Bitcoin Forum
April 23, 2021, 07:37:58 AM *
News: Latest Bitcoin Core release: 0.21.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 [27] 28 29 30 31 32 33 34 35 36 37 »
  Print  
Author Topic: [XPM] Primecoin Built-in Miner Sieve Performance Issue  (Read 69059 times)
dudeguy
Member
**
Offline Offline

Activity: 182
Merit: 10



View Profile
July 12, 2013, 07:35:01 PM
 #521

I am running aroun 4k pps myself and not a block found since 3 am. I also have the processes spread over 4 separate boxes so it has to be REALLY bad luck, or someone is just growling up all the blocks (I think BlastBob prob runs around 100k pps right now)..
Not just Blastbob Tongue .. Running 60K here.. Still finding blocks but there's still 30-100 blocks between the blocks I find.

Now that's just naughty! Who's stealin 'dem blocks!  Shocked


I hope the block size becomes 0.000145 XPM by next week, so we've got a pre-manufactured scarcity on our hands just like BitBar--combine that with the new POF of this coin and resistance to GPU mining I think we've got a winner  Wink


Btw, how the F%^K do you mine 60K PPS? Server farm? I'm not even at 1K with 3 processors!
1619163478
Hero Member
*
Offline Offline

Posts: 1619163478

View Profile Personal Message (Offline)

Ignore
1619163478
Reply with quote  #2

1619163478
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
eule
Hero Member
*****
Offline Offline

Activity: 728
Merit: 501


View Profile
July 12, 2013, 07:36:51 PM
 #522

My last block: "confirmations" : 7263
Around 4k pps...
Based on your rate i should get one every 450 - 1500 blocks. Seems I have bad luck now, need to get more servers. Yesterday. Grin

TheSwede75
Full Member
***
Offline Offline

Activity: 224
Merit: 100



View Profile
July 12, 2013, 07:41:24 PM
 #523

My latest Windows builds. From Chemisist source:

Tuned for Sandy and Ivy Intel Core processors (AVX), O3:

https://www.dropbox.com/s/18bgecwqzsmwsh2/primecoin0712v2-avx.zip


Ivy Bridge ONLY build:

https://www.dropbox.com/s/f7fu0u0yk4i09il/primecoin0712v2-ivyonly.zip

XPM: AR2BpBnitqXudN67Ncuc9FfYVT8u9jNe7a

I have a 32core system, Intel® Xeon® Processor E7-8837

It doesn't get above 1150PPS, other systems go to 2500PPS with less powerful and older cores.

Any idea why? Smiley

CPU details: http://ark.intel.com/products/53576/

I have noticed that setting an actual core number works better then just default for some processors.
TheSwede75
Full Member
***
Offline Offline

Activity: 224
Merit: 100



View Profile
July 12, 2013, 07:45:37 PM
 #524

Really odd that I have a 50% PPS increase with the newest codes but haven't found a block in several hours. With the old code PPS was far lower but I found blocks ever few hours at least.
tyrion70
Legendary
*
Offline Offline

Activity: 934
Merit: 1000



View Profile
July 12, 2013, 07:46:52 PM
 #525

Well, If I get a block every 50 blocks that means total pps rate should be around 50X60K == 3M pps...

@TheSwede, the latest build gave me a block just 32 confirms ago.. so no issue there here..

Sunny King
Legendary
*
Offline Offline

Activity: 1205
Merit: 1003



View Profile WWW
July 12, 2013, 07:48:07 PM
 #526

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.
TheSwede75
Full Member
***
Offline Offline

Activity: 224
Merit: 100



View Profile
July 12, 2013, 07:50:22 PM
 #527

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

Understood. Now we just need someone to build a CUDA implementation and release it to the public before some basement hacker gets one working and runs it for himself. The theoretical speed of CUDA Mersenne solving is something like 1000x faster then CPU. A single SLI Nvidia rig could throw 25 million PPS if optimized on CUDA.
TheSwede75
Full Member
***
Offline Offline

Activity: 224
Merit: 100



View Profile
July 12, 2013, 07:51:01 PM
 #528

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

How can I see block-rate in the debug console?

yoshiyoshi
Member
**
Offline Offline

Activity: 89
Merit: 10



View Profile
July 12, 2013, 07:51:10 PM
 #529

Really odd that I have a 50% PPS increase with the newest codes but haven't found a block in several hours. With the old code PPS was far lower but I found blocks ever few hours at least.

difficulty is exponential, and now it gets very very hard to find block just with couple of CPUs.. Next station - porting algo for GPU
drummerjdb666
Full Member
***
Offline Offline

Activity: 244
Merit: 101



View Profile
July 12, 2013, 07:52:51 PM
 #530

for my 3770k should i use this?
https://www.dropbox.com/s/f7fu0u0yk4i09il/primecoin0712v2-ivyonly.zip

or sunny's recent update?   getting about 1k to 1300 pps..     tired of not finding blocks though.. 18 total since the release date.. but in the past 48 hrs nothing but 5 orphans
oroqen
Sr. Member
****
Offline Offline

Activity: 280
Merit: 250



View Profile
July 12, 2013, 07:53:08 PM
 #531

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

Understood. Now we just need someone to build a CUDA implementation and release it to the public before some basement hacker gets one working and runs it for himself. The theoretical speed of CUDA Mersenne solving is something like 1000x faster then CPU. A single SLI Nvidia rig could throw 25 million PPS if optimized on CUDA.
Gotta sort out the API before that
Sunny King
Legendary
*
Offline Offline

Activity: 1205
Merit: 1003



View Profile WWW
July 12, 2013, 07:55:06 PM
 #532

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

How can I see block-rate in the debug console?

Unfortunately this is not currently measured in the client. It can be tested on testnet though.
anonppcoin
Newbie
*
Offline Offline

Activity: 48
Merit: 0


View Profile
July 12, 2013, 08:10:14 PM
 #533

for my 3770k should i use this?
https://www.dropbox.com/s/f7fu0u0yk4i09il/primecoin0712v2-ivyonly.zip

or sunny's recent update?   getting about 1k to 1300 pps..     tired of not finding blocks though.. 18 total since the release date.. but in the past 48 hrs nothing but 5 orphans

That build is tailored for your CPU generation. It should work well.
tocket
Newbie
*
Offline Offline

Activity: 9
Merit: 0


View Profile
July 12, 2013, 08:21:32 PM
 #534

I also notice a decline in PPS using Chemisist's code compare to the official (0.11). Getting an average of about 1800 PPS vs 2200 with official. I'm only running 12 threads as well, so it's not just affecting users with high thread counts.
Chemisist
Member
**
Offline Offline

Activity: 99
Merit: 10



View Profile
July 12, 2013, 08:33:13 PM
 #535

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

How can I see block-rate in the debug console?

Unfortunately this is not currently measured in the client. It can be tested on testnet though.

Sunny, I've got my code running on testnet right now and it's happily generating blocks on two different machines.  The things that I checked were that my code was (1) generating blocks on the test net and (2) that the pps was higher than the built in code.  Though it seems that the higher pps might not actually be an improvement?I suppose a better comparison would be to run both on the same machine and then check the number of blocks generated per hour, right?

The blocktime value in the listtransactions command is in what unit of time? (I'm assuming seconds?)

e.g.
        "blocktime" : 1373660991,
        "blocktime" : 1373661018,
        "blocktime" : 1373661039,
        "blocktime" : 1373661052,
        "blocktime" : 1373661071,

btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF     ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd     xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
TheSwede75
Full Member
***
Offline Offline

Activity: 224
Merit: 100



View Profile
July 12, 2013, 08:33:58 PM
 #536

I also notice a decline in PPS using Chemisist's code compare to the official (0.11). Getting an average of about 1800 PPS vs 2200 with official. I'm only running 12 threads as well, so it's not just affecting users with high thread counts.

12 is prob 'fairly high' though. Seems the latest builds have problems with anything over 6/8 threads.
drummerjdb666
Full Member
***
Offline Offline

Activity: 244
Merit: 101



View Profile
July 12, 2013, 08:34:22 PM
 #537

k well either way.. for the noobs... if you install sunny's new client..

don't use this https://www.dropbox.com/s/f7fu0u0yk4i09il/primecoin0712v2-ivyonly.zip

it is making the wallet hang on splash screen..

I just copied the files into the wallet folder.. no worky


EDIT.  deleted sunny's newest qt folder...    (strange seeing how he's the real bad ass here and all!!!)

but am using anonppcoin's v0710 again.. replaced those files with the ivyonly build.. and well.. hellz yea!   1600pps @ true 4?  haha before i was getting 600 @ true 4

and um.. guy's WAS there a fork? or was that a bork in one of the modded wallet files?  because I too have only found 4 blocks in the past 48hrs.. and 3 of them orphaned... saw you other guy's having that issue


And Prop's to all you dev's working on project's like this!  Sunny king <3   The people optimizing..  I have had an obsession with opensource software for yrs.. but never really got into any kind of coding.. this whole community really makes me want to devote every minute of my life into learning some form of language..   Qt in itself being a mixture of languages you can implement.. c++. java, html, and the 2 others i can't remember.. seems like it would be impossible to learn qt without knowing the 5 languages it consists of... where should I start.. trying to learn at this point.. standard c?  but wouldn't that be like trying to learn a dead language? or is C++ something you could never learn without knowing C..   anyway done blabbing.. looking forward to being able to gpu mine this for a couple days..  with 2 6950's i don't it see it making me rich.. but damnit a few btc's on top of my current like .8 would be awesome..  yea yea ik ik poor noob !!!!  Smiley
Chemisist
Member
**
Offline Offline

Activity: 99
Merit: 10



View Profile
July 12, 2013, 08:36:43 PM
 #538

I also notice a decline in PPS using Chemisist's code compare to the official (0.11). Getting an average of about 1800 PPS vs 2200 with official. I'm only running 12 threads as well, so it's not just affecting users with high thread counts.

Well, you're running with more threads than I was able to test (highest I got to was 8 on my core i7-950), so I'm not entirely sure why this is, though it could be that all threads are trying to access a single variable which is determining how long to let the sieve be woven for.  I suppose that all these threads could end up blocking each other and cause a significant portion of idle time.  It would be far more effective to have a sieve weaving time variable (see line 11 in my version of the prime.cpp: static volatile int sieveBuildTime = 0;) for each individual boost thread, but I'm not sure how to do this as I am entirely unfamiliar with the Boost library (I'm not a c++ programmer  Sad )

btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF     ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd     xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
fabrizziop
Hero Member
*****
Offline Offline

Activity: 506
Merit: 500



View Profile
July 12, 2013, 08:39:21 PM
 #539

I also notice a decline in PPS using Chemisist's code compare to the official (0.11). Getting an average of about 1800 PPS vs 2200 with official. I'm only running 12 threads as well, so it's not just affecting users with high thread counts.

Well, you're running with more threads than I was able to test (highest I got to was 8 on my core i7-950), so I'm not entirely sure why this is, though it could be that all threads are trying to access a single variable which is determining how long to let the sieve be woven for.  I suppose that all these threads could end up blocking each other and cause a significant portion of idle time.  It would be far more effective to have a sieve weaving time variable for each individual boost thread, but I'm not sure how to do this as I am entirely unfamiliar with the Boost library (I'm not a c++ programmer  Sad )

Actually with my FX 8350 I get just a bit less speed with your code than the official 0.11, and with my sempron 145 I get around 20% less with your code. Maybe it doesn't work well for AMD architectures?.
drummerjdb666
Full Member
***
Offline Offline

Activity: 244
Merit: 101



View Profile
July 12, 2013, 08:39:28 PM
 #540

Please note primespersecond is not an accurate measure of actual performance. It has some correlations but if sieve round is reduced too short you could see inflated pps but not really faster performance. Only block rate is an accurate measure of true performance.

Understood. Now we just need someone to build a CUDA implementation and release it to the public before some basement hacker gets one working and runs it for himself. The theoretical speed of CUDA Mersenne solving is something like 1000x faster then CPU. A single SLI Nvidia rig could throw 25 million PPS if optimized on CUDA.


AWWWW!!!   This is going to make me really regret trading my gtx660 for a 6950 +30$  Sad Sad Sad



isn't it.     ISN'T IT!!!!....  ISN'T IT!?!!!.....  ISN'T IT?!!!  

lmao.  grr.
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 [27] 28 29 30 31 32 33 34 35 36 37 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!