[XPM] Primecoin Built-in Miner Sieve Performance Issue

[dudeguy]

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 .. 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!  


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  


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

[1715038286]

1715038286

[1715038286]

1715038286

[1715038286]

1715038286

[1715038286]

1715038286

[eule]

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. 

[TheSwede75]

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?

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]

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]

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]

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]

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]

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]

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]

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]

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]

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]

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]

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]

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,

[TheSwede75]

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]

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 !!!!  

[Chemisist]

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   )

[fabrizziop]

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   )

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]

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$  



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

lmao.  grr.