[ANN][RIC] Riecoin: constellations POW *CPU* HARD FORK successful, world record

[CremeFraiche]

running b11-version on 3 windows machines for 24 hours now without problems (ypool)

[dga]

Ahh, thanks for the very helpful debug output.

I've updated the miner to b12 and committed into the repository.

b12 fixes a known bug with not regenerating the nonce time frequently enough for ypool.  The miner now resets its time nonce every minute instead of every block.

This slows down the miner a bit.  Sorry - but it should compensate by not having invalid shares as often.  You might find that reducing -s slightly helps.

But there may still be lingering buts causing this "not a valid prime" error.  I'm going to keep digging, and will work on getting some of that speed back in b13.

  -Dave

Are people seeing this on Linux also?

yes.

Code:

2014-03-23 08:58:23.532409175 [07:39:52] Share found! (Blockheight: 24610)
2014-03-23 08:58:25.432824705 [07:39:54] 2ch/s: 30.3795 3ch/s: 1.9480 4ch/s: 0.0681 Shares total: 532 / 511
2014-03-23 08:58:33.432838812 [07:40:02] 2ch/s: 30.3808 3ch/s: 1.9483 4ch/s: 0.0681 Shares total: 532 / 511
2014-03-23 08:58:34.672120815 [07:40:03] Share found! (Blockheight: 24610)
2014-03-23 08:58:41.434825115 [07:40:10] 2ch/s: 30.3809 3ch/s: 1.9485 4ch/s: 0.0682 Shares total: 533 / 512
2014-03-23 08:58:49.435824174 [07:40:18] 2ch/s: 30.3814 3ch/s: 1.9489 4ch/s: 0.0682 Shares total: 533 / 512
2014-03-23 08:58:56.817584732 [07:40:25] Share found! (Blockheight: 24610)
2014-03-23 08:58:57.435830082 [07:40:26] 2ch/s: 30.3820 3ch/s: 1.9490 4ch/s: 0.0684 Shares total: 534 / 513
2014-03-23 08:58:57.751832129 Invalid share
2014-03-23 08:58:57.751847632 Reason: CheckProofOfWork() : not valid pow
2014-03-23 08:59:05.435828594 [07:40:34] 2ch/s: 30.3828 3ch/s: 1.9487 4ch/s: 0.0683 Shares total: 534 / 512
... no new shares found till I restarted the process
2014-03-23 09:01:21.381338929 [07:42:50] Share found! (Blockheight: 24612)
2014-03-23 09:01:21.450835838 [07:42:50] 2ch/s: 30.3662 3ch/s: 1.9490 4ch/s: 0.0681 Shares total: 535 / 513
2014-03-23 09:01:22.209827519 Invalid share
2014-03-23 09:01:22.209846968 Reason: CheckProofOfWork() : n not prime
2014-03-23 09:01:29.450829867 [07:42:58] 2ch/s: 30.3649 3ch/s: 1.9488 4ch/s: 0.0681 Shares total: 535 / 512
2014-03-23 09:01:31.656831570 Ping 656.0ms (Average 912.0)
2014-03-23 09:01:37.451826894 [07:43:06] 2ch/s: 30.3659 3ch/s: 1.9492 4ch/s: 0.0681 Shares total: 535 / 512
2014-03-23 09:01:45.453825500 [07:43:14] 2ch/s: 30.3656 3ch/s: 1.9491 4ch/s: 0.0681 Shares total: 535 / 512
2014-03-23 09:01:53.455857748 [07:43:22] 2ch/s: 30.3648 3ch/s: 1.9488 4ch/s: 0.0681 Shares total: 535 / 512
2014-03-23 09:02:01.073502585 [07:43:30] Share found! (Blockheight: 24612)
2014-03-23 09:02:01.289827815 Invalid share
2014-03-23 09:02:01.289841922 Reason: CheckProofOfWork() : n not prime
...
2014-03-23 09:24:07.587949574 Reason: CheckProofOfWork() : n not prime
2014-03-23 09:24:09.612825465 [08:05:38] 2ch/s: 30.2689 3ch/s: 1.9400 4ch/s: 0.0685 Shares total: 560 / 512

restarted, all works ok:

2014-03-23 09:25:18.179824577 [00:00:56] 2ch/s: 23.3326 3ch/s: 1.0971 4ch/s: 0.0000 Shares total: 0 / 0
2014-03-23 09:25:18.686221800 [00:00:56] Share found! (Blockheight: 24628)
2014-03-23 09:25:26.179830232 [00:01:04] 2ch/s: 24.1280 3ch/s: 1.2160 4ch/s: 0.0640 Shares total: 1 / 1

[goodluck0319]

I am wondering if u can fix the windows leak problem on ur miner. does b12 fix the "invalid share, not a valid prime and Reason: CheckProofOfWork() : not valid pow problem?

[antonio8]

How are you all using the avx2 version?

I downloaded it and it is just a file and not a program like the sse4 version.

If it needs compiled does anyone have a compiled version for windows?

[Aboroth]

How are you all using the avx2 version?

I downloaded it and it is just a file and not a program like the sse4 version.

If it needs compiled does anyone have a compiled version for windows?

I think the avx2 version is Linux only.

[surfer43]

How are you all using the avx2 version?

I downloaded it and it is just a file and not a program like the sse4 version.

If it needs compiled does anyone have a compiled version for windows?

It's a linux executable program.

The names of Windows programs always end in .exe
Linux programs usually do not have a file type ending.

[antonio8]

@ aboroth and surfer43

Thanks.

I need to learn how to compile.

[kobol21]

Hey guys, where can I see RIECOIN difficulty chart ?! Cant find it anywhere!

[istudy92]

Hey guys, where can I see RIECOIN difficulty chart ?! Cant find it anywhere!

http://da-data.blogspot.com/?view=classic

[dga]

The ypool chat room was full of people noticing that b12 slowed down too much, so I've pushed b13 as a compromise between safety and speed.  But it also includes an important bug fix (which should be masked by the other one, but fixing bugs is good).

 http://www.cs.cmu.edu/~dga/crypto/ric/
and in github:
 https://github.com/dave-andersen/fastrie/

Also, there's now a ChangeLog that covers all of this stuff:  https://github.com/dave-andersen/fastrie/blob/master/ChangeLog

It may be slightly slower than b11 when a block takes a long time, but it's not too far off.  It might even be a little faster for fast blocks, but overall, it's probably 5% slower than b11, but it won't get into the "invalid prime" problem.

There were two bugs in b11:
  - If it went too long before a new block was found, it could overflow the 256 bit riecoin nonce field -- I changed some constants and didn't re-do the math on them.  This is now computed based upon those constants, so the bug shouldn't reappear.

  - It could run for too long without updating its block time field, which would cause ypool to reject it even if it contained a valid share or block.  I've mostly kinda sorta fixed this, but the current version reflects a compromise between safety (updating the time field frequently) and efficiency (running as long as possible after computing a target).  This one's a "let's run it and see how it does" kind of thing.

For people who like to read the source of these things, the internals have gotten a little cleaner, which hopefully makes the logic a bit more clear.  In particular, I think the offset computation should be easier to read and understand now, for those seeking to understand what makes these miners tick.

  -Dave

[goodluck0319]

so what does b13 fix?

[dga]

so what does b13 fix?

The most important fix is that it addresses the lack of a changelog to answer questions like this.

https://github.com/dave-andersen/fastrie/blob/master/ChangeLog

(I suppose the changelog should include that b13 includes a changelog. :-)

[istudy92]

MY PC uses like 5.1GB memory with this miner, is this normal?

[one4many]

MY PC uses like 5.1GB memory with this miner, is this normal?

Seems to be normal on machine with several cores.
Here is the mem usage on one of my systems:

Code:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND                                                                                                                             
 3104 root      20   0 31.5g  28g 1052 S 4810.3 74.4  28803:26 xptminer-sse4-b                                                                                                                    

Is yours an 8 core machine?

    one4many

[istudy92]

Well I know on my Win 7 I have a 8 core 8320, it uses 5.1 gb.

On my 4p 64 core ubunto it would crash and had to limit the mem to 4gb per processor, it REALLY wants more and more.  I am not sure if giving it more would benifit the miner?  So in general I have allocated a total of 15.7 GB to the 4p opposed to the 21gb it sorta wants.

[northranger79510]

Quick Update: RiecoinFoundation.org will be ready by next Monday

[dogaatac]

the price is killing me

[relm9]

The ypool chat room was full of people noticing that b12 slowed down too much, so I've pushed b13 as a compromise between safety and speed.  But it also includes an important bug fix (which should be masked by the other one, but fixing bugs is good).

 http://www.cs.cmu.edu/~dga/crypto/ric/
and in github:
 https://github.com/dave-andersen/fastrie/

Also, there's now a ChangeLog that covers all of this stuff:  https://github.com/dave-andersen/fastrie/blob/master/ChangeLog

It may be slightly slower than b11 when a block takes a long time, but it's not too far off.  It might even be a little faster for fast blocks, but overall, it's probably 5% slower than b11, but it won't get into the "invalid prime" problem.

There were two bugs in b11:
  - If it went too long before a new block was found, it could overflow the 256 bit riecoin nonce field -- I changed some constants and didn't re-do the math on them.  This is now computed based upon those constants, so the bug shouldn't reappear.

  - It could run for too long without updating its block time field, which would cause ypool to reject it even if it contained a valid share or block.  I've mostly kinda sorta fixed this, but the current version reflects a compromise between safety (updating the time field frequently) and efficiency (running as long as possible after computing a target).  This one's a "let's run it and see how it does" kind of thing.

For people who like to read the source of these things, the internals have gotten a little cleaner, which hopefully makes the logic a bit more clear.  In particular, I think the offset computation should be easier to read and understand now, for those seeking to understand what makes these miners tick.

  -Dave

For some reason, when starting this up on my server it hangs after the "Launching miner..." and just exits. I'm on CentOS 6.4 and had to compile against custom built gmp as the one CentOS had was outdated. No errors or anything though. I also tried your prebuilt binaries, same thing.

[JohnnyCrypto]

Memory usage is related to the sieve size (-s), and by the amount of cores your CPU has, or threads you use. Default sieve size is -s 500000000. If you find it is using too much memory, or freezes or crashes upon running the program, just decrease the sieve amount to a lower size. If you have more memory you might want to increase it, it may help performance. For reference i'm using -s 675000000 for the 8 core i7, which uses about 6.5 GB of RAM. On other systems with 4 core i5's i can use -s 900000000 which uses about 6 GB.


https://www.cs.cmu.edu/~dga/crypto/ric/readme.html

[gpools]

Why price riecoin down