FPGA development board "Icarus" - DisContinued/ important announcement

[ngzhang]

How often is that error happening?

Not really a constant rate but I would say somewhere between min. 10 and max 30 times per hour.

i checked 3 board's log files with 144MB size here, and just find 8 "Read Miner ID ERROR" s. it seems really not a common error when i test them. because 144MB log file is a record of hundreds hrs. i test them under windows7 64bit.

this issue means  FPGA pushed a work result to PC, but with a wrong work id. so PC think there will be some problem and restart the script and FPGA.

i don't  know why it happy unusually frequently over there. some advise is check the USB connection or change to a USB port on the mainboard instead of at the front panel or change to another PC for a test. 

but i will fix this bug by a fully rewrite the u_mining_ctrl.v and the miner script. in order to resolve a lot of small problems and support regular miner software instead of a poor TCL script.

please keep me informed.

I'd prefer them as a quiet and power cheap alternative to GPU mining.

In my "review" I wrote that the fans made noise but I have to correct this, it was only in the first hours.
I think the fans moved during transport and the blades of one fan touched the heatsink.

Now I hardly hear the fans anymore, I think it's between 10-15 db, way less then a normal pc.

i'm planing to change the 2 X 25mm fans to 1X 40mm lower speed fan installed at the middle of 2 heatsinks. i will test this fashion on the 2nd batch.

about the heatsink, there is another issue. i must test the boards for hundreds hours before ship them, so i must install a heatsink on it. if a use thermal grease, it will get the boards dirty and negative affect when install a heatsink with 3M thermal tape, because the grease is hard to remove completely.

any suggestion?

[Dexter770221]

Toilet paper + 95% ethanol alcohol.

[server]

i checked 3 board's log files with 144MB size here, and just find 8 "Read Miner ID ERROR" s. it seems really not a common error when i test them. because 144MB log file is a record of hundreds hrs. i test them under windows7 64bit.

this issue means  FPGA pushed a work result to PC, but with a wrong work id. so PC think there will be some problem and restart the script and FPGA.

i don't  know why it happy unusually frequently over there. some advise is check the USB connection or change to a USB port on the mainboard instead of at the front panel or change to another PC for a test. 

but i will fix this bug by a fully rewrite the u_mining_ctrl.v and the miner script. in order to resolve a lot of small problems and support regular miner software instead of a poor TCL script.

Ok, I will check my cable and test this on another pc, but for your information, in the last 12 hours I counted 54 times error 26 and 52 times error 254.

i'm planing to change the 2 X 25mm fans to 1X 40mm lower speed fan installed at the middle of 2 heatsinks. i will test this fashion on the 2nd batch.

about the heatsink, there is another issue. i must test the boards for hundreds hours before ship them, so i must install a heatsink on it. if a use thermal grease, it will get the boards dirty and negative affect when install a heatsink with 3M thermal tape, because the grease is hard to remove completely.

any suggestion?

I would prefer to use Arctic Silver 5 grease but this means that you need holes in the board to hold the heatsink.
But heat is not really an issue atm ... as long as you're using the I version of XC6SLX150 that has Tj.max of 100C.
If you want to use the commercial version which has a Tj.max of 85C, then you need better way of cooling I guess.

[DeathAndTaxes]

i'm planing to change the 2 X 25mm fans to 1X 40mm lower speed fan installed at the middle of 2 heatsinks. i will test this fashion on the 2nd batch.

about the heatsink, there is another issue. i must test the boards for hundreds hours before ship them, so i must install a heatsink on it. if a use thermal grease, it will get the boards dirty and negative affect when install a heatsink with 3M thermal tape, because the grease is hard to remove completely.

any suggestion?

One larger fan is likely good idea.  No issue w/ YOU installing thermal tape but if you have standard mounting holes then the heat sink & thermal tape could be replaced different heatsink and/or thermal grease.

[ngzhang]

looks like lots of things @ the TODO list.
now some information about the 2nd batch, these days many people ask me about this via email or PM.
the 2nd batch order will start @ 1/1/2012, and ship before 7/1/2012 (dd/mm/yyyy).
the price is 569$ each, the maximum quantity of one order is 3. the package include Icarus mining board, adapter, usb cable, heatsink and fan. openbox for use.
the dev kit is sell independently, the price is 69$ each, the package include: platform cable, some other cables, software USB stick(include ISE13.2, altium10, synplify 2011.03, modelsim 10. all with crack but use as your own risk, commercial use is  illegal in my country (but legal for personal use in my country ))
all these above not include shipment fee.
no large amount discount this time, because: there are only 30 boards this batch, and i only can test 10 of them at one time. also, i wish more people to use and evaluation fpga mining.
if someone wants to build a large cluster, a even more powerful and low-priced customized product is under development and HOPEFUL can release in 2012 second half.
payment accept:
BTC, by the dynamic exchange rate, no extra fees.
mtgox $, no extra fees.
LR account, no extra fees.
paypal , add 10% charge. (because paypal charges nearly 10%)

[allinvain]

if someone wants to build a large cluster, a even more powerful and low-priced customized product is under development and HOPEFUL can release in 2012 second half.
payment accept:
BTC, by the dynamic exchange rate, no extra fees.
mtgox $, no extra fees.
LR account, no extra fees.
paypal , add 10% charge. (because paypal charges nearly 10%)

Just to be clear you mean it will be priced lower than the existing dual fpga board yet be more powerful? - ie cheaper than $569

[ngzhang]

Just to be clear you mean it will be priced lower than the existing dual fpga board yet be more powerful? - ie cheaper than $569

means the cost-performance rate, or the $/MH

[allinvain]

Awesome..looking forward to that. If you can get the cost down to $1/Mhash or lower that would be really good news.

[server]

Awesome..looking forward to that. If you can get the cost down to $1/Mhash or lower that would be really good news.

This is not realistic, I'm now getting 350-360 MH/s and the 2 Spartan6 chips alone cost over $200..

[ngzhang]

Awesome..looking forward to that. If you can get the cost down to $1/Mhash or lower that would be really good news.

This is not realistic, I'm now getting 350-360 MH/s and the 2 Spartan6 chips alone cost over $200..

Awesome..looking forward to that. If you can get the cost down to $1/Mhash or lower that would be really good news.

This is not realistic, I'm now getting 350-360 MH/s and the 2 Spartan6 chips alone cost over $200..

BFL already reached a better than 1$/1MH，  BFL are make up by aliens？

[CA Coins]

The combined BFL thread is almost 100 pages long, so not sure what is going on there.

Quick question:  What happens if the fan fails?  Does the software shut the board down?

[ngzhang]

The combined BFL thread is almost 100 pages long, so not sure what is going on there.

Quick question:  What happens if the fan fails?  Does the software shut the board down?

if there are only 1 fan fail, no serious fault will happen at correct setup. the board can survival with the other good fan.

and with 2 fan fail, i don't know what will happen, spartan6 didn't include system monitor circuit on chip. if this function is really needed, some component must add to the boards.

the PCB of Icarus will not changed (to add new functions or add mounting holes). so may be you can build a temperature sensor, and connect to Icarus with its GPIO, and write some code to let it work. this is not a complex work.

and here is the TODO list.

1,this week i will test about 8-10 boards. (7X24 hours full load mining to insure stabilize) and next week another 8-10, going on until finish up 30 borads.
2,change the miner software to TheSeven's opensource miner. a fully re-write of mining_ctrl.v is also needed. this maybe take 2-4weeks to finish.
3,improve speed . (low priority)

[CA Coins]

Thanks for the reply.  I think it would be nice to have some redundancy in case the fan dies.  I wouldn't want the FPGA to get too hot or fry the board if the fan fails. But I guess if you are keeping 2 fans and it will run with 1 fan, then I don't think that will be a problem.

The suggestion for a temperature sensor is a good one.  I will take a look when I am able to get a board from you.  However, I am not an EE/programmer so chance of success is low.  Either way, good luck with all the work you have planned!

[aTg]

Well, I love this thread but is very focused on marketing, is also an open source project and I'm sure there are people interested in replicating the circuit at home.
- Print the 6-layer PCB at home is impossible, but it is not difficult to find a manufacturer in your city.
- Welding of BGA FPGAs can do it with a small oven or using the technique "rph" not having to buy an infrared soldering station ...
- SMD soldering and finally we could do it with a heat gun or an air jet welder can cost us about 50 euros.

If I'm wrong I hope the experts correct me, I leave a list of the basic components to make the Icarus at home.



On the subject of an Icarus disconnect the system when a fan stops working, I think of anything simpler than a temperature sensor, there are fans with a third wire that sends a signal every time the fan turns around, there is only to program a loop in a microcontroller on the board or another integrated signal when it receives input 1 second, disconnect or turn off the circuit, right?

[pazor]

hello,

its a good question!

how much is the kit (PCB, components, etc.) without work of soldering ?

[ngzhang]

no matter the cost.
every one's pleasure is different, i think.
so any manually assembling question is welcome, but not include like "where/ howmuch did you buy them" or "could you buy some for me " question.

[DeathAndTaxes]

As far as overheat detection ... I think ztex indicated as the chip overheats hashing rate declines.  Since under normal conditions hashing rate shouldn't decline more than 15% a miner could be programmed to halt operation in the event hashing rate declines (i.e. don't keep feeding data to a chip running at  20MH/S cooking it). Obviously some hardware sensor would be better but mining monitoring has the advantage of not requiring any new hardware.

[ngzhang]

As far as overheat detection ... I think ztex indicated as the chip overheats hashing rate declines.  Since under normal conditions hashing rate shouldn't decline more than 15% a miner could be programmed to halt operation in the event hashing rate declines (i.e. don't keep feeding data to a chip running at  20MH/S cooking it). Obviously some hardware sensor would be better but mining monitoring has the advantage of not requiring any new hardware.

yes, you are right.
add to TODO list.

[BkkCoins]

Watching.

[ngzhang]

hit and run