| Author |
Message |
|---|
John Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
                 
|
|
Welcome to the GFN Prime Search PRPNet Challenge
Taking the lead from rogue's November post about the anniversary of Pierre de Fermat's death, we decided to make January a focus on the PSA's Generalized Fermat Number Prime Search. The project is offered through PrimeGrid's PRPNet. NOTE: PrimeGrid's BOINC Challenge series doesn't start until 3 February 2011.
A 3 Day PRPNet Challenged will take place from 12 Jan 2011 00:00 to 15 Jan 2011 00:00. While these PRPNet Challenges will become more organized this year, they remain relatively an informal event...a gathering of resources for the goal of finding primes. The specific goal of this Challenge will be to find PrimeGrid's first GFN prime.
PRPClient v4.1.1c with CUDA support is still being tested and can be found here: PRPClient v4.1.1c beta. (EDIT: File server is experiencing bandwidth problems...please download from this location.) Builds are available for Linux64, Mac, and Windows. v4.1.1c also supports team affiliations but unfortunately, the servers have yet to be updated so we will not be able to take advantage of this feature for the first Challenge.
All 4 GFN servers are available for the Challenge.
- server=GFN32768:0:1:prpnet.primegrid.com:12005
- server=GFN65536:0:1:prpnet.primegrid.com:12003
- server=GFN262144:0:1:prime2u.com:11002
- server=GFN524288:0:1:prime2u.com:11001
Challenge Stats
The GFN Prime Search will be featured for the entire month of January. The monthly goal will be to find a GFN prime for each of the following N's: 32768, 65536, 262144, 524288. We have 5 deeply sieved files and have yet to find a single GFN prime. The known primes listed in the Top 5000 primes database are:
With the breakthrough of porting Genefer 2.2.0 to the GPU (CUDA), the search has become much more attractive. Testing times for the higher N's have been drastically reduced. NOTE: GPU app requires double precision support. DP capable CUDA cards are: nVidia GTX260-295, 400 and 500.
The GFN Prime Search is open to the following platforms.
- GeneferCUDA - used on Windows and Linux64, CUDA 3.2+
- GeneferCUDAMacIntel - used on MacIntel running OS 10.5 or later, CUDA 3.2+
- GenefX64 - used on Windows 64
- Genefer - used on Linux 32&64, Windows 32&64, MacIntel, MacPPC
- Genefer80 - used on Windows 32&64
However, the GeneferCUDA and GenefX64 applications offer the most efficient testing. It is these two applications that provide the fastest path to finding GFN primes.
About the Generalized Fermat Number Prime Search
This search is for primes of the form b^2^n+1. The numbers F(b,n) = b^2^n+1 (with n and b integers, b greater than one) are called generalized Fermat numbers. In the special case where b=2, they are called Fermat numbers, named after Pierre de Fermat who first studied them.
The original Generalized Fermat Prime Search by Yves Gallot was very active from 2001-2004. It was a premier project ranking second only to GIMPS in organization and size of primes found. Now a decade after its debut, it's about time to revitalize the search. Thanks in large part to David Underbakke, Mark Rodenkirch, and Shoichiro Yamada, the GFN prime search is ready to make a comeback.
David Underbakke's AthGfn64 for sieving and his Genefx64 for primality testing, along with Shoichiro Yamada's GeneferCUDA will be the main tools utilized in the search. These can be used to test relatively small bases and are the most efficient GFN primality testing applications for the CPU and GPU respectively. Yves Gallot's Genefer and Genefer80 along with pfgw will also be used for the higher bases.
The search range is as follows:
N = 2, 4, 8, 16 for b < 1,000,000,000
N = 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768, 65536, 131072, 262144, 524288, 1048576, 2097152, 4194304, 8388608, 16777216 for b < 100,000,000
Much of the smaller N work has already been completed. The main search will start at N=32768 in PRPNet (prpnet.primegrid.com:12005). We'll test some gaps and then start testing above 3149688^32768+1.
As sieving progresses, 65536, 131072*, 262144, and 524288 will be added to PRPNet so users can have a selection of testing times from which to choose. NOTE: There's only been 1 prime found at n=262144 and none found at n=524288. Therefore, finds at those levels will be quite significant. Of course, they'll take some time to test as well. :)
NOTE: *131072 is currently reserved.
Sieve depths as of 29 August 2010 are:
n = 32768; p=1000P
n = 65536; p=1500P
n = 131072; p=2000P
n = 262144; p=2510P
n = 524288; p=3070P
For more information about generalized Fermat numbers and primes, please visit these links:
For more information about Fermat numbers and primes, please visit these links:
For more information about Pierre de Fermat, please visit these links:
A special thanks to David Underbakke for his AthGfn6v and Genefx64 and to Mark Rodenkirch for PRPNet. Also, thanks to them both for their collaboration in updating Genefx64, Genefer, and Genefer80 to prepare the programs for a distributed effort. And finally a special thanks to Shoichiro Yamada for his GeneferCUDA.
____________
|
|
|
 Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
                    
|
|
Just as it looks straight out, it should be possible to offer some Challenge stats like on "Last 10 dayes of 2010" Challenge. As this competition is spread across 4 ports, the effort is a little larger but I make good progress to solve the problems.
Thanks for your patience
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
 |
|
|
|
|
|
It seems whenever I terminate the client by pressing ctrl c it stop as normal but when I re start the client it starts the checking process from the beginning. It does not seem to re start the test from when I terminated it. Is this correct
PS in the file there is a .sav file but the prpclient say starting from the .in file |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
Can you tell us which ranges have already been tested by PrimeGrid? The min B on the status pages clearly does not reflect where PrimeGrid started the search.
I think that you meant "b < 100,000,000" instead of "n < 100,000,000" in the first post.
Does PrimeGrid intend to get a server for n = 1048576 or n = 2097152 up and running? |
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
Just as it looks straight out, it should be possible to offer some Challenge stats like on "Last 10 dayes of 2010" Challenge. As this competition is spread across 4 ports, the effort is a little larger but I make good progress to solve the problems.
Thanks for your patience
Challenge stats have come to a publishable state.
They are reachable via URL http://vcn94.homelinux.org/PRPNet/challenge_stats.php
EDIT
to prevent questions: yes the stats are empty until challenge starts ...
A bold Thank You! to all crunchers at the GFN ports in the last days, you have support my testings and made it in the beta test stats (beta winner is s-yama ;) )
you can get an impress here (availability not guaranteed).
If you have any suggestion, feel free to send me a Private Message
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Can you tell us which ranges have already been tested by PrimeGrid? The min B on the status pages clearly does not reflect where PrimeGrid started the search.
I think that you meant "b < 100,000,000" instead of "n < 100,000,000" in the first post.
Does PrimeGrid intend to get a server for n = 1048576 or n = 2097152 up and running?
The general approach to the search can be seen below.
Thanks for the typo catch. It's been corrected. As for N = 1048576 and N = 2097152, it was part of our original sieve. However, testing times seemed so long that we decided to focus on the lower N first. Now that GeneferCUDA brought those times down, we should probably revisit those sieves. Sieves were stopped at the following depths:
N = 1048576, p=49453T
N = 2097152, p=53893T
N = 4194304, p=54245T
N = 8388608, p=15000T
N = 16777216, p=15017T
We'll be testing the following for Top 5000 primes:
NOTE: top b limits are listed as GeneferCUDA, Genefx64, Genefer, Genefer80
N=32768
553602<b<743788
855124<b<999236
1277444<b<1519380
1519380<b<1755378
1755378<b<1909372
2034902<b<2147196
2167350<b<2279250
2696506<b<2973894
3000336<b<3109540
3149688<b<10021136
10021136<b
Top limits: 1.87M, 1.5M, 2.1M, 65M+
N=65536
671600<b<843832
1057476<b<1266062
1540550<b<1828502
1874512<b<2162068
2187182<b<17629398
17684828<b<19502212
19502212<b
Top limits: 1.53M, 1.2M, 1.7M, 51M+
N=131072
130816<b<386892
386892<b<572186
572186<b<1176694
1176694<b<1361244
1372930<b
Top limits: 1.255M, 1M, 1.4M, 42M+
N=262144
24518<b
Top limits: 815K, 865K, 1,2M, 35M+
N=524288
2<b
Top limits: 710K, 735K, 975K, 29M+
____________
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1871
ID: 352
Credit: 2,945,079,561
RAC: 1,561,307
                           
|
|
In the other thread, I asked if we need to sieve deeper.
I haven't done any more sieving since then, ranges are as described.
What about lower Ns primality testing? I have done all for N<=2048 up to b=100M, N=4096 is up b<6M5
I guess the question now would be: do we go wider with testing? Or are we gonna finish lower Ns first?
Do we open higher Ns now or once a prime is found for lower N? For example there is known prime for N=524288, we open N = 1048576?
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Here's a time comparison chart. GeneferCUDA was timed on a GTX 570 at stock speeds and the rest were timed on a Q6600 at 2.4Ghz 2G RAM. The machine is Windows 64. Time is ms/mul.
GeneferCUDA GenefX64 Genefer Genefer80
N=256 - 0.00381 0.00572 0.01000
N=512 - 0.00858 0.01240 0.02100
N=1024 - 0.01720 0.02670 0.04860
N=2048 - 0.03630 0.05530 0.10100
N=4096 - 0.08010 0.12200 0.23300
N=8192 0.272 0.20600 0.27500 0.53400
N=16384 0.278 0.42800 0.56400 1.11000
N=32768 0.300 0.91400 1.22000 2.38000
N=65536 0.367 1.89000 2.69000 5.31000
N=131072 0.481 4.39000 5.61000 11.70000
N=262144 0.614 9.28000 13.00000 26.60000
N=524288 1.010 22.00000 28.30000 61.50000
N=1048576 1.790 51.80000 73.30000 141.00000
N=2097152 3.450 115.00000 164.00000 299.00000
As can be seen, anything less than N=16384 should be tested on the CPU. Anything N=16384 is more efficient on at GPU (CUDA).
Approximate b limits are in the next chart:
GeneferCUDA GenefX64 Genefer Genefer80
N=32768 1.87M 1.5M 2.1M 63.6M
N=65536 1.53M 1.2M 1.7M 50.9M
N=131072 1.255M 1M 1.4M 41.8M
N=262144 815K 865K 1.2M 34.7M
N=524288 710K 735K 975K 28.8M
N=1048576 ??? 600K 790K 23.5M
N=2097152 ??? 495K 660K 19.9M
Once these limits are reached, the remaining will be completed with pfgw.
____________
|
|
|
|
|
|
if we could get GPU seiveing for GFN numbers, that migth make further sieving worthwhile..
without knowing how long seiveing takes
i'd suggest more cpu sieving may not be worthwhile. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
if we could get GPU seiveing for GFN numbers, that migth make further sieving worthwhile..
without knowing how long seiveing takes
i'd suggest more cpu sieving may not be worthwhile.
Using AthGfn64 on a CPU, sieving is definitely still worthwhile for all N > 16384. We have a large b range that we're sieving. Therefore, if the intentions are to primality test the entire range, for the larger N, we are far from optimal.
We decided to start primality testing because we were at a moderate depth with the sieve...but more practically, we simply wanted to find a GFN prime. :)
____________
|
|
|
|
|
|
I'm running GeneferCUDA on my GTX 460 with the 64 bit Linux version of PRPClient v4.1.1c beta. The client gets work from the test port (GFN262144TEST) which is set up as backup port (0:0) but is unable to get work from port 11001 (GFN524288) which is set up as as one of two main ports (50:1) to query for work in conjunction with port 11002 (GFN262144, also 50:1) which currently doesn't seem to have work.
The card completes one GFN262144 task from the test port in about an hour @ stock clocks, but I would like to know why it doesn't get work from the GFN524288 port (11001).
____________
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1871
ID: 352
Credit: 2,945,079,561
RAC: 1,561,307
|
As can be seen, anything less than N=16384 should be tested on the CPU. Anything N=16384 is more efficient on at GPU (CUDA).
Well, the other aspects to consider.
1. CUDAGenefer has it's limits too, some ranges (for exaplme 32768) will fall to Genefer80.
2. Q6600 has 4 cores, so it can do 4x more work.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
I'm running GeneferCUDA on my GTX 460 with the 64 bit Linux version of PRPClient v4.1.1c beta. The client gets work from the test port (GFN262144TEST) which is set up as backup port (0:0) but is unable to get work from port 11001 (GFN524288) which is set up as as one of two main ports (50:1) to query for work in conjunction with port 11002 (GFN262144, also 50:1) which currently doesn't seem to have work.
The card completes one GFN262144 task from the test port in about an hour @ stock clocks, but I would like to know why it doesn't get work from the GFN524288 port (11001).
Update:
On port 11001 (GFN524288) an older server version is listening compared to the test port 13000 (GFN262144TEST).
[2011-01-09 16:36:39 CET] GFN524288: Getting work from server prime2u.com at port 11001
...
[2011-01-09 16:36:39 CET] 4: received [ServerInfo: 8 3.3.0]
[2011-01-09 16:36:40 CET] 4: select indicates socket has data, but recv returned nothing
...
[2011-01-09 16:36:40 CET] 4: received [INFO: No available candidates are left on this server.]
...
[2011-01-09 16:36:44 CET] 4: received [ServerInfo: 8 4.1.1 0]
[2011-01-09 16:36:44 CET] 4: received [WorkUnit: 5488^262144+1 1294587424 5488 262144]
[2011-01-09 16:36:45 CET] 4: select indicates socket has data, but recv returned nothing
[2011-01-09 16:36:45 CET] 4: received [End of Message]
____________
|
|
|
|
|
|
a question! unfortunately my registration email up PrimeGrid has been closed and now i have another. If, in PRP client, I put my name but I put the new email (different from the my PrimeGrid's profile settings) my credits are added to the same? |
|
|
|
|
|
Not one to miss a challenge, but with only useless 9800GT and and ATi 4800 cards in my hotbox, I bit the bullet & splashed out on a GTS450 to give me some more GPU-power.
Replaced the ATi with the new GTS450 & uninstalled ATi drivers, leaving latest (260.99) NVidia drivers installed.
Ran latest prpclient-4.1.1cbeta-windows - but GeneferCUDA gripes about that missing cudart32_32_16.dll
Tried renaming that suggested cudart.dll to cudart32_32_16.dll - but then it just gripes about a cufft32_32_16.dll
Does this mean I now have to go thru the painful process of downloading & installing hundreds of megs of Cuda-Toolkit?
Or is there an easier way to get those missing DLLs??
____________
 |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1871
ID: 352
Credit: 2,945,079,561
RAC: 1,561,307
|
|
IIRC, latest PRPNet Client should have .dll incluced.
Perhaps you can download it directly via http://www.primegrid.com/download/cudart.dll
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Ran latest prpclient-4.1.1cbeta-windows - but GeneferCUDA gripes about that missing cudart32_32_16.dll
Tried renaming that suggested cudart.dll to cudart32_32_16.dll - but then it just gripes about a cufft32_32_16.dll
Both cudart.dll and cufft.dll are provided in the 4.1.1c prpclient packages. If anyone has a solution, please post as others are affected by this exact same problem.
____________
|
|
|
|
|
|
I got the same error.
Downloaded correct package with correct dll.
Still got same error.
Renamed it like our friend tried to, then got a different
error for a different dll name.
Bit the bullet and downloaded the CUDA toolkit at
over +100MB.
Now, no more missing dll errors, however still does not work.
It starts up fine but then says no data, and suggests
I am missing genfer.log.
I am not.
I would suggest if possible to do a disk image or backup if you can
and download toolkit...see if that fixes the missing dll problem.
It did for me but still not working for some other reason.
Cheers |
|
|
|
|
|
I think I found my problem.
I have a GTS250.....which I
believe is not double precision.
Oh well...will have to save up to play
with the big boys. |
|
|
|
|
|
This is the error I got trying to run with CUDA toolkit
and a GTS250........
2011-01-09 16:34:38 CET] GFN262144TEST: No data in file [genefer.log]. Is genefer broken?
(debuglevel=2).
|
|
|
|
|
This is the error I got trying to run with CUDA toolkit
and a GTS250........
2011-01-09 16:34:38 CET] GFN262144TEST: No data in file [genefer.log]. Is genefer broken?
(debuglevel=2).
I get these messages here too but otherwise the GPU works (at least on the test port)...
ralf@quadriga ~/Beta/prpclient-4.1.1cbeta-linux_64-GTX-460/prpclient-1 $ cat prpclient.log | grep broken
[2011-01-09 16:34:38 CET] GFN262144TEST: No data in file [genefer.log]. Is genefer broken?
...
[2011-01-10 20:03:41 CET] GFN262144TEST: No data in file [genefer.log]. Is genefer broken?
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
Each version of genefer will output the test result to a file called genefer.log. The client reads from that file to determine if the test revealed a PRP or a composite. If a composite is found, then it returns the residue (from the file) to the server.
If the file is not created, it implies the genefer was terminated with a ctrl-C or that genefer terminated for some other reason without creating that file.
If the file is not found, then the client will output the error.
I see a problem in genefer where it doesn't output the roundoff issue to the log. It should be doing that so that the client can see it and act appropriately.
I think I found the problem with the client deleting the checkpoint file. I need to work out a solution with John and David and the Shoichiro. |
|
|
|
|
I'm running GeneferCUDA on my GTX 460 with the 64 bit Linux version of PRPClient v4.1.1c beta. The client gets work from the test port (GFN262144TEST) which is set up as backup port (0:0) but is unable to get work from port 11001 (GFN524288) which is set up as as one of two main ports (50:1) to query for work in conjunction with port 11002 (GFN262144, also 50:1) which currently doesn't seem to have work.
The card completes one GFN262144 task from the test port in about an hour @ stock clocks, but I would like to know why it doesn't get work from the GFN524288 port (11001).
Update:
On port 11001 (GFN524288) an older server version is listening compared to the test port 13000 (GFN262144TEST).
[2011-01-09 16:36:39 CET] GFN524288: Getting work from server prime2u.com at port 11001
...
[2011-01-09 16:36:39 CET] 4: received [ServerInfo: 8 3.3.0]
[2011-01-09 16:36:40 CET] 4: select indicates socket has data, but recv returned nothing
...
[2011-01-09 16:36:40 CET] 4: received [INFO: No available candidates are left on this server.]
...
[2011-01-09 16:36:44 CET] 4: received [ServerInfo: 8 4.1.1 0]
[2011-01-09 16:36:44 CET] 4: received [WorkUnit: 5488^262144+1 1294587424 5488 262144]
[2011-01-09 16:36:45 CET] 4: select indicates socket has data, but recv returned nothing
[2011-01-09 16:36:45 CET] 4: received [End of Message]
Update 2:
I've identified the source of the problem:
Server 4.1.1 accepts if the client sends only genefercuda but Server 3.3.0 also needs genefer or otherwise it will send no work. I had commented out the genefer entry and the server 3.3.0 doesn't like this ;)
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
I've identified the source of the problem:
Server 4.1.1 accepts if the client sends only genefercuda but Server 3.3.0 also needs genefer or otherwise it will send no work. I had commented out the genefer entry and the server 3.3.0 doesn't like this ;)
That is likely. There are definitely some problems with server 3.x and genefer. Lennart hasn't had time to upgrade to 4.x. |
|
|
|
|
Both cudart.dll and cufft.dll are provided in the 4.1.1c prpclient packages. If anyone has a solution, please post as others are affected by this exact same problem.
I tried the same renaming trick with cufft.dll as well, but then I get..
"The procedure entry point cudaFuncSetCacheConfig could not be located in the dynamic link library cudart32_32_16.dll"
- so obviously that supplied cudart.dll is not as complete as it needs to be...
It looks like for now the only solution is to install "The Beast" Cuda-toolkit so I'm going to be biting that particular bullet as well!
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
PRPClient Update
PRPNet v4.1.2beta has been released for testing which addresses most of the issues experienced the past few days. You can download PrimeGrid's full PRPClient packages or just client only files. They are available here:
http://uwin.mine.nu/PRPNet_files/
Thanks for all the help in troubleshooting.
[EDIT] Windows package now includes cudart32_32_16.dll and cufft32_32_16.dll from the CUDA Toolkit 3.2 XP32 installation.
____________
|
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
|
Stats are running :)
The computing for the Score deduction of pending tests will take place today evening.
You may watch a decrease of Score, if your PRPNet clients was running on GFN ports before 00:00 UTC.
Have fun with the challenge
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
               
|
|
Nice job on the stats page.
Neo
____________
 |
|
|
|
|
|
Are the amount of points on the stats page the amount of point s that will be given to your score at prime grid or are the amount of project staging primegrid points given in a different way?
|
|
|
|
|
Are the amount of points on the stats page the amount of point s that will be given to your score at prime grid or are the amount of project staging primegrid points given in a different way?
PRPNet points don't translate 1:1 into PSA credits / BOINC cobblestones.
____________
|
|
|
|
|
Are the amount of points on the stats page the amount of point s that will be given to your score at prime grid or are the amount of project staging primegrid points given in a different way?
PRPNet points don't translate 1:1 into PSA credits / BOINC cobblestones.
So how are the credits translated from prpnet to primegrid? I understand the timing is off but was just wondering how the amount of credits differ.
____________
@AggieThePew
 |
|
|
|
|
Are the amount of points on the stats page the amount of point s that will be given to your score at prime grid or are the amount of project staging primegrid points given in a different way?
PRPNet points don't translate 1:1 into PSA credits / BOINC cobblestones.
So how are the credits translated from prpnet to primegrid? I understand the timing is off but was just wondering how the amount of credits differ.
This depends on the ports you crunch and various other factors. When I began crunching PRPnet (LLR tests only) I ended up with a(n estimated) ratio of 1 cobblestone per 18 PRPnet points. I guess that more than 90% of my 4.98M PRPnet points are the result of the tpsieve / ppsieve PSA work. I sieved 9.1T (PPR3M) and 740G (RSP2M) and found 5,589,990 factors.
____________
|
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
|
There was an unplanned update to a new PRPNet-Server version on GFN ports :(
This causes a missbehavior of handling pending credits, which can not be fixed during the challenge. Unfortunately I had to turn off this functionality, so all returned work is influencing challenge stats, whether it was downloaded before the challenge start or not.
Sorry for this
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
|
|
|
Looks like a prime number was found on the GFN65536 port by Lumiukko.
Lumiukko
Points: 123719
Tests: 82
PRPs: 0
Primes: 1
and Sysadm's stats lists it as new.
Congratulations!
____________
|
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
|
http://prpnet.primegrid.com:12003/user_primes.html
until now it is hidden ...
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
|
|
|
I think the PrimeSearchTeam should change it's name to PrimeFindTeam ;)
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
Looks like a prime number was found on the GFN65536 port by Lumiukko.
Lumiukko
Points: 123719
Tests: 82
PRPs: 0
Primes: 1
and Sysadm's stats lists it as new.
Congratulations!
I'm guessing that this one, http://primes.utm.edu/primes/page.php?id=12 is the one found. I suspect that it hadn't been removed from the database. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Looks like a prime number was found on the GFN65536 port by Lumiukko.
Lumiukko
Points: 123719
Tests: 82
PRPs: 0
Primes: 1
and Sysadm's stats lists it as new.
Congratulations!
Sadly, this is an existing prime. Part of the current work is double checking some open ranges that were left behind by the previous effort 6 years ago. Unfortunately, it looks like a prime slipped into the data set.
[EDIT] Yes, rogue is correct.
[EDIT2] I haven't checked yet but b=857678 may be in there as well.
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
Out of curiosity, who has 131072 reserved? |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
BTW, can the false primes/PRPs be removed from the 262144 and 524288 databases? The server_stats page says that primes were found, but there are no user_primes.
And you might want to consider fixing the HTML title on the various servers. |
|
|
|
|
|
ok...how come my GPU cruncher will only get GFN work on GFN3631444TEST:13000 even though I uncommented ALL of the GFN servers?
It's definitely working, that's for sure - absolutely tearing through tasks, maybe 50 minutes per, but as far as this challenge goes, they aren't the right ones!
To make things even more confusing, my Mac won't get any GFN work for any of its 4 CPUs.
Basically, PRPNet is working but I can't seem to do any work related to the challenge. :P
Since the first post seems to imply that GeneferCUDA can be used in this challenge, I'd like to know why exactly the client simply fails to start if it is told to use anything but the GFNTEST server...
____________
 |
|
|
|
|
ok...how come my GPU cruncher will only get GFN work on GFN3631444TEST:13000 even though I uncommented ALL of the GFN servers?
It's definitely working, that's for sure - absolutely tearing through tasks, maybe 50 minutes per, but as far as this challenge goes, they aren't the right ones!
To make things even more confusing, my Mac won't get any GFN work for any of its 4 CPUs.
Basically, PRPNet is working but I can't seem to do any work related to the challenge. :P
Since the first post seems to imply that GeneferCUDA can be used in this challenge, I'd like to know why exactly the client simply fails to start if it is told to use anything but the GFNTEST server...
My best guess is the %'s of the uncommented servers
I think your ini file looks something like this:
server=GFN65536:0:1:prpnet.primegrid.com:12003
server=GFN32768:0:1:prpnet.primegrid.com:12005
server=GFN262144:0:1:prime2u.com:11002
server=GFN524288:0:1:prime2u.com:11001
server=GFN262144TEST:100:1:prime2u.com:13000
this has 100% of the going to GFN262144TEST
with the other servers being a backup when there is no work on port 13000
It should look something like this:
server=GFN65536:25:1:prpnet.primegrid.com:12003
server=GFN32768:25:1:prpnet.primegrid.com:12005
server=GFN262144:25:1:prime2u.com:11002
server=GFN524288:25:1:prime2u.com:11001
server=GFN262144TEST:0:1:prime2u.com:13000
with the work going to the 4 GFN servers 25% each
=> you can change the % to what ever you want,
Just as long they all add up to exactly 100
Hope this will fix your problem.
If not, Just post your ini file.
Help will be there before you know it
Crunch On,
Steve Martin
____________
From the High Desert in New Mexico
 |
|
|
|
|
Sadly, this is an existing prime. Part of the current work is double checking some open ranges that were left behind by the previous effort 6 years ago. Unfortunately, it looks like a prime slipped into the data set.
Now we know it's a genuine high quality double (or even triple?) checked prime complete with the PG seal of approval ;)
____________
|
|
|
LumiukkoVolunteer tester
Send message
Joined: 7 Jul 08
Posts: 160
ID: 25183
Credit: 698,645,928
RAC: 113,033
                      
|
Now we know it's a genuine high quality double (or even triple?) checked prime complete with the PG seal of approval ;)
So this prime was now rechecked with GeneferCUDA using a GTX275-series GPU in a Win7 machine.
--
Lumiukko
|
|
|
|
|
|
Got it. Thanks Steve. I somehow didn't notice that.
____________
|
|
|
|
|
|
I must be blind or something but where do I get the 64 bit apps for a Windows platform? I looked at the download site and like I said I must be blind cause I didnt see it.
____________
@AggieThePew
|
|
|
|
|
I must be blind or something but where do I get the 64 bit apps for a Windows platform? I looked at the download site and like I said I must be blind cause I didnt see it.
The windows archive contains a few 64 bit apps (pfgw64, etc.). Since 64 bit apps are (most of the time) not faster when it comes to primality testing no 64 bit builds are distributed.
____________
|
|
|
|
|
|
Copied prpnet folder to a different server from a server that is working and now get this error
uwin.min.nu:10000 connect to socket failed with error 10060
any ideas and is it possible it's looking for a proxy server or something
____________
@AggieThePew
|
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
Looks like we have another; hopefully a new discovery.
____________
|
|
|
|
|
Looks like we have another; hopefully a new discovery.
It's most likely this:
[EDIT2] I haven't checked yet but b=857678 may be in there as well.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Looks like we have another; hopefully a new discovery.
It's most likely this:
[EDIT2] I haven't checked yet but b=857678 may be in there as well.
Yes, that is correct.
However, we do have two new primes at N=32768. More info to come
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
Exciting. I just hope that they are not of the "rediscovered" variety. |
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
|
prpnet.primegrid.com:12005 will dry out soon
some workunit are on the run, but there is no new work ...
http://prpnet.primegrid.com:12005/server_stats.html
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
|
|
|
Who has 131072 reserved ?
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
Who has 131072 reserved ?
David Underbakke, the brains behind genefx64. |
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1180
ID: 18646
Credit: 469,176,790
RAC: 117,742
|
|
challenge stats are final!
Congratulation to s-yama for the first place in over all, followed by Ross* and Lumiukko.
at single ports:
GFN32768: first is s-yama followed by Ross* and Lumiukko
GFN65536: first is LookAS followed by Ross* and samuel7
GFN262144: first is pschoefer followed by Steve* and biancaw
GFN524288: first is s-yama followed by Greg and Ross*
more details at
http://vcn94.homelinux.org/PRPNet/Januarys_PRPNet_Challenge_Project_on_GFN.html
I hope you enjoy it and I look forward to seeing you at the next challenge.
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
|
|
|
|
|
|
Obviously this GeneferCuda lark is a lot trickier than I imagined..
Got it working after installing toolkit, and set up all 8 clients the same... - 25% to each of the 4 servers with all genefer* options active, and left overnight to crunch during the challenge.
Next morning, all 8 were still crunching the same GeneferCuda task they had downloaded the evening before and their countdown rate implied they would still be at it for weeks to come!
- so I heaved a great sigh, trashed the lot, and went back to Boinc. :-(
I assume they don't like fighting each other for Cuda-core-time and I should only enable 1 client for Cuda??
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1194
ID: 12001
Credit: 18,565,548
RAC: 0
|
Obviously this GeneferCuda lark is a lot trickier than I imagined..
Got it working after installing toolkit, and set up all 8 clients the same... - 25% to each of the 4 servers with all genefer* options active, and left overnight to crunch during the challenge.
Next morning, all 8 were still crunching the same GeneferCuda task they had downloaded the evening before and their countdown rate implied they would still be at it for weeks to come!
- so I heaved a great sigh, trashed the lot, and went back to Boinc. :-(
I assume they don't like fighting each other for Cuda-core-time and I should only enable 1 client for Cuda??
If you have only one GPU, then yes. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Looks like we have another; hopefully a new discovery.
It's most likely this:
[EDIT2] I haven't checked yet but b=857678 may be in there as well.
Yes, that is correct.
However, we do have two new primes at N=32768. More info to come
A third N=32768 prime has been found. We are awaiting crediting questions at the Prime Pages for all three primes. As soon as that's resolved, they will be submitted.
____________
|
|
|
|
|
|
Where is ATI suport? They have really fast DP. |
|
|
|
|
Where is ATI suport? They have really fast DP.
Looks like the current version of their FFT library (OpenCL) is not up to the task ("Single precision FFTs; forward and reverse") though this may change. The chip architectures (caches / memory hierarchy, etc.) could pose some additional difficulties to make efficient use of the DP capabilities especially when larger FFTs are required.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
GFN Prime Search PRPNet Challenge - Final Results
The GFN Prime Search PRPNet Challenge was a nice success. Congratulations to Senji Yamashita (s-yama) as the top individual scorer. Additional congratulations goes out to Amy Chambers (Buckeye74) who found PrimeGrid's first GFN prime in this search. This prime was quickly followed less than an hour and a half later by another prime discovered by Ross Goudie (Ross*). The goal of finding at least one GFN prime was achieved. For complete scores, please see here.
The search continues with the month long goals of finding a prime at each of the following N's: 65536, 262144, 524288. For more information, please see the GFN Prime Search. BTW, the GPU really shines at the higher N's. :)
NOTE: We are still awaiting word from Chris at the prime pages on proper crediting of these primes.
Overall
rank UID tests done Score
1 s-yama 3,001 6,359,292
2 Ross* 1,546 4,908,561
3 Greg 85 3,287,170
4 pschoefer 614 2,911,967
5 Lumiukko 1,397 2,449,542
6 Steve* 127 2,342,252
7 samuel7 641 2,089,545
8 biancaw 108 1,895,465
9 LookAS 725 1,335,832
10 MiHost 63 1,327,102
N=32768
rank UID tests done Score
1 s-yama 2,618 1,197,423
2 Lumiukko 980 448,421
3 Ross* 786 359,824
4 Genn 687 314,230
5 pschoefer 248 113,418
6 Buckeye74 210 96,106
7 BiBi 207 94,714
8 Steve_Martin@SETI.USA 149 68,137
9 Ralf_Recker 66 30,166
10 Robert_Collins 49 22,415
N=65536
rank UID tests done Score
1 LookAS 693 1,047,674
2 Ross* 616 933,110
3 samuel7 585 885,623
4 [PNT]_toms83 477 722,247
5 BiBi 428 646,961
6 Lumiukko 356 538,446
7 Steve_Martin@SETI.USA 304 459,886
8 Buckeye74 246 372,261
9 pschoefer 234 354,131
10 s-yama 221 333,576
N=262144
rank UID tests done Score
1 pschoefer 127 2,227,719
2 Steve* 123 2,163,692
3 biancaw 108 1,895,465
4 Ross* 103 1,808,879
5 s-yama 86 1,508,302
6 Meiner_Einer 63 1,104,956
7 sm5ymt 54 945,653
8 cpuid 48 842,061
9 samuel7 46 807,655
10 Lumiukko 46 805,678
N=524288
rank UID tests done Score
1 s-yama 76 3,319,991
2 Greg 75 3,272,078
3 Ross* 41 1,806,748
4 Artist 17 734,395
5 Sashixi 16 700,611
6 MiHost 15 660,905
7 Lumiukko 15 656,997
8 msft 11 479,172
9 Sideshow_Larry 10 434,571
10 samuel7 9 395,810
____________
|
|
|
|
|
|
Once all testing is completed on the GeneferCuda app will it automatically install like the CPU apps do or will GPU app require manual installing? |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1871
ID: 352
Credit: 2,945,079,561
RAC: 1,561,307
|
Once all testing is completed on the GeneferCuda app will it automatically install like the CPU apps do or will GPU app require manual installing?
If I understood correctly your question... PRPNet shoule be able to do testing using both GPU and CPU using one settings. What app will be used depends on b limits (see there)
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
Once all testing is completed on the GeneferCuda app will it automatically install like the CPU apps do or will GPU app require manual installing?
If I understood correctly your question... PRPNet shoule be able to do testing using both GPU and CPU using one settings. What app will be used depends on b limits (see there)
Honza, you almost understood correctly. Let me reword. After testing of GPU app FOR Genefer 2.2.0 will it be automatically downloaded from the Applications page? Like the Proth Prime Search (Sieve) GPU apps are installed now. 1.38 (cuda23) & 1.38 (ati13ati). I have a GTX470. |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1871
ID: 352
Credit: 2,945,079,561
RAC: 1,561,307
|
|
Nope, PRPNet is manual effort unlike BOINC.
(on the other side, no need to install anything, just run from a folder).
You will need to download package (which has correct app and DLLs) and set prpclient.ini
geneferexe=genefercuda.exe
Download http://uwin.mine.nu/PRPNet_files/prpclient-4.1.4beta-windows-gpu.7z
See Generalized Fermat Prime Search
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
