Bitcoin Forum
December 08, 2016, 04:34:00 PM *
News: Latest stable version of Bitcoin Core: 0.13.1  [Torrent].
 
   Home   Help Search Donate Login Register  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 [16] 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 »
  Print  
Author Topic: Algorithmically placed FPGA miner: 255MH/s/chip, supports all known boards  (Read 109584 times)
Garr255
Legendary
*
Offline Offline

Activity: 952


What's a GPU?


View Profile
June 01, 2012, 07:25:58 AM
 #301

Quote
We don't want board developers rushing this process and doing a crappy job (which will just take longer to clean up) in the hurry to be the first one with a TML bitstream. Therefore, we will not start reviewing BDK submissions until 7-Jun-2012. Take your time, do this right.

He is using his wisdom as a prof. Tongue

“First they ignore you, then they laugh at you, then they fight you, then you win.”  -- Mahatma Gandhi

Average time between signing on to bitcointalk: Two weeks. Please don't expect responses any faster than that!
1481214840
Hero Member
*
Offline Offline

Posts: 1481214840

View Profile Personal Message (Offline)

Ignore
1481214840
Reply with quote  #2

1481214840
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1481214840
Hero Member
*
Offline Offline

Posts: 1481214840

View Profile Personal Message (Offline)

Ignore
1481214840
Reply with quote  #2

1481214840
Report to moderator
1481214840
Hero Member
*
Offline Offline

Posts: 1481214840

View Profile Personal Message (Offline)

Ignore
1481214840
Reply with quote  #2

1481214840
Report to moderator
1481214840
Hero Member
*
Offline Offline

Posts: 1481214840

View Profile Personal Message (Offline)

Ignore
1481214840
Reply with quote  #2

1481214840
Report to moderator
BTCurious
Hero Member
*****
Offline Offline

Activity: 714


^SEM img of Si wafer edge, scanned 2012-3-12.


View Profile
June 01, 2012, 07:29:27 AM
 #302

18.4 MH/J, so to compare to BFL Singles, which run 850 Mhash/sec at a stated 80 Watt (YMMV), that gives about 46 Watts for 850 Mhash/sec, or about half of what BFL Singles use.
I don't know if this is accurate, and if the PSU is taken into account in various cases, though.

Dexter770221
Legendary
*
Offline Offline

Activity: 1026


View Profile
June 01, 2012, 07:34:09 AM
 #303

I had similiar idea, payment with hashing power. But little bit different way. Discounts. You're byuing with 10% discount but 10% of hashing power comes to my account....

Under development Modular UPGRADEABLE Miner (MUM). Looking for investors.
Changing one PCB with screwdriver and you have brand new miner in hand... Plug&Play, scalable from one module to thousands.
arklan
Legendary
*
Offline Offline

Activity: 1204


Just along for the ride...


View Profile
June 01, 2012, 07:43:22 AM
 #304

...now THAT is interesting.
eldentyrell
Donator
Legendary
*
Offline Offline

Activity: 966


felonious vagrancy, personified


View Profile WWW
June 01, 2012, 09:03:32 AM
 #305

But 270mh/sec? I wonder what the amps on that is.

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.

Plus, hes going to have to make an Artix 7 or whatever core as well.

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.

The printing press heralded the end of the Dark Ages and made the Enlightenment possible, but it took another three centuries before any country managed to put freedom of the press beyond the reach of legislators.  So it may take a while before cryptocurrencies are free of the AML-NSA-KYC surveillance plague.
DiabloD3
Legendary
*
Offline Offline

Activity: 1162


DiabloMiner author


View Profile WWW
June 01, 2012, 09:12:16 AM
 #306

But 270mh/sec? I wonder what the amps on that is.

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.  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.

Plus, hes going to have to make an Artix 7 or whatever core as well.

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.

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

arklan
Legendary
*
Offline Offline

Activity: 1204


Just along for the ride...


View Profile
June 01, 2012, 09:14:56 AM
 #307

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.
eldentyrell
Donator
Legendary
*
Offline Offline

Activity: 966


felonious vagrancy, personified


View Profile WWW
June 01, 2012, 09:15:09 AM
 #308

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

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...

The printing press heralded the end of the Dark Ages and made the Enlightenment possible, but it took another three centuries before any country managed to put freedom of the press beyond the reach of legislators.  So it may take a while before cryptocurrencies are free of the AML-NSA-KYC surveillance plague.
DiabloD3
Legendary
*
Offline Offline

Activity: 1162


DiabloMiner author


View Profile WWW
June 01, 2012, 09:23:34 AM
 #309

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

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

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.

nedbert9
Sr. Member
****
Offline Offline

Activity: 252

Inactive


View Profile
June 01, 2012, 12:22:30 PM
 #310



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.
arklan
Legendary
*
Offline Offline

Activity: 1204


Just along for the ride...


View Profile
June 01, 2012, 12:56:04 PM
 #311

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.
norulezapply
Sr. Member
****
Offline Offline

Activity: 475


View Profile
June 01, 2012, 01:24:25 PM
 #312



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.

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)

If my post helped, I'll happily accept a few bitmills!   15rGg6A1JFZV3b7TTbtpAaiYGdUD1e1oAm
BTCurious
Hero Member
*****
Offline Offline

Activity: 714


^SEM img of Si wafer edge, scanned 2012-3-12.


View Profile
June 01, 2012, 02:20:17 PM
 #313

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? Smiley)

rjk
Sr. Member
****
Offline Offline

Activity: 420


1ngldh


View Profile
June 01, 2012, 02:25:27 PM
 #314

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

Mining Rig Extraordinaire - the Trenton BPX6806 18-slot PCIe backplane [PICS] Dead project is dead, all hail the coming of the mighty ASIC!
DiabloD3
Legendary
*
Offline Offline

Activity: 1162


DiabloMiner author


View Profile WWW
June 01, 2012, 02:32:26 PM
 #315

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? Smiley)

No, they don't use Spartan 6s.

nedbert9
Sr. Member
****
Offline Offline

Activity: 252

Inactive


View Profile
June 01, 2012, 02:54:04 PM
 #316



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.

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)

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.
rjk
Sr. Member
****
Offline Offline

Activity: 420


1ngldh


View Profile
June 01, 2012, 02:57:20 PM
 #317

Things change.
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.

Mining Rig Extraordinaire - the Trenton BPX6806 18-slot PCIe backplane [PICS] Dead project is dead, all hail the coming of the mighty ASIC!
norulezapply
Sr. Member
****
Offline Offline

Activity: 475


View Profile
June 01, 2012, 03:17:44 PM
 #318

Things change.
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.

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

If my post helped, I'll happily accept a few bitmills!   15rGg6A1JFZV3b7TTbtpAaiYGdUD1e1oAm
rjk
Sr. Member
****
Offline Offline

Activity: 420


1ngldh


View Profile
June 01, 2012, 03:38:36 PM
 #319

Things change.
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.

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

Mining Rig Extraordinaire - the Trenton BPX6806 18-slot PCIe backplane [PICS] Dead project is dead, all hail the coming of the mighty ASIC!
cablepair
Hero Member
*****
Offline Offline

Activity: 854


https://btc-republic.com/index.php?ref=cablepair


View Profile WWW
June 01, 2012, 03:50:57 PM
 #320

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


Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 [16] 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 »
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!