Thanks, that makes sense. Maybe an older gen FPGA with lots of LUTs. Or maybe an even older gen ASIC. The older gen FPGA makes the most sense. I wonder how many chips are there wherever they came from. I have to imagine they can secure more than the tray they took a picture of to go into business, but who knows.
I apologize not reading up on all the BFL threads. It is too tedious to read. I read it when I want entertainment. I think they are getting better now that there are 2 units in the wild (1 delivered?), but they use to be so emotionally charged it's like watching a soap opera.
EDIT: So nobody knows exactly and BFL isn't telling is the answer to my question.
Topic: FPGA - 1st Quarter 2012 Summary (BFL vs. Icarus ?) (Read 5620 times)
Is it a FPGA or an ASCI?
Hint: An sASIC would use less power than an FPGA. A custom ASIC would use even less power than that.
Spartan-6 LX150 is a 45nm FPGA. It gets ~20MH/W (and is very consistent among 3 different board designs). BFL chip gets 10MH/W.
Thus BFL chip is likely?
a) 45nm Custom ASIC (should have 4x to 5x the performance per watt)
b) 45nm sASIC (should have 2x to 3x the performance per watt)
c) nextgen 28nm FPGA (should have ~2x the performance per watt)
d) 45nm FPGA (should have similar performance per watt)
e) 65nm FPGA (should have half the performance per watt)
Does anybody know for sure what chip is used in the BFL single? The 400 or so MH/s and the >40Watt power per chip doesn't quite make sense. There is some mention that it is an Altera chip but anybody know what chip exactly? Is it a FPGA or an ASCI? It is like buying a computer and not knowing who made the CPU.
There's already 2 topics with same question. For 90% it's an older gen FPGA with 200k+ LUTs. Exact model is unknown. As D&A mention BFL propably get those chips in some insane discount. Maybe from former employee?
Does anybody know for sure what chip is used in the BFL single? The 400 or so MH/s and the >40Watt power per chip doesn't quite make sense. There is some mention that it is an Altera chip but anybody know what chip exactly? Is it a FPGA or an ASCI? It is like buying a computer and not knowing who made the CPU.
Theres another important factor: resale value. BFL have almost zero resale value. It can only mine, so only miners will be interested in buying it. Icarus is a development board, cheapest on the market (with 2 LX150). Not only miners may want to buy it. And I think that 200MH/s from LX150 its not maks. BFL proved that for their product it is.
This isn't true butterfly labs also offers drivers for other purposes
Quote
Drivers available:
Computational research
Medical imaging
Packet integrity verification
Generic fingerprinting engine
Block mining (2 step SHA256)
Computational research
Medical imaging
Packet integrity verification
Generic fingerprinting engine
Block mining (2 step SHA256)
http://www.butterflylabs.com/drivers/
But if don't know what chips are on board then you will not implement your design.
And what that drivers does? How do you connect equipment for medical imaging since theres only simple USB connection to PC? Or fingerprint sensor? In FPGA you can implement many, many useful designs, but board have to be equiped at least with bunch of I/Os. Do you see any on that board?
Would be interesting to see how the competition pans out.
I debated between the various boards when Icarus was ready for its first batch, X6500 was already on the market, and no one was sure if BFL was a scam. Glad I went with ztex, now hashing happily at 4GH/S. The product is solid and he is a pro.
I debated between the various boards when Icarus was ready for its first batch, X6500 was already on the market, and no one was sure if BFL was a scam. Glad I went with ztex, now hashing happily at 4GH/S. The product is solid and he is a pro.
Theres another important factor: resale value. BFL have almost zero resale value. It can only mine, so only miners will be interested in buying it. Icarus is a development board, cheapest on the market (with 2 LX150). Not only miners may want to buy it. And I think that 200MH/s from LX150 its not maks. BFL proved that for their product it is.
This isn't true butterfly labs also offers drivers for other purposes
Quote
Drivers available:
Computational research
Medical imaging
Packet integrity verification
Generic fingerprinting engine
Block mining (2 step SHA256)
Computational research
Medical imaging
Packet integrity verification
Generic fingerprinting engine
Block mining (2 step SHA256)
http://www.butterflylabs.com/drivers/
So BFL has delivered 2 whole units and now X6500 and ztex are obsolete?
That's comedy.
Unlike BFL, both Ztex and the X6500 crew are contributors to this board (especially the X6500 crew) and have been totally upfront and honest since Day 1.
We'll see who's still around a year from now.
That's comedy.
Unlike BFL, both Ztex and the X6500 crew are contributors to this board (especially the X6500 crew) and have been totally upfront and honest since Day 1.
We'll see who's still around a year from now.
ztek board is interesting but 200MH/s is just too low in face of 800 MH/s from BFL. There is something to be said about the simplicity of a single large hashing engine. But don't count ztek out (OP I think your BFL vs Icarus comparison is too narrow).
ztek has:
highest MH/W
highest MH out of a Spartan-6 LX150 (most efficient bitstream).
the company has also been around a long time and provides non Bitcoin products (some permanence is Bitcoin market dries up).
EU regulations require them to offer a 2 year warranty.
Regular firmware and software updates boosting miner performance.
overheat detection, and floating mhz adjustment in realtime based on error rate.
The problem is 200 MH/s is just too small. It makes it very difficult to get costs down. Likely nobody is beating BFL on cost but they need to get close.
I think a 4x LX150 board would be interesting. It would have higher unit cost but power consumption should be ~35W
850 MH/s for 35W would be very impressive. Now the LX150 costs ~$150 ea in bulk so that puts $600+ just for FPGAs. In large volume the boards cost $100 over FPGA cost. A 4 FPGA unit should have some reduction in parts (i.e. 1 USB controller vs 4, 1 40W DC PSU vs 4x 10W. It may be possible to put a single large heatsink with 120mm (or 1400mm fan) across all 4 FPGAs ? reducing heatsink and fan costs.
Although it wouldn't be as cheap as BFL single, a 850 MH/s @ 35W board made by ztek for $800 would be a solid competitor. Even $900 would be viable if BFL raises their price to $700 after pre-orders.
ztek has:
highest MH/W
highest MH out of a Spartan-6 LX150 (most efficient bitstream).
the company has also been around a long time and provides non Bitcoin products (some permanence is Bitcoin market dries up).
EU regulations require them to offer a 2 year warranty.
Regular firmware and software updates boosting miner performance.
overheat detection, and floating mhz adjustment in realtime based on error rate.
The problem is 200 MH/s is just too small. It makes it very difficult to get costs down. Likely nobody is beating BFL on cost but they need to get close.
I think a 4x LX150 board would be interesting. It would have higher unit cost but power consumption should be ~35W
850 MH/s for 35W would be very impressive. Now the LX150 costs ~$150 ea in bulk so that puts $600+ just for FPGAs. In large volume the boards cost $100 over FPGA cost. A 4 FPGA unit should have some reduction in parts (i.e. 1 USB controller vs 4, 1 40W DC PSU vs 4x 10W. It may be possible to put a single large heatsink with 120mm (or 1400mm fan) across all 4 FPGAs ? reducing heatsink and fan costs.
Although it wouldn't be as cheap as BFL single, a 850 MH/s @ 35W board made by ztek for $800 would be a solid competitor. Even $900 would be viable if BFL raises their price to $700 after pre-orders.
Yeah. This is music to my ears. A EU based BFL competitor is desperately needed.
4 chips on one big FPGA board = hardcore mining !
Ah, "der Brutzler" is back 
. As far as i know, you are the only one that destroyed 2 boards in a row. Ha ha
. As far as i know, you are the only one that destroyed 2 boards in a row. Ha haYeah, think of it as an extreme quality control.
German is weak, US and chinese is ok.
It's ZTEX's fault ...he should take my advice and put this label on all the boxes he ships.
Ah, "der Brutzler" is back . As far as i know, you are the only one that destroyed 2 boards in a row. Ha ha
. As far as i know, you are the only one that destroyed 2 boards in a row. Ha ha
Ah, "der Brutzler" is back . As far as i know, you are the only one that destroyed 2 boards in a row. Ha ha
. As far as i know, you are the only one that destroyed 2 boards in a row. Ha ha
Gusti,
U're so wrong....
U're so wrong....
Ztex is a very nice guy and very helpful...
Gusty,if you don't know what a power source is,then stay away from electronics ( eg: FPGA boards).
Don't blame ZTEX for your 10 year old kid's knowledge OR tests?!
Gusty,if you don't know what a power source is,then stay away from electronics ( eg: FPGA boards).
Don't blame ZTEX for your 10 year old kid's knowledge OR tests?!
Thanks for your insights, Stefan.
ztek board is interesting but 200MH/s is just too low in face of 800 MH/s from BFL. There is something to be said about the simplicity of a single large hashing engine. But don't count ztek out (OP I think your BFL vs Icarus comparison is too narrow).
ztek has:
highest MH/W
highest MH out of a Spartan-6 LX150 (most efficient bitstream).
the company has also been around a long time and provides non Bitcoin products (some permanence is Bitcoin market dries up).
EU regulations require them to offer a 2 year warranty.
Regular firmware and software updates boosting miner performance.
overheat detection, and floating mhz adjustment in realtime based on error rate.
The problem is 200 MH/s is just too small. It makes it very difficult to get costs down. Likely nobody is beating BFL on cost but they need to get close.
I think a 4x LX150 board would be interesting. It would have higher unit cost but power consumption should be ~35W
850 MH/s for 35W would be very impressive. Now the LX150 costs ~$150 ea in bulk so that puts $600+ just for FPGAs. In large volume the boards cost $100 over FPGA cost. A 4 FPGA unit should have some reduction in parts (i.e. 1 USB controller vs 4, 1 40W DC PSU vs 4x 10W. It may be possible to put a single large heatsink with 120mm (or 1400mm fan) across all 4 FPGAs ? reducing heatsink and fan costs.
Although it wouldn't be as cheap as BFL single, a 850 MH/s @ 35W board made by ztek for $800 would be a solid competitor. Even $900 would be viable if BFL raises their price to $700 after pre-orders.
ztek has:
highest MH/W
highest MH out of a Spartan-6 LX150 (most efficient bitstream).
the company has also been around a long time and provides non Bitcoin products (some permanence is Bitcoin market dries up).
EU regulations require them to offer a 2 year warranty.
Regular firmware and software updates boosting miner performance.
overheat detection, and floating mhz adjustment in realtime based on error rate.
The problem is 200 MH/s is just too small. It makes it very difficult to get costs down. Likely nobody is beating BFL on cost but they need to get close.
I think a 4x LX150 board would be interesting. It would have higher unit cost but power consumption should be ~35W
850 MH/s for 35W would be very impressive. Now the LX150 costs ~$150 ea in bulk so that puts $600+ just for FPGAs. In large volume the boards cost $100 over FPGA cost. A 4 FPGA unit should have some reduction in parts (i.e. 1 USB controller vs 4, 1 40W DC PSU vs 4x 10W. It may be possible to put a single large heatsink with 120mm (or 1400mm fan) across all 4 FPGAs ? reducing heatsink and fan costs.
Although it wouldn't be as cheap as BFL single, a 850 MH/s @ 35W board made by ztek for $800 would be a solid competitor. Even $900 would be viable if BFL raises their price to $700 after pre-orders.
Ztex is a very nice guy and very helpful...
Gusty,if you don't know what a power source is,then stay away from electronics ( eg: FPGA boards).
Don't blame ZTEX for your 10 year old kid's knowledge OR tests?!
Gusty,if you don't know what a power source is,then stay away from electronics ( eg: FPGA boards).
Don't blame ZTEX for your 10 year old kid's knowledge OR tests?!
Also ztex has a warranty and its made by respectable company in Germany.
Not at all, that was not my own experience.Both Icarus and X6500 manufacturers, gave me far better support and respect than ztex.
I do NOT recommend going with ztex if you have another option.
cgminer 2.3.0 supports icarus/
That is a huge bonus. Good to see.
X6500: https://bitcointalksearch.org/topic/x6500-custom-fpga-miner-40058
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
Unfortunately I don't see either of those 2 products/projects as a very attractive solution as they really don't stand out in any way, but rather they are simply too average (or not extreme enough in a specific area....like, no focus)....?
bitlane,
a x6500 is as good as Icarus, performs as well and costs as much, so why don't you consider it as a good solution?
spiccioli.
Oh that. That is because of his ignorance or bias.
Also ztex has a warranty and its made by respectable company in Germany.
Theres another important factor: resale value. BFL have almost zero resale value. It can only mine, so only miners will be interested in buying it. Icarus is a development board, cheapest on the market (with 2 LX150). Not only miners may want to buy it. And I think that 200MH/s from LX150 its not maks. BFL proved that for their product it is.
cgminer 2.3.0 supports icarus/
That is a huge bonus. Good to see.
X6500: https://bitcointalksearch.org/topic/x6500-custom-fpga-miner-40058
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
Unfortunately I don't see either of those 2 products/projects as a very attractive solution as they really don't stand out in any way, but rather they are simply too average (or not extreme enough in a specific area....like, no focus)....?
bitlane,
a x6500 is as good as Icarus, performs as well and costs as much, so why don't you consider it as a good solution?
spiccioli.
Ztex was always one step ahead. He still has the fastest board per chip. It is expensive but this boards and the software are (were) proven to work from the start. It's the most professional product in my eyes. But the others are catching up
BFL had to add a second fan due to overheating, and even then it starts slowing down if room temperature isn't cool. This does not bode well long-term for them.
Icarus looks like a far better engineered product that will stay profitable a lot longer. Icarus also has a volume discount.
It comes down to how long do you plan on mining before you throw the FPGA away? 6 months? Then BFL if you can get one. 1+ years? Icarus.
Icarus looks like a far better engineered product that will stay profitable a lot longer. Icarus also has a volume discount.
It comes down to how long do you plan on mining before you throw the FPGA away? 6 months? Then BFL if you can get one. 1+ years? Icarus.
cgminer 2.3.0 supports icarus/
That is a huge bonus. Good to see.
X6500: https://bitcointalksearch.org/topic/x6500-custom-fpga-miner-40058
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
Unfortunately I don't see either of those 2 products/projects as a very attractive solution as they really don't stand out in any way, but rather they are simply too average (or not extreme enough in a specific area....like, no focus)....?
The BitForce Single and the Icarus both do SOMETHING extremely well, on opposite sides of the 'spectrum' and are great competitors both on paper and in use. 1 being extremely power-usage-friendly for the performance achieved (the Icarus) and the other being extremely powerful for the money (Butterfly Labs).
Neither the Ztex nor X6500 units are real stand-outs unfortunately and unless something drastic were to happen (as in major price cuts, major performance increases or something rediculous such as single-digit max (W) power-draw for existing performance figures) with either project/product, I really couldn't see too many people ordering them over the other 2 above.....it simply wouldn't make sense.
You should also consider availability. Icarus is available right now (I think ?) and BFL's delivery date is kind of uncertain. Shipping costs are also important for people who live outside the US/China.
The BFL units have clearly began shipping, as there are a couple users on this forum who are currently running their pre-order units.
Availability is the one thing that could really KILL the BFL product, because they clearly have the 'bang-for-buck' covered....now they just need to come up with quantity and immediate availability.
Regardless of the BFL's raw power, I still am interested in the Icarus due to it's low power consumption and think that along with slight performance increases, the Icarus project should really focus on overall power draw, because I don't think they will be capable of competing with Butterfly Labs for outright horsepower on a single unit-vs-unit basis.....
My 2 bitcents...
bitlane.
Dont forget,none but the BFL has ANY warranty.At least BFL offers a 6 month & Ztex offers up to 2 years,maybe,Icarus & X6500 have NONE.So for $600 my money is on BFL
cgminer 2.3.0 supports icarus/
That is a huge bonus. Good to see.
X6500: https://bitcointalksearch.org/topic/x6500-custom-fpga-miner-40058
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
ztex: https://bitcointalksearch.org/topic/ztex-usb-fpga-modules-115x-and-115y-215-and-860-mhs-fpga-boards-49180
Unfortunately I don't see either of those 2 products/projects as a very attractive solution as they really don't stand out in any way, but rather they are simply too average (or not extreme enough in a specific area....like, no focus)....?
The BitForce Single and the Icarus both do SOMETHING extremely well, on opposite sides of the 'spectrum' and are great competitors both on paper and in use. 1 being extremely power-usage-friendly for the performance achieved (the Icarus) and the other being extremely powerful for the money (Butterfly Labs).
Neither the Ztex nor X6500 units are real stand-outs unfortunately and unless something drastic were to happen (as in major price cuts, major performance increases or something rediculous such as single-digit max (W) power-draw for existing performance figures) with either project/product, I really couldn't see too many people ordering them over the other 2 above.....it simply wouldn't make sense.
You should also consider availability. Icarus is available right now (I think ?) and BFL's delivery date is kind of uncertain. Shipping costs are also important for people who live outside the US/China.
The BFL units have clearly began shipping, as there are a couple users on this forum who are currently running their pre-order units.
Availability is the one thing that could really KILL the BFL product, because they clearly have the 'bang-for-buck' covered....now they just need to come up with quantity and immediate availability.
Regardless of the BFL's raw power, I still am interested in the Icarus due to it's low power consumption and think that along with slight performance increases, the Icarus project should really focus on overall power draw, because I don't think they will be capable of competing with Butterfly Labs for outright horsepower on a single unit-vs-unit basis.....
My 2 bitcents...
bitlane.
380 MH/s
20W
$570 USD (shipping not included)
**Specific Software (so far)
20W
$570 USD (shipping not included)
**Specific Software (so far)
cgminer 2.3.0 supports icarus/
You should also consider availability. Icarus is available right now (I think ?) and BFL's delivery date is kind of uncertain. Shipping costs are also important for people who live outside the US/China.
So, just to get this right, as I might be a bit 'behind' the times as it were.....
As of February 2012, it seems that we clearly have 2 winning FPGA Designs/Projects that are essentially 'Production' ready and each have quite opposite strengths and merits worth taking note of.
Please correct me if anything below is WRONG.
-------------------------------------
Butterfly Labs BitForce Single
830 MH/s
85W
$600 USD (shipping not included)
**Sexy Software Support (CGMiner etc)
-------------------------------------
$0.72/MegaHash ..or.. 1.38 MegaHash/$Dollar
0.102W/MegaHash ..or.. 9.76 MegaHash/W
.......VS.......
-------------------------------------
ngzhang's "Icarus"
380 MH/s
20W
$570 USD (shipping not included)
**Specific Software (so far) Now with CGMiner 2.3.0 support.
-------------------------------------
$1.50/MegaHash ..or.. 0.67 MegaHash/$Dollar
0.053W/MegaHash ..or.. 19 MegaHash/W
I guess it's almost safe to say that when/IF choosing between the 2 designs/projects above, POWER CONSUMPTION, POWER PRICE and INITIAL INVESTMENT might be what it all comes down to.....(ie. Is your power more expensive over X amount of time, making the option of purchasing a BitForce Single less cost effective than purchasing a pair of Icarus units even though 'out of pocket' is more ?).....
Comparing them at face value....
BitForce = 2x MegaHash/$Dollar
Icarus = 2x MegaHash/Watt
I think for current GPU Miners, the Bitforce is the clear winner....but what about NEW Miners ? Might Icarus keep reoccurring costs down (even though initial is higher) due to ever increasing difficulty and decreasing mining profit ?
Is there anything else on the horizon even worth a mention ?
As of February 2012, it seems that we clearly have 2 winning FPGA Designs/Projects that are essentially 'Production' ready and each have quite opposite strengths and merits worth taking note of.
Please correct me if anything below is WRONG.
-------------------------------------
Butterfly Labs BitForce Single
830 MH/s
85W
$600 USD (shipping not included)
**Sexy Software Support (CGMiner etc)
-------------------------------------
$0.72/MegaHash ..or.. 1.38 MegaHash/$Dollar
0.102W/MegaHash ..or.. 9.76 MegaHash/W
.......VS.......
-------------------------------------
ngzhang's "Icarus"
380 MH/s
20W
$570 USD (shipping not included)
**Specific Software (so far)
-------------------------------------
$1.50/MegaHash ..or.. 0.67 MegaHash/$Dollar
0.053W/MegaHash ..or.. 19 MegaHash/W
I guess it's almost safe to say that when/IF choosing between the 2 designs/projects above, POWER CONSUMPTION, POWER PRICE and INITIAL INVESTMENT might be what it all comes down to.....(ie. Is your power more expensive over X amount of time, making the option of purchasing a BitForce Single less cost effective than purchasing a pair of Icarus units even though 'out of pocket' is more ?).....
Comparing them at face value....
BitForce = 2x MegaHash/$Dollar
Icarus = 2x MegaHash/Watt
I think for current GPU Miners, the Bitforce is the clear winner....but what about NEW Miners ? Might Icarus keep reoccurring costs down (even though initial is higher) due to ever increasing difficulty and decreasing mining profit ?
Is there anything else on the horizon even worth a mention ?
Jump to: