[BBR] Boolberry: Privacy and Security - Guaranteed Since 2014

[dga]

I spent some time auditing simpleminer for missing failure tests;  only found one.  This patch updates it (it should be something nobody would encounter in any normal situation, but I wanted to be sure to cover it in case of weird pool failures / timing so the client wouldn't fail).

https://github.com/dave-andersen/boolberry/commit/903f81cf13f57f023fdc61a55c5f3ab7d49ef1aa

will test overnight and push tomorrow if nobody notices problems.

The plan after that is to integrate long-polling, which should finally bring simpleminer into nearly-full speed parity with solo.

[antonio8]

I might be one of the few but Christian, do what you do. You deserve it.

Isn't this like what happened with Prime Coin? People knew a CPU miner existed but no one would claim it and they mined the heck out of it. When did a miner finally get released? When it wasn't profitable.

[oranges411]

The plan after that is to integrate long-polling, which should finally bring simpleminer into nearly-full speed parity with solo.

-clap- -clap-

[Alpharisc]

Ignore the haters Christian. Without you, Nvidia users would be still in the dark when it came to mining. This guy has bought more equitable mining software to the table than anyone else. The problem here was using an algo that could be adapted to GPU's without putting in the work to make such any such mining software before release. With a big block reward, and the current price, the boolberry devs should be getting off the asses and making things happen.

My question would be, what are you doing with all the coins?? Not that it's any of my business, but the trading volume seems to suggest you are holding them.

[cbuchner1]

My question would be, what are you doing with all the coins?? Not that it's any of my business, but the trading volume seems to suggest you are holding them.

A combination of trickle sale and putting them for sale above the current market price.
I don't think we've ever held more than 10000 BBR at once.

Christian

[onemorebtc]

My question would be, what are you doing with all the coins?? Not that it's any of my business, but the trading volume seems to suggest you are holding them.

A combination of trickle sale and putting them for sale above the current market price.
I don't think we've ever held more than 10000 BBR at once.

Christian

well, as you seem honest, this is nice to know

[dewdeded]

I started a general thread on CryptoNote, please join discussion under:
https://bitcointalk.org/index.php?topic=658884.

[sunanchi]

who can teach me how to mine the bbr with Online mining
i have many machine but it just solo by themsleves.
can i use them to mine same block
thanks.

[otila]

patch to disable Nagle's algorithm, reduces simpleminer request latency by ~0.2s-0.5s per request, or up to ~1s per share because it waits for response to block submit instead of sending the getjob command in the same time as doing block submit  

Code:

diff --git a/contrib/epee/include/net/net_helper.h b/contrib/epee/include/net/net_helper.h
index 4f7ebfa..3699858 100644
--- a/contrib/epee/include/net/net_helper.h
+++ b/contrib/epee/include/net/net_helper.h
@@ -159,8 +159,8 @@ namespace net_utils
 					boost::asio::ip::tcp::endpoint local_endpoint(boost::asio::ip::address::from_string(addr.c_str()), 0);
 					m_socket.bind(local_endpoint);
 				}
+				m_socket.set_option(boost::asio::ip::tcp::no_delay(true));
 
-				
 				m_deadline.expires_from_now(boost::posix_time::milliseconds(m_connect_timeout));
 
 

I would prefer to have TCP_CORK socket option, does boost have it?  Now it just sends more packets than is necessary.  Or fix epee..?
Linux 3.14 has tcp_autocorking enabled by default since 3.14..

before:

Code:

2014-06-20T06:34:19.184956Z IP (tos 0x2,ECT(0), ttl 255, id 12996, offset 0, flags [none], proto TCP (6), length 126)
    192.168.1.102.52306 > 216.119.175.87.5555: Flags [P.], cksum 0x4a4e (incorrect -> 0x86f7), seq 3237694011:3237694085, ack 2743690248, win 16384, options [nop,nop,TS val 986498185 ecr 366660], length 74
        0x0000:  4502 007e 32c4 0000 ff06 3ed6 c0a8 0166  E..~2.....>....f
        0x0010:  d877 af57 cc52 15b3 c0fb 4a3b a389 6408  .w.W.R....J;..d.
        0x0020:  8018 4000 4a4e 0000 0101 080a 3acc c489  ..@.JN......:...
        0x0030:  0005 9844 4745 5420 2f6a 736f 6e5f 7270  ...DGET./json_rp
        0x0040:  6320 4854 5450 2f31 2e31 0d0a 486f 7374  c.HTTP/1.1..Host
        0x0050:  3a20 6262 722e 6578 7472 656d 6570 6f6f  :.bbr.extremepoo
        0x0060:  6c2e 6f72 670d 0a43 6f6e 7465 6e74 2d4c  l.org..Content-L
        0x0070:  656e 6774 683a 2032 3531 0d0a 0d0a       ength:.251....
2014-06-20T06:34:19.501248Z IP (tos 0x0, ttl 44, id 28904, offset 0, flags [none], proto TCP (6), length 52)
    216.119.175.87.5555 > 192.168.1.102.52306: Flags [.], cksum 0x9588 (correct), ack 3237694085, win 1452, options [nop,nop,TS val 367927 ecr 986498185], length 0
        0x0000:  4500 0034 70e8 0000 2c06 d3fe d877 af57  E..4p...,....w.W
        0x0010:  c0a8 0166 15b3 cc52 a389 6408 c0fb 4a85  ...f...R..d...J.
        0x0020:  8010 05ac 9588 0000 0101 080a 0005 9d37  ...............7
        0x0030:  3acc c489                                :...
2014-06-20T06:34:19.501309Z IP (tos 0x2,ECT(0), ttl 255, id 12997, offset 0, flags [none], proto TCP (6), length 303)
    192.168.1.102.52306 > 216.119.175.87.5555: Flags [P.], cksum 0x4aff (incorrect -> 0x3468), seq 3237694085:3237694336, ack 2743690248, win 16384, options [nop,nop,TS val 986498502 ecr 367927], length 251
        0x0000:  4502 012f 32c5 0000 ff06 3e24 c0a8 0166  E../2.....>$...f
        0x0010:  d877 af57 cc52 15b3 c0fb 4a85 a389 6408  .w.W.R....J...d.
        0x0020:  8018 4000 4aff 0000 0101 080a 3acc c5c6  ..@.J.......:...
        0x0030:  0005 9d37 7b0d 0a20 2022 6964 223a 2022  ...7{...."id":."
        0x0040:  3022 2c0d 0a20 2022 6a73 6f6e 7270 6322  0",...."jsonrpc"
        0x0050:  3a20 2232 2e30 222c 0d0a 2020 226d 6574  :."2.0",...."met
        0x0060:  686f 6422 3a20 2273 7562 6d69 7422 2c0d  hod":."submit",.
        ... snip

after:

Code:

2014-06-20T15:45:39.984776Z IP (tos 0x2,ECT(0), ttl 255, id 64704, offset 0, flags [none], proto TCP (6), length 126)
    192.168.1.102.62860 > 216.119.175.87.7777: Flags [P.], cksum 0x4a4e (incorrect -> 0x5aad), seq 2923004606:2923004680, ack 2280487023, win 16384, options [nop,nop,TS val 1019578985 ecr 8629187], length 74
	0x0000:  4502 007e fcc0 0000 ff06 74d9 c0a8 0166  E..~......t....f
	0x0010:  d877 af57 f58c 1e61 ae39 82be 87ed 786f  .w.W...a.9....xo
	0x0020:  8018 4000 4a4e 0000 0101 080a 3cc5 8a69  ..@.JN......<..i
	0x0030:  0083 abc3 4745 5420 2f6a 736f 6e5f 7270  ....GET./json_rp
	0x0040:  6320 4854 5450 2f31 2e31 0d0a 486f 7374  c.HTTP/1.1..Host
	0x0050:  3a20 6262 722e 6578 7472 656d 6570 6f6f  :.bbr.extremepoo
	0x0060:  6c2e 6f72 670d 0a43 6f6e 7465 6e74 2d4c  l.org..Content-L
	0x0070:  656e 6774 683a 2032 3531 0d0a 0d0a       ength:.251....
2014-06-20T15:45:39.985071Z IP (tos 0x2,ECT(0), ttl 255, id 64705, offset 0, flags [none], proto TCP (6), length 303)
    192.168.1.102.62860 > 216.119.175.87.7777: Flags [P.], cksum 0x4aff (incorrect -> 0x54d0), seq 2923004680:2923004931, ack 2280487023, win 16384, options [nop,nop,TS val 1019578986 ecr 8629187], length 251
	0x0000:  4502 012f fcc1 0000 ff06 7427 c0a8 0166  E../......t'...f
	0x0010:  d877 af57 f58c 1e61 ae39 8308 87ed 786f  .w.W...a.9....xo
	0x0020:  8018 4000 4aff 0000 0101 080a 3cc5 8a6a  ..@.J.......<..j
	0x0030:  0083 abc3 7b0d 0a20 2022 6964 223a 2022  ....{...."id":."
	0x0040:  3022 2c0d 0a20 2022 6a73 6f6e 7270 6322  0",...."jsonrpc"
	0x0050:  3a20 2232 2e30 222c 0d0a 2020 226d 6574  :."2.0",...."met
	0x0060:  686f 6422 3a20 2273 7562 6d69 7422 2c0d  hod":."submit",.
        ...snip

[dga]

patch to disable Nagle's algorithm, reduces simpleminer request latency by ~0.2s-0.5s per request, or up to ~1s per share because it waits for response to block submit instead of sending the getjob command in the same time as doing block submit  

Argh - thanks, I had thought to myself that I should check that and completely forgot.

I'll start testing this now.  Appreciate the patch!

(update:  I'm going to apply this only within simpleminer for now to keep the risk reduced of changing it globally.)

If CORK isn't an option, I'll go through it and add some manual buffering of the writes and/or change it to writev if needed.

[33zer0w0lf]

patch to disable Nagle's algorithm, reduces simpleminer request latency by ~0.2s-0.5s per request, or up to ~1s per share because it waits for response to block submit instead of sending the getjob command in the same time as doing block submit  

Argh - thanks, I had thought to myself that I should check that and completely forgot.

I'll start testing this now.  Appreciate the patch!

(update:  I'm going to apply this only within simpleminer for now to keep the risk reduced of changing it globally.)

If CORK isn't an option, I'll go through it and add some manual buffering of the writes and/or change it to writev if needed.

awesome let me know when this fix is out and I will start testing! 

Pool is starting to pick up some momentum, should start seeing regular blocks!

http://bbr.extremepool.org

[otila]

31233/7810=4

Code:

2014-Jun-20 20:08:32.835339 Job didn't change
2014-Jun-20 20:08:35.149099 Share found: nonce=1236027 for job=766127159120515, diff: 1000000
, PoW:<b0483585675a1eccf65eed2bd221f6438280b2e657c064e25b40829558060000>, height:24250, submitting...
2014-Jun-20 20:08:35.444838 ERROR /c/boolberry/contrib/epee/include/storages/http_abstract_invoke.h:112[bool epee::net_utils::invoke_http_json_rpc(const string&, const string&, t_request&, t_response&, t_transport&, unsigned int, const string&, const string&) [with t_request = mining::COMMAND_RPC_SUBMITSHARE::request; t_response = mining::COMMAND_RPC_SUBMITSHARE::response; t_transport = epee::net_utils::http::http_simple_client; std::string = std::basic_string<char>]]RPC call of "submit" returned error: -1, message: Low difficulty share
2014-Jun-20 20:08:35.444892 Failed to submit share!  Updating job.
2014-Jun-20 20:08:36.445021 Connecting bbr.extremepool.org:7777....
2014-Jun-20 20:08:36.695119 Connected bbr.extremepool.org:7777 OK
2014-Jun-20 20:08:36.955066 Getting scratchpad...
2014-Jun-20 20:08:53.493900 Scratchpad received ok, size: 31233 KiB, height=24249
2014-Jun-20 20:08:53.494763 Getting next job...
2014-Jun-20 20:08:53.754985 Job didn't change
2014-Jun-20 20:09:01.027921 Share found: nonce=1653696 for job=292234379635192, diff: 1000000
, PoW:<9f657c02ae13a1cd24b5a0e0bbf850842c2b4d1befeeb29e50a07a995a0f0000>, height:24250, submitting...
2014-Jun-20 20:09:01.315045 ERROR /c/boolberry/contrib/epee/include/storages/http_abstract_invoke.h:112[bool epee::net_utils::invoke_http_json_rpc(const string&, const string&, t_request&, t_response&, t_transport&, unsigned int, const string&, const string&) [with t_request = mining::COMMAND_RPC_SUBMITSHARE::request; t_response = mining::COMMAND_RPC_SUBMITSHARE::response; t_transport = epee::net_utils::http::http_simple_client; std::string = std::basic_string<char>]]RPC call of "submit" returned error: -1, message: Low difficulty share
2014-Jun-20 20:09:01.315138 Failed to submit share!  Updating job.
2014-Jun-20 20:09:01.315175 Too many submission failures.  Something is very wrong.
2014-Jun-20 20:09:01.315204 Excessive failures.  Sleeping 10-ish seconds and restarting...
2014-Jun-20 20:09:11.822174 Mining with 4 threads 
2014-Jun-20 20:09:11.822227 Connecting bbr.extremepool.org:7777....
2014-Jun-20 20:09:12.075822 Connected bbr.extremepool.org:7777 OK
2014-Jun-20 20:09:12.335865 Getting scratchpad...
2014-Jun-20 20:09:30.463601 Scratchpad received ok, size: 7810 KiB, height=24249
2014-Jun-20 20:09:30.464554 Getting next job...
2014-Jun-20 20:09:49.054149 Share found: nonce=4052003 for job=799678992363624, diff: 1000000
, PoW:<ea12ab15b2e4df5f939a864da6b92d958b61d216b46401a6ecf1285b510d0000>, height:24250, submitting...
2014-Jun-20 20:09:49.394855 Share submitted successfully!

[dga]

- I've pushed the NODELAY change into the open pull request with the reliability fix:

   https://github.com/cryptozoidberg/boolberry/pull/28

It's simple enough that I think it's worth doing now.  Whether you see any benefits from this will depend a lot on your system and the timing of writes.

Otila, can you confirm that this is as effective for you as your patch was?

- Thanks for the /4 comment about the scratchpad corruption.  That bug is still on my todo list to track down, and that's helpful.  I'm relieved to see that the bug masking is still working, but we shouldn't have to fall back to a full restart like that.

[otila]

- I've pushed the NODELAY change into the open pull request with the reliability fix:

I applied both of those and it works OK (tried also with reconnect (new socket)).

[33zer0w0lf]

hey dga where is the latest repo at with the latest updates?

[crypto_zoidberg]

Hi folk!
Last 3 days i spent experementig with CPU miner optimizations.
I took original optimized code from http://keccak.noekeon.org, spent a lot of time adding macros for wild keccak, spent a lot of time finding a bugs in macros (finding bugs in macros it's real nighmare ), and then i've got some interesting results that i want to share.

On this graph vertical axis shows time spent for calculating 100000 hashes in two different algos and it horizontal axis shows how scratchpad is growing (for about one year).



Green lines it's a windows versions, as you could see optimzed version showed about 25% boost compared with original algo.
But, as the same code makes reverse effect in linux(blue lines) - optimized version works slower.

I guess it because ms compiler sucks gcc version is faster and doesn't need some manual optimization. 
But, as i talked with dga, he also had some linux specific optimizations, that makes 10-15% boost, he is going to contribute it also.

This graph is built on my core i5, i think it would be interesting for all, if we get differen graphs for different CPUs.
HowTo get graph:
1. git pull
2. make -j
3. ./build/release/tests/performance_tests

after few performance tests you'll see something this:

Code:

sz	                 original	 opt       
992                 	2057      	2270      
100992              	2058      	2268      
200992              	2065      	2273      
300992              	2073      	2267      
400992              	2075      	2268      
500992              	2091      	2281      
....

Now wait .... about 30-minutes, until numbers will be close to 100000000 (100MB scratchpad).
After that, just copy this text into any spreadsheets (i used google online spreadsheets) and make a graph from this.


PS: This graph should also explain why some people complain that new build have "bad" hash rate - here you can see how hash calculation is slowing down.

What next:
1. Tomorrow i'll make windows build that use optimized version and works a little faster
2. Also tomorrow we going to release GUI (if apple developer will finally create a bundle for macos precompiled binaries)
2. I'll start to work on stratum pool.

[33zer0w0lf]

I have spoken to LucasJones about stratum, the pool he code he released, same that is running @ http://bbr.extremepool.org, does support stratum, so that part is already done.  What we are missing is a cpuminer like cpuminer-multi from either LucasJones or Wolf that can mine with the wild keccak algo.

[btc-mike]

I have spoken to LucasJones about stratum, the pool he code he released, same that is running @ http://ultrbooCrybbr.extremepool.org, does support stratum, so that part is already done.  What we are missing is a cpumer like cpuminer-multi from either LucasJones or Wolf that can mine with the wild keccak algo.

Is it full stratum? Don't the other CN coins use a broken version of stratum?

[33zer0w0lf]

I have spoken to LucasJones about stratum, the pool he code he released, same that is running @ http://ultrbooCrybbr.extremepool.org, does support stratum, so that part is already done.  What we are missing is a cpumer like cpuminer-multi from either LucasJones or Wolf that can mine with the wild keccak algo.

Is it full stratum? Don't the other CN coins use a broken version of stratum?

that's a good question as far as I am aware it is full stratum, and lucas confirmed the pool code does support startum that we just need a miner that was capable of mining such as minerd

[thinker-jone]

good coin but i have missed....   the price is too high for buying.