Author

Topic: [ANN][BLC] Blakecoin Blake-256 for GPU/FPGA With Merged Mined Pools Stable Net - page 135. (Read 409571 times)

legendary
Activity: 1509
Merit: 1030
Solutions Architect
See this is how I know this is a good plan, there is room for improvement. I don't even know if it's possible, but I have the feeling this is where crypto currencies could evolve towards!

have you ever used Xilinx ISE to do a build for FPGA or do you use the pre-built bitstreams only?

No, only used pre-built stuff. That's why I don't know, if I knew I would have tried to build a "dynamic" hashing algo already.

good luck you are going to need it plus without independent test you could make an algo that is flawed

try building from source on FPGA before you get too exited about the wild idea Wink
hero member
Activity: 725
Merit: 503
See this is how I know this is a good plan, there is room for improvement. I don't even know if it's possible, but I have the feeling this is where crypto currencies could evolve towards!

have you ever used Xilinx ISE to do a build for FPGA or do you use the pre-built bitstreams only?

No, only used pre-built stuff. That's why I don't know, if I knew I would have tried to build a "dynamic" hashing algo already.

But I can imagine you need some core elements, that could be provided as inputs to the "ASIC" to be required inside the "ASIC" for performance reasons.

"ASIC" = the bitstream
legendary
Activity: 1509
Merit: 1030
Solutions Architect
See this is how I know this is a good plan, there is room for improvement. I don't even know if it's possible, but I have the feeling this is where crypto currencies could evolve towards!

have you ever used Xilinx ISE to do a build for FPGA or do you use the pre-built bitstreams only?

crypto currencies should just stick to the work by academics in the field and try to use best practice if possible  Cheesy
hero member
Activity: 725
Merit: 503
if you mod the algo dynamically like you are saying its not going to be power efficient as you are trying to create resistance  Roll Eyes

and if you can put it on FPGA then some part of it could be put into silicon!

Yes, but if you modify the bitstream you would have to make a new ASIC. So I suggest inventing a new hashing algo that requires a new bitstream every difficulty change (or so) not like quark (that uses a set of known algos) but a totally new and fundamentally dynamic algo.

quark uses an awful waterfall method for its algo's its terribly inefficient!  

try building the bitstream from source via Xilinx ISE and then think about it again Roll Eyes

takes days/weeks to get a good fmax build  Shocked

See this is how I know this is a good plan, there is room for improvement. I don't even know if it's possible, but I have the feeling this is where crypto currencies could evolve towards! We need hashing, so we should improve hashing!
legendary
Activity: 1509
Merit: 1030
Solutions Architect

not sure if you are looking at same graphs as me but the price is set by the market and atm it is super cheap!  Undecided

But I guess Bitcoin had this in the early days too 10k BTC pizza for example Wink
legendary
Activity: 1509
Merit: 1030
Solutions Architect
if you mod the algo dynamically like you are saying its not going to be power efficient as you are trying to create resistance  Roll Eyes

and if you can put it on FPGA then some part of it could be put into silicon!

Yes, but if you modify the bitstream you would have to make a new ASIC. So I suggest inventing a new hashing algo that requires a new bitstream every difficulty change (or so) not like quark (that uses a set of known algos) but a totally new and fundamentally dynamic algo.

quark uses an awful waterfall method for its algo's its terribly inefficient!  

try building the bitstream from source via Xilinx ISE and then think about it again Roll Eyes

takes days/weeks to get a good fmax build  Shocked
hero member
Activity: 725
Merit: 503
if you mod the algo dynamically like you are saying its not going to be power efficient as you are trying to create resistance  Roll Eyes

and if you can put it on FPGA then some part of it could be put into silicon!

Yes, but if you modify the bitstream you would have to make a new ASIC. So I suggest inventing a new hashing algo that requires a new bitstream every difficulty change (or so) not like quark (that uses a set of known algos) but a totally new and fundamentally dynamic algo.
legendary
Activity: 1509
Merit: 1030
Solutions Architect
so when scrypt/asics come out and when GPU's become redundant for scrypt mining. Do you think people will turn to blake-256?

edit: scratch that thought, the new FPGAS will take over and be an ASIC like race  all over again.

Just look at the name Asic = Application-specific integrated circuit Roll Eyes

No algo can avoid getting an Asic if the demand is there  Wink

That's why I have always said it is pointless to be resistant, if an Asic was done for Blake-256 due to not trying to be resistant it would be a smaller chip that is cheaper to make and uses less power yet would also have a fast hashing rate e.g. it would be faster than the current SHA-256 at the same manufacturing process

Blake-256 = Small, Fast, Simple, Power efficient  Grin

I also like FPGA's they are re-programmable and lots of ex-SHA-256 boards about that can be reused for Blake-256  Cool

This is where I think one should just appreciate the flexibility and power efficiency of FPGA and build a coin that modifies the hashing algo in such a fundamental way every difficulty change that it can't be made into an ASIC.

Blakecoin is a great hedge against weaknesses in SHA256, that's all really (just like any altcoin).

if you mod the algo dynamically like you are saying its not going to be power efficient as you are trying to create resistance  Roll Eyes

and if you can put it on FPGA then some part of it could be put into silicon!

not seen another coin that tries to be efficient they all follow the resistance path  Huh

Scrypt/Skein/Keccak/and multi algo's hashes like QRK none of these look at power usage or hash/watt really  Cry

Blake
"Advantages
Design
• simplicity of the algorithm
• interface for hashing with a salt

Performance
• fast in both software and hardware
• parallelism and throughput/area trade-off for hardware implementation
• simple speed/confidence trade-off with the tunable number of rounds

Security
• based on an intensively analyzed component (ChaCha)
• resistant to generic second-preimage attacks
• resistant to side-channel attacks
• resistant to length-extension"  
https://131002.net/blake/blake.pdf

Tuned the rounds to 8 and that is the Blake-256 algo used in Blakecoin

I took the the adequate approach to the security buffer 2200! which djb talks about in one of his latest talks: http://cr.yp.to/talks/2014.01.18/slides-dan+tanja-20140118-a4.pdf  <-- see page 60
hero member
Activity: 725
Merit: 503
so when scrypt/asics come out and when GPU's become redundant for scrypt mining. Do you think people will turn to blake-256?

edit: scratch that thought, the new FPGAS will take over and be an ASIC like race  all over again.

Just look at the name Asic = Application-specific integrated circuit Roll Eyes

No algo can avoid getting an Asic if the demand is there  Wink

That's why I have always said it is pointless to be resistant, if an Asic was done for Blake-256 due to not trying to be resistant it would be a smaller chip that is cheaper to make and uses less power yet would also have a fast hashing rate e.g. it would be faster than the current SHA-256 at the same manufacturing process

Blake-256 = Small, Fast, Simple, Power efficient  Grin

I also like FPGA's they are re-programmable and lots of ex-SHA-256 boards about that can be reused for Blake-256  Cool

This is where I think one should just appreciate the flexibility and power efficiency of FPGA and build a coin that modifies the hashing algo in such a fundamental way every difficulty change that it can't be made into an ASIC.

Blakecoin is a great hedge against weaknesses in SHA256, that's it really (just like any non SHA256 altcoin). But I agree that litecoin miners will probably move to any non SHA256 or scrypt mining if the price of those currencies doesn't rise as that hedge is worth something.
sr. member
Activity: 384
Merit: 250
Ok, so I got it working but now it won't get shares from ny1.blakecoin.com:8337.

It connects, sees longpoll but then doesn't get shares.

It was never tested against a pool. Atavacron got it working solo (in the pre-pool days), see from here to here.

To get this debugged, I suggest the first thing to check is that the FPGA is returning valid hashes. So you'll need to uncomment some of the debug output in mine.py (lines 91-102) which prints out the getwork data received from the pool, and the nonce returned from the FPGA, then plugs that into checkblake

There may be an issue with difficulty/target values, though these are now the same as bitcoin, so perhaps this is unlikely. Anyway, see how you get on with that for now.

[Aside: the X6500 port was only ever very experimental as it was done blind. Getting it to work with cgminer will be even more difficult as there was never a driver written for it, though bfgminer does have one, but that's another can of worms to either port it to cgminer or get bfgminer working with blakecoin].
legendary
Activity: 1509
Merit: 1030
Solutions Architect
so when scrypt/asics come out and when GPU's become redundant for scrypt mining. Do you think people will turn to blake-256?

edit: scratch that thought, the new FPGAS will take over and be an ASIC like race  all over again.

GPU's will still mine Blake-256 for some time to come, FPGA don't beat GPU's for speed only power usage!

Just look at the name Asic = Application-specific integrated circuit Roll Eyes

No algo can avoid getting an Asic if the demand is there  Wink

That's why I have always said it is pointless to be resistant, if an Asic was done for Blake-256 due to not trying to be resistant it would be a smaller chip that is cheaper to make and uses less power yet would also have a fast hashing rate e.g. it would be faster than the current SHA-256 at the same manufacturing process

Blake-256 = Small, Fast, Simple, Power efficient  Grin

I also like FPGA's they are re-programmable and lots of ex-SHA-256 boards about that can be reused for Blake-256  Cool
full member
Activity: 132
Merit: 100
BitCoin
hero member
Activity: 644
Merit: 500
P2P The Planet!
so when scrypt/asics come out and when GPU's become redundant for scrypt mining. Do you think people will turn to blake-256?

edit: scratch that thought, the new FPGAS will take over and be an ASIC like race  all over again.
legendary
Activity: 1140
Merit: 1000
The Real Jude Austin
I've not able to even get the driver running need to retrace my steps again. might move to ubuntu if no luck still trying win764 and a x6500.

works a sha256 miner with cgminer. do ineed to unload something from that?

Let me know how it works for you on Linux.

legendary
Activity: 1470
Merit: 1001
Use Coinbase Account almosanywhere with Shift card
I've not able to even get the driver running need to retrace my steps again. might move to ubuntu if no luck still trying win764 and a x6500.

works a sha256 miner with cgminer. do ineed to unload something from that?
legendary
Activity: 1140
Merit: 1000
The Real Jude Austin
Ok, so I got it working but now it won't get shares from ny1.blakecoin.com:8337.

It connects, sees longpoll but then doesn't get shares.

Code:
Long-poll: connected to ny1.blakecoin.com:8337
0: Error getting work! Retrying...
1: Error getting work! Retrying...
0: Error getting work! Retrying...
1: Error getting work! Retrying...
0: Error getting work! Retrying...
1: Error getting work! Retrying...
0: Error getting work! Retrying...
1: Error getting work! Retrying...
0: Error getting work! Retrying...
1: Error getting work! Retrying...
0: Error getting work! Retrying...

Haha, I just cant win!
legendary
Activity: 1509
Merit: 1030
Solutions Architect
Just spam the launch string a few more times..
Code:
C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:55 | Exiting...
Run Summary:
-------------
Traceback (most recent call last):
  File "mine.py", line 271, in
    logger.printSummary(settings)
  File "C:\x6500-miner-master\ConsoleLogger.py", line 210, in printSummary
    self.say('Device: %d' % self.devicenum, True, True)
AttributeError: 'ConsoleLogger' object has no attribute 'devicenum'

C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:56 | Exiting...
Run Summary:
-------------
Traceback (most recent call last):
  File "mine.py", line 271, in
    logger.printSummary(settings)
  File "C:\x6500-miner-master\ConsoleLogger.py", line 210, in printSummary
    self.say('Device: %d' % self.devicenum, True, True)
AttributeError: 'ConsoleLogger' object has no attribute 'devicenum'

C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:58 | Device 0 opened (AH01A6C3)
2014-02-02 00:36:58 | Connected to 2 FPGAs
2014-02-02 00:36:58 | FPGA 0 is running at 125MHz
2014-02-02 00:36:58 | FPGA 1 is running at 125MHz
2014-02-02 00:36:58 | Connected to server
2014-02-02 00:36:59 | Long-poll: connected to 192.168.1.12:8337
0 kH/s | 0/0/0 0.00%/0.00%

Otherwise, unplug the miner usb connection and plug it back in, spam the launch string.

ahh the magic touch lol  Grin
sr. member
Activity: 409
Merit: 250
Ok, getting it programmed just fine:
But when I run  mine.py I get:

Code:
Run Summary:
-------------
Traceback (most recent call last):
  File "mine.py", line 271, in
    logger.printSummary(settings)
  File "C:\Users\block_000\Desktop\x6500-miner-master\x6500-miner-master\Console
Logger.py", line 210, in printSummary
    self.say('Device: %d' % self.devicenum, True, True)
AttributeError: 'ConsoleLogger' object has no attribute 'devicenum'

Hmm.

Just spam the launch string a few more times..
Code:
C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:55 | Exiting...
Run Summary:
-------------
Traceback (most recent call last):
  File "mine.py", line 271, in
    logger.printSummary(settings)
  File "C:\x6500-miner-master\ConsoleLogger.py", line 210, in printSummary
    self.say('Device: %d' % self.devicenum, True, True)
AttributeError: 'ConsoleLogger' object has no attribute 'devicenum'

C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:56 | Exiting...
Run Summary:
-------------
Traceback (most recent call last):
  File "mine.py", line 271, in
    logger.printSummary(settings)
  File "C:\x6500-miner-master\ConsoleLogger.py", line 210, in printSummary
    self.say('Device: %d' % self.devicenum, True, True)
AttributeError: 'ConsoleLogger' object has no attribute 'devicenum'

C:\x6500-miner-master>"c:/python26/python.exe" mine.py -c 2 -i 5 -u 192.168.1.12:8337 -w user:x
2014-02-02 00:36:58 | Device 0 opened (AH01A6C3)
2014-02-02 00:36:58 | Connected to 2 FPGAs
2014-02-02 00:36:58 | FPGA 0 is running at 125MHz
2014-02-02 00:36:58 | FPGA 1 is running at 125MHz
2014-02-02 00:36:58 | Connected to server
2014-02-02 00:36:59 | Long-poll: connected to 192.168.1.12:8337
0 kH/s | 0/0/0 0.00%/0.00%

Otherwise, unplug the miner usb connection and plug it back in, spam the launch string.
Jump to: