't throw a term like greed around lightly, (we're all here mining bitcoins intending to make money after all, aren't we?) the potential for the commision could be substantial, yea. but then, as the last couple weeks clearly show, there could be some other development just around the corner that leads to everyone dropping this bitstream over night, since it's freely available software. no one's going to hesitate to update ot the next big bitstream. i don't hold any kind of malice towards somone wanting to make what they can from their work, while they can, knowing it could evaproate like that.

I dislike the approach of a perpetual transaction based relationship.

The profit potential of this is too big (if BFL doesn't take too many operations)

I sense greed.

If they can deliver on their promise of a Single 2.0 that does less than 50w and is as fast or faster than existing Singles, then I'll let them slide on it. Cost isn't an issue if we can't get the average person to buy dedicated Bitcoin mining hardware and proudly display it on their desk and show it to others, no matter who manufactures it.

People need skin in the game.

Assuming their ASIC actually exists.  And also assuming that when they say "ASIC" they don't really mean "old FPGA" like the last time they said they had an ASIC.  I've been mining with Virtex-II Pros for a month now and I don't go around telling people that I have ASICs...

Fool me once, shame on you -- fool me twice...

28nm fpga's out doing BFL's asic... man, if that's actually potentially possible (which by all i've read seems it's a bit of a toss up right now) the next 6 months or so are gonna prove fascinating.

If you can manage to eldentyrell BFL's ASIC with 28nm FPGAs... this will have come full circle.

Yes; I should address this.  I need to apologize for the slightly imprecise nature of the performance figures at this exact moment… there's a bit of a background story here that doesn't really fit into the website.

When I started the "countdown clock" I genuinely believed that 255MH/s was the limit for this design.  Then, a few days after starting the clock, I unexpectedly stumbled across a way to free up another 8 SLICEL's and how to keep a bunch of the SLICEX->SLICEL routes from leaving the switchbox.  These changes had a ripple effect on routing congestion, and made a massive, massive, massive difference.  It was great that I was getting these unexpected improvements, but exploiting this new stuff wreaked havoc on my timeline and schedule.  I didn't want to change the deadline, so during the last week I've spent very little time in the lab actually running this stuff and checking power/heat/etc and most of my time in front of fpga_editor.  And the last day or so has been really crazy getting the website and signcryption network deployed.  Anyways, the bottom line is that this is all going to come into much clearer focus over the course of the next week, and my "lame excuse" for not having everything neat-and-tidy when midnight hit is the major performance improvement compared to what I thought I was going to get when I initially set the deadline.


Already done that.  It isn't tuned, but I'll definitely have an Artix-7 bitstream the day they ship the chips (whenever that is…).  Porting this design to other Xilinx devices is really easy.  Porting to other fpga architectures like Altera would be a major effort.