Author

Topic: [ANN][YAC] YACoin ongoing development - page 105. (Read 379983 times)

sr. member
Activity: 310
Merit: 250
In Crypto I trust.
December 08, 2013, 10:45:35 PM
I downloaded the new client and ran it using the switch but yacoin still hasn't opened after 15 minutes. Huh
sr. member
Activity: 406
Merit: 250
One does not simply mine Bitcoins
December 08, 2013, 09:22:13 PM
it is a problem with pushpool itself as it somehow keeps connections established instead of closing it (and yes - I am using block notify scripts Smiley). After few days it is so overloaded (EDIT: sometimes there is 200000 established connections to yac port!) that there are more and more rejects - the only solution is to restart pushpool daemon.

There is no such problems with stratum server so thats why this is important (no only for me but also for miners)
IMHO that should be fairly easy to fix.
I can take a look after wrapping up the YAC client. BTW, already achieved massive speedup on startup - need more donations! Wink
YAC price will surely spike after we release it, haha! Grin

To the Moon! ┗(°0°)┛

Yac price is recovering nicely...usd value has increased afaik.

Now I need to get a proper wallet, mine isn't updating the blockchain after a certain point Huh

We have a working prototype of a new wallet ready. Not sure if it's bug-free, though. However, works for me so far.
You can get the latest windows binaries here: https://www.dropbox.com/s/pwa2lymjl6igh6q/yacoin-cc-latest.zip

NOTE: you NEED to move/rename (backup) everything except wallet.dat and peers.dat in the yacoin data directory and then run yacoin with -loadblock=C:\blk0001.dat (assuming you put the blk0001.dat file in C:\). This is needed because we had to change the way blocks are indexed and now the wallet starts in LESS THAN 30 SECONDS (down from more than 30 minutes). However, this initial load will take somewhere around 2 hours (still faster than downloading it through the p2p network).

If you don't mind redownloading all the blocks, you can just remove the unneeded files and start yacoin normally (less hassle, but takes more time).
hero member
Activity: 812
Merit: 1000
December 08, 2013, 09:11:02 PM
it is a problem with pushpool itself as it somehow keeps connections established instead of closing it (and yes - I am using block notify scripts Smiley). After few days it is so overloaded (EDIT: sometimes there is 200000 established connections to yac port!) that there are more and more rejects - the only solution is to restart pushpool daemon.

There is no such problems with stratum server so thats why this is important (no only for me but also for miners)
IMHO that should be fairly easy to fix.
I can take a look after wrapping up the YAC client. BTW, already achieved massive speedup on startup - need more donations! Wink
YAC price will surely spike after we release it, haha! Grin

To the Moon! ┗(°0°)┛

Yac price is recovering nicely...usd value has increased afaik.

Now I need to get a proper wallet, mine isn't updating the blockchain after a certain point Huh
sr. member
Activity: 406
Merit: 250
One does not simply mine Bitcoins
December 08, 2013, 05:51:10 PM
it is a problem with pushpool itself as it somehow keeps connections established instead of closing it (and yes - I am using block notify scripts Smiley). After few days it is so overloaded (EDIT: sometimes there is 200000 established connections to yac port!) that there are more and more rejects - the only solution is to restart pushpool daemon.

There is no such problems with stratum server so thats why this is important (no only for me but also for miners)
IMHO that should be fairly easy to fix.
I can take a look after wrapping up the YAC client. BTW, already achieved massive speedup on startup - need more donations! Wink
YAC price will surely spike after we release it, haha! Grin

To the Moon! ┗(°0°)┛
legendary
Activity: 1197
Merit: 1000
December 08, 2013, 05:05:02 PM
Hi,

As promised I opened 2 new LP ports on http://yac.coinmine.pl pool to handle the load:

8088 - not recommended - heavily used
8080 - not recommended - heavily used
8881 - recommended - new
8882 - recommended - new

Stratum Port: 9088 - recommended

feeleep



umm... are you sure you're not just running into the default "socket max conn" limit? the default in linux is set to 128, which limits the number of incoming connections per one socket (one pool port) to 128, which is absurdly low. you can check this with
Code:
cat /proc/sys/net/core/somaxconn

or, if this is not the case, you must be getting some errors in the kernel log (dmesg) - if so, what are they? it's also possible you've reached max number of file handles per process (can be increased, too).

it is a problem with pushpool itself as it somehow keeps connections established instead of closing it (and yes - I am using block notify scripts Smiley). After few days it is so overloaded (EDIT: sometimes there is 200000 established connections to yac port!) that there are more and more rejects - the only solution is to restart pushpool daemon.

There is no such problems with stratum server so thats why this is important (no only for me but also for miners)
sr. member
Activity: 406
Merit: 250
One does not simply mine Bitcoins
December 08, 2013, 04:50:05 PM
Hi,

As promised I opened 2 new LP ports on http://yac.coinmine.pl pool to handle the load:

8088 - not recommended - heavily used
8080 - not recommended - heavily used
8881 - recommended - new
8882 - recommended - new

Stratum Port: 9088 - recommended

feeleep



umm... are you sure you're not just running into the default "socket max conn" limit? the default in linux is set to 128, which limits the number of incoming connections per one socket (one pool port) to 128, which is absurdly low. you can check this with
Code:
cat /proc/sys/net/core/somaxconn

or, if this is not the case, you must be getting some errors in the kernel log (dmesg) - if so, what are they? it's also possible you've reached max number of file handles per process (can be increased, too).
legendary
Activity: 1197
Merit: 1000
December 08, 2013, 04:39:34 PM
Hi,

As promised I opened 2 new LP ports on http://yac.coinmine.pl pool to handle the load:

8088 - not recommended - heavily used
8080 - not recommended - heavily used
8881 - recommended - new
8882 - recommended - new

Stratum Port: 9088 - recommended

feeleep

legendary
Activity: 1197
Merit: 1000
December 08, 2013, 08:38:05 AM

I think the lack of a mining client with stratum support is a pretty big freakin deal, right? 

Well, not that much. Stratum protocol aims to reduce the load on pool servers with a huge number of miners by letting the miners:
  • create unique valid work for themselves when they run out of shares and there was not a new block on the network
  • listen for changes instead of asking for them every couple of seconds, thus reducing the bandwidth used by miners/pools by disabling longpoll and making the pool push notifications to miner

The first point doesn't need correcting yet as it's not slowing down miners (because YAC mining is so slow anyway), and the second is not a big deal unless the pools run out of bandwidth due to a HUGE amount of miners.

I'd say a bigger deal would be integrating CPU mining into more sophisticated clients like cgminer (AFAIK there's yacminer with only GPU support). This way you would have the option to failover from one pool to another automagically, which is needed badly right now with pools crashing every couple of minutes (no idea why they're crashing).

its not pool which is crashing - pushpool is not able to handle so many connections... I will open few more ports today evening to support higher interest of yac mining
sr. member
Activity: 406
Merit: 250
One does not simply mine Bitcoins
December 08, 2013, 06:54:03 AM

I think the lack of a mining client with stratum support is a pretty big freakin deal, right? 

Well, not that much. Stratum protocol aims to reduce the load on pool servers with a huge number of miners by letting the miners:
  • create unique valid work for themselves when they run out of shares and there was not a new block on the network
  • listen for changes instead of asking for them every couple of seconds, thus reducing the bandwidth used by miners/pools by disabling longpoll and making the pool push notifications to miner

The first point doesn't need correcting yet as it's not slowing down miners (because YAC mining is so slow anyway), and the second is not a big deal unless the pools run out of bandwidth due to a HUGE amount of miners.

I'd say a bigger deal would be integrating CPU mining into more sophisticated clients like cgminer (AFAIK there's yacminer with only GPU support). This way you would have the option to failover from one pool to another automagically, which is needed badly right now with pools crashing every couple of minutes (no idea why they're crashing).
hero member
Activity: 809
Merit: 501
December 08, 2013, 02:42:41 AM

I think the lack of a mining client with stratum support is a pretty big freakin deal, right? 
sr. member
Activity: 406
Merit: 250
One does not simply mine Bitcoins
hero member
Activity: 693
Merit: 500
December 07, 2013, 03:50:13 AM
It's something in the code - I compiled mikaelh's code with the same options as the code with stratum, and it's about 15% faster.  Too tired to dig through the code tonight to see where the differences are.
hero member
Activity: 693
Merit: 500
December 06, 2013, 12:02:33 PM
Thanks for the coins folks.

I've taken feeleep's work, merged it with what I did in my earlier versions, and updated the scrypt-jane library from floodyberry's latest push and put it all together into a bunch of executables.

I made one zip that has x64 binaries for each of the following architectures: avx2, avx, corei7, core2, sse3 and athlon.  I'm going to work on getting the github source updated for those who will want to compile their own and see the changes - though they are significant as this is based on cpuminer 2.3.1 vs the old one being 2.2.3.

I pushed it to Mega - if someone else wants to host it or suggest someplace else to host it, feel free.
https://mega.co.nz/#!nk4FkZhQ!G0P9605rdfhaE6e1CkXp2GQB1I9BSATwYBKZtWncDKQ


not sure if I am right but this version is 2x slower than previous one... could anyone doublecheck?

anyone else willing to try?  I'm not feeling so confident about this... I have yet to find a share using these binaries even though it's reporting the correct hashrate.  The binary I compiled previous was finding them like mad.  Try at your own risk today, but just in case, I'm posting the binary that was for sure finding shares, compiled from the source feeleep posted as well (AVX version only):
https://mega.co.nz/#!O0wDUaKK!SOp9J2oIh9a8TvImtQQdS0tQ7Hgi8DRku7LJDpoComw

so "my version" is also slower than this one: http://yacointalk.com/index.php?topic=264.0

feeleep
I've never actually used that version, but the avx version is noticeably faster for me than the ones I've compiled myself (stratum or otherwise).  He did post the code, maybe it's in the compiler optimizations, or maybe it's in the code.  I'll see if I can spot the differences...
legendary
Activity: 1197
Merit: 1000
December 06, 2013, 02:07:38 AM
Thanks for the coins folks.

I've taken feeleep's work, merged it with what I did in my earlier versions, and updated the scrypt-jane library from floodyberry's latest push and put it all together into a bunch of executables.

I made one zip that has x64 binaries for each of the following architectures: avx2, avx, corei7, core2, sse3 and athlon.  I'm going to work on getting the github source updated for those who will want to compile their own and see the changes - though they are significant as this is based on cpuminer 2.3.1 vs the old one being 2.2.3.

I pushed it to Mega - if someone else wants to host it or suggest someplace else to host it, feel free.
https://mega.co.nz/#!nk4FkZhQ!G0P9605rdfhaE6e1CkXp2GQB1I9BSATwYBKZtWncDKQ


not sure if I am right but this version is 2x slower than previous one... could anyone doublecheck?

anyone else willing to try?  I'm not feeling so confident about this... I have yet to find a share using these binaries even though it's reporting the correct hashrate.  The binary I compiled previous was finding them like mad.  Try at your own risk today, but just in case, I'm posting the binary that was for sure finding shares, compiled from the source feeleep posted as well (AVX version only):
https://mega.co.nz/#!O0wDUaKK!SOp9J2oIh9a8TvImtQQdS0tQ7Hgi8DRku7LJDpoComw

so "my version" is also slower than this one: http://yacointalk.com/index.php?topic=264.0

feeleep
hero member
Activity: 809
Merit: 501
December 06, 2013, 01:43:24 AM
\cpuminer-stratum-feeleep_x64_avx\minerd.exe -q -a scrypt-jane -o stratum+tcp://yac.coinmine.pl:9088 -O Workername.1:password

Program just opens and closes immediately

What am I doing wrong?

feeleep just left the -a option as "scrypt", so the following should work with that executable.


\cpuminer-stratum-feeleep_x64_avx\minerd.exe -q -a scrypt -o stratum+tcp://yac.coinmine.pl:9088 -O Workername.1:password

Thank you for the reply.

But still not working Sad
hero member
Activity: 693
Merit: 500
December 06, 2013, 01:33:35 AM
\cpuminer-stratum-feeleep_x64_avx\minerd.exe -q -a scrypt-jane -o stratum+tcp://yac.coinmine.pl:9088 -O Workername.1:password

Program just opens and closes immediately

What am I doing wrong?

feeleep just left the -a option as "scrypt", so the following should work with that executable.


\cpuminer-stratum-feeleep_x64_avx\minerd.exe -q -a scrypt -o stratum+tcp://yac.coinmine.pl:9088 -O Workername.1:password
hero member
Activity: 809
Merit: 501
December 06, 2013, 01:13:38 AM
\cpuminer-stratum-feeleep_x64_avx\minerd.exe -q -a scrypt-jane -o stratum+tcp://yac.coinmine.pl:9088 -O Workername.1:password

Program just opens and closes immediately

What am I doing wrong?
hero member
Activity: 693
Merit: 500
December 05, 2013, 06:26:32 PM
Thanks for the coins folks.

I've taken feeleep's work, merged it with what I did in my earlier versions, and updated the scrypt-jane library from floodyberry's latest push and put it all together into a bunch of executables.

I made one zip that has x64 binaries for each of the following architectures: avx2, avx, corei7, core2, sse3 and athlon.  I'm going to work on getting the github source updated for those who will want to compile their own and see the changes - though they are significant as this is based on cpuminer 2.3.1 vs the old one being 2.2.3.

I pushed it to Mega - if someone else wants to host it or suggest someplace else to host it, feel free.
https://mega.co.nz/#!nk4FkZhQ!G0P9605rdfhaE6e1CkXp2GQB1I9BSATwYBKZtWncDKQ


not sure if I am right but this version is 2x slower than previous one... could anyone doublecheck?

anyone else willing to try?  I'm not feeling so confident about this... I have yet to find a share using these binaries even though it's reporting the correct hashrate.  The binary I compiled previous was finding them like mad.  Try at your own risk today, but just in case, I'm posting the binary that was for sure finding shares, compiled from the source feeleep posted as well (AVX version only):
https://mega.co.nz/#!O0wDUaKK!SOp9J2oIh9a8TvImtQQdS0tQ7Hgi8DRku7LJDpoComw
hero member
Activity: 693
Merit: 500
December 05, 2013, 06:24:40 PM
Code forked from pooler and updated to support yacoin added to my repository for those who wish to peruse the changes.
https://github.com/thirtybird/cpuminer
legendary
Activity: 1197
Merit: 1000
December 05, 2013, 06:24:03 PM
Thanks for the coins folks.

I've taken feeleep's work, merged it with what I did in my earlier versions, and updated the scrypt-jane library from floodyberry's latest push and put it all together into a bunch of executables.

I made one zip that has x64 binaries for each of the following architectures: avx2, avx, corei7, core2, sse3 and athlon.  I'm going to work on getting the github source updated for those who will want to compile their own and see the changes - though they are significant as this is based on cpuminer 2.3.1 vs the old one being 2.2.3.

I pushed it to Mega - if someone else wants to host it or suggest someplace else to host it, feel free.
https://mega.co.nz/#!nk4FkZhQ!G0P9605rdfhaE6e1CkXp2GQB1I9BSATwYBKZtWncDKQ


not sure if I am right but this version is 2x slower than previous one... could anyone doublecheck?
Jump to: