Pages:
Author

Topic: FPGA development board "Icarus" - DisContinued/ important announcement - page 48. (Read 207285 times)

rph
full member
Activity: 176
Merit: 100
how to compete with scammers?

Take the GH number they're promising and double it at the same cost & power.
Promise a 13-20 week ship date allowing you to use 28nm engineering sample FPGAs.
Then, start charging credit cards for pre-orders right away, and send the buyers a couple pics of empty PCBs
so they don't dispute the charge before the cutoff. Congratulations, now you are a proper Bitcoin scammer!  Grin

-rph
donator
Activity: 1218
Merit: 1079
Gerald Davis
How is your board going to compete with BFL Labs ? Hopefully they are not scammers.

how to compete with scammers?

LOLZ
hero member
Activity: 592
Merit: 501
We will stand and fight.
How is your board going to compete with BFL Labs ? Hopefully they are not scammers.

how to compete with scammers?
hero member
Activity: 518
Merit: 500
How is your board going to compete with BFL Labs ? Hopefully they are not scammers.
hero member
Activity: 592
Merit: 501
We will stand and fight.
Icarus will soon open pre-order.

the pre-order fee is 1 BTC. and this 1BTC will change to a gift pack for the first batch buyer.

hero member
Activity: 592
Merit: 501
We will stand and fight.
no, my mining software is a simple TCL script, and the FPGA config bitstream is stored in the on board flash, which will boot the FPGA when powerup in 500ms.

That is a nice setup.  Can the onboard-flash be reprogrammed in the field if you improve the bitstream or the flash becomes corrupt?

certainly, by using a XILINX platform cable and the Labtooles software.
donator
Activity: 1218
Merit: 1079
Gerald Davis
no, my mining software is a simple TCL script, and the FPGA config bitstream is stored in the on board flash, which will boot the FPGA when powerup in 500ms.

That is a nice setup.  Can the onboard-flash be reprogrammed in the field if you improve the bitstream or the flash becomes corrupt?
hero member
Activity: 592
Merit: 501
We will stand and fight.
Is this or any FPGA board going to be usable to someone without an electrical engineering or computer science degree?  Just reading FPGA threads makes my head hurt.

  There is very little end-user work to be done on this and several other fpga projects out now.  Being that it is USB means that the developer has likely done the work of programming their mining software to program the chip for you when you run it. Ng will have to verify his software does this, but I know for sure the Ztex software does and if BFL ever release a product their's will be plug and play as well.

  So yes, very usable to common folks like us without PHDs. ;p

no, my mining software is a simple TCL script, and the FPGA config bitstream is stored in the on board flash, which will boot the FPGA when powerup in 500ms.
hero member
Activity: 714
Merit: 500
Hope the price rise in 2012, then FPGA will dominate hasing farms.
hero member
Activity: 504
Merit: 500
Is this or any FPGA board going to be usable to someone without an electrical engineering or computer science degree?  Just reading FPGA threads makes my head hurt.

  There is very little end-user work to be done on this and several other fpga projects out now.  Being that it is USB means that the developer has likely done the work of programming their mining software to program the chip for you when you run it. Ng will have to verify his software does this, but I know for sure the Ztex software does and if BFL ever release a product their's will be plug and play as well.

  So yes, very usable to common folks like us without PHDs. ;p
legendary
Activity: 1022
Merit: 1000
BitMinter
$600 for 360MH/s.  Pretty impressive. 

Not where electricity is pretty much dead cheap and where 5870s are available !

Maybe when you hash at 10 MH/s... Cheesy
donator
Activity: 798
Merit: 500
Is this or any FPGA board going to be usable to someone without an electrical engineering or computer science degree?  Just reading FPGA threads makes my head hurt.
hero member
Activity: 518
Merit: 500
$600 for 360MH/s.  Pretty impressive. 

Not where electricity is pretty much dead cheap and where 5870s are available !
donator
Activity: 1218
Merit: 1079
Gerald Davis
$600 for 360MH/s.  Pretty impressive. 
hero member
Activity: 592
Merit: 501
We will stand and fight.
 Grin heated disscussion.

Those are pretty cool. Would love to know what the final price is going to be.

  He gave us enough of a range on price to do some math with. Less than $648 and more than $550(since he stated $500 was a bit less than his costs).

  I am still gonna stick to within 10% or so of my orig speculation once he crunches all the math on parts and labor costs involved. Unless he plans to order 500+ chips at once and contracts with an ultra cheap assembler in hopes you forum peoples will buy them before a cheaper design can be built.....

less than 600$, first batch.

  That still puts X6500 more expensive. ;p  Not tring to argue with ya, m8. But your orginal comment of it in comparison tot he x6500 is off a bit.  Also, X6500 started out at 200. Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

  Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

It won't be better (from a purely myopic $/MH basis) unless it's equal to or lower than the x6500 unless he undercuts the x6500 price. Which is kinda sad because it looks like a much nicer board (for reasons unrelated to $/MH).

Or would you like to take a bet that a board with the _lower_ speed grade version of the same part is going to deliver higher performance?   I might take the opposite of that bet.


Icrus uses a  -2I fpga, "I " means the industrial temperature. speed/voltage spec are guaranteed across the industrial temperature range. by my testing, -2I  is very alike with -3C, and the price is also the same, bug much easier to buy in my country.


Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

Well not exactly the same boat.  The theoretical performance for a LX150 is ~200MH/s.

Ztek is getting 190MH/s on 1x LX150 or 90% of peak.
Icarus is getting 360MH/s on 2x LX150 or 80% of peak.
X6500 is getting 250MH/s on 2x LX150 or 62.5% of peak.

So X6500 has the POTENTIAL to make larger increases in performance and thus larger changes in MH/$.  Ztek is pretty much at peak performance.  Looks like he is looking at changing the voltage that might pick up another ~4 or 5 MH but it is pretty optomized.  Icarus could pick up some speed (and thus moderate reduction in MH/$) but not much.

Granted I wouldn't buy a board only on potential but the X6500 is interesting simply because it has a low cost per FPGA which potentially could delivery higher performance per $.  Still they got their work cut out for them because all that really matters in the end if MH/$.

Quote
Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

Hinted but I would like to see a firm price list w/ volume pricing.

Of course if BF Labs 1GH/s for $500 isn't a scam well that pretty much kills all these projects.

I think a well designed PCB, enhanced on board power regulator, and high-quality assembling is also important for a long run. and that's why people do not buy the cheapest brands of GPUs / PSUs for mining.
staff
Activity: 4284
Merit: 8808
 Unless I am really missing something in my math here, it IS better from a $/MH at the price we have to go by assuming Icarus price.
  Can you show me how $580/250 is cheaper than $648/360?

  X6500= $580/250MH= $2.32 per MH
  Icarus= $648/360MH= $1.80 per MH

  If, Ng's reply to me was misunderstood and the buyer price is higher, it would have to be  $835.2 to be the same price as X6500 per MH.  That's  $2.32*360MH=$835.20

  I'm not sure where we are misunderstanding each other at here. I am completely non-biased to either of these two projects. I think every one of the guys working on FPGA builds are absolutly amazing!

Because $580/380 is $1.52.  380 is the speed we should expect from two of the the -3 parts, because thats the speed that the currently best available open source code typically gets on them.

It would be surprising if the X6500 is unable to reach the same speed ZTEX is getting with _the same part_ after adopting the same double stage pipeline design spending the couple hours brute-forcing the optimizer settings to get it to route for 200mhz. It might be the case that the -2 part also ends up just as fast due to the vulgarities of binning, but I know of no reason to expect it to be to be faster.

(and keep in mind, of these three the ztex is the only one you can click-and-buy for immediate shipment right _now_ so if you want to talk about dollars per MH _this moment_ rather than what we can reasonably expect— then Icarus and X6500 are both ~$∞/mh. Smiley )

And yes, agreed. It's all good work. And regardless of how the $/mh ultimately works out when the boards are generally available. The Icarus boards look like they'll be an excellent value due to their increase flexibility.
hero member
Activity: 504
Merit: 500
 That still puts X6500 more expensive. ;p  Not trying to argue with ya, m8. But your orginal comment of it in comparison to the x6500 is off a bit.  Also, X6500 started out at 200. Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

  Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

It won't be better (from a purely myopic $/MH basis) unless it's equal to or lower than the x6500 unless he undercuts the x6500 price. Which is kinda sad because it looks like a much nicer board (for reasons unrelated to $/MH).

Or would you like to take a bet that a board with the _lower_ speed grade version of the same part is going to deliver higher performance?   I might take the opposite of that bet.

  Unless I am really missing something in my math here, it IS better from a $/MH at the price we have to go by assuming Icarus price.
  Can you show me how $580/250 is cheaper than $648/360?

  X6500= $580/250MH= $2.32 per MH
  Icarus= $648/360MH= $1.80 per MH

  If, Ng's reply to me was misunderstood and the buyer price is higher, it would have to be  $835.2 to be the same price as X6500 per MH.  That's  $2.32*360MH=$835.20

  I'm not sure where we are misunderstanding each other at here. I am completely non-biased to either of these two projects. I think every one of the guys working on FPGA builds are absolutly amazing!

  Kind regards,
      Derek
hero member
Activity: 504
Merit: 500
Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

Well not exactly the same boat.  The theoretical performance for a LX150 is ~200MH/s.
 Quite right you are. By same boat, I was referring more to the initial public sales release aspect.

Ztek is getting 190MH/s on 1x LX150 or 90% of peak.
Icarus is getting 360MH/s on 2x LX150 or 80% of peak.
X6500 is getting 250MH/s on 2x LX150 or 62.5% of peak.
 In my haste I had neglected to view the Ic/X from this perspective of potential. Thanks for taking the time to sum that up.

So X6500 has the POTENTIAL to make larger increases in performance and thus larger changes in MH/$.  Ztek is pretty much at peak performance.  Looks like he is looking at changing the voltage that might pick up another ~4 or 5 MH but it is pretty optomized.  Icarus could pick up some speed (and thus moderate reduction in MH/$) but not much.

Granted I wouldn't buy a board only on potential but the X6500 is interesting simply because it has a low cost per FPGA which potentially could delivery higher performance per $.  Still they got their work cut out for them because all that really matters in the end if MH/$.
 Yep and yep. =)

Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

Hinted but I would like to see a firm price list w/ volume pricing.
Of course if BF Labs 1GH/s for $500 isn't a scam well that pretty much kills all these projects.

  He did a bit more than hint when he replied to my price speculation with. "relax, it will not reach $648.  Grin"
  Volume price for parts is very easy to evaluate but we've no idea what he would want per license if he chooses to keep it as a private 'for sale' project.

  Frigging BF Labs, the 10 ton unicorn in the room for sure.  They better put some fire under it one way or the other. I'm about 1 quote away from going ahead with my endeavours without them.  Wink



    Cheers m8
donator
Activity: 1218
Merit: 1079
Gerald Davis
Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

Well not exactly the same boat.  The theoretical performance for a LX150 is ~200MH/s.

Ztek is getting 190MH/s on 1x LX150 or 90% of peak.
Icarus is getting 360MH/s on 2x LX150 or 80% of peak.
X6500 is getting 250MH/s on 2x LX150 or 62.5% of peak.

So X6500 has the POTENTIAL to make larger increases in performance and thus larger changes in MH/$.  Ztek is pretty much at peak performance.  Looks like he is looking at changing the voltage that might pick up another ~4 or 5 MH but it is pretty optomized.  Icarus could pick up some speed (and thus moderate reduction in MH/$) but not much.

Granted I wouldn't buy a board only on potential but the X6500 is interesting simply because it has a low cost per FPGA which potentially could delivery higher performance per $.  Still they got their work cut out for them because all that really matters in the end if MH/$.

Quote
Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

Hinted but I would like to see a firm price list w/ volume pricing.

Of course if BF Labs 1GH/s for $500 isn't a scam well that pretty much kills all these projects.
staff
Activity: 4284
Merit: 8808
  That still puts X6500 more expensive. ;p  Not tring to argue with ya, m8. But your orginal comment of it in comparison tot he x6500 is off a bit.  Also, X6500 started out at 200. Icarus is in the same boat and has been stated as such by the developer that he is still working on optimizing the mining code.

  Point is, from the range Ng has hinted to thus far, his mark-up will be much less than others and his end-user price per MH is better...

It won't be better (from a purely myopic $/MH basis) unless it's equal to or lower than the x6500 unless he undercuts the x6500 price. Which is kinda sad because it looks like a much nicer board (for reasons unrelated to $/MH).

Or would you like to take a bet that a board with the _lower_ speed grade version of the same part is going to deliver higher performance?   I might take the opposite of that bet.
Pages:
Jump to: