| Author |
Message |
|---|
John Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
                 
|
|
Welcome to a week of Wieferich and Wall-Sun-Sun
This first week of 2012 (1 Jan - 7 Jan) is a home warming party for the new Wieferich and Wall-Sun-Sun prime search projects at PrimeGrid. These projects are looking for extremely rare and unique primes. For more information, please click on the individual project links above. PSA credit is available.
Project stats are available here: Wieferich | Wall-Sun-Sun
They are only available for 64bit OS's (Mac, Linux, Windows). They also require the updated PRPNet 5.0 release. New PRPClient packages are available in the PRPNet thread.
For existing v4.3.7 PRPNet users, it's probably best to get a clean install as there are several updates:
- prpclient v5.0.2 & readme_prpclient.txt
- master-prpclient-ini (new servers and wwww application)
- pfgw v3.6.0 & readme_pfgw.txt
- wwww v1.3 & readme_wwww.txt
For any help or issues with prpclient, please post in the PRPNet Discussion thread.
In order to participate in the two new projects, make sure you give them some project share. Below is an example of 50/50 share:
server=WIEFERICH:50:1:prpnet.mine.nu:13000
server=WALLSUNSUN:50:1:prpnet.mine.nu:13001 Remember that percent across all active projects must equal 100%.
NOTE: As Wall-Sun-Sun is already in first pass territory, please consider setting Wieferich with a higher percent (80/20 for example).
New to PRPNet
If anyone needs help setting up PRPNet, PM me and I can provide an individualized "pre-set" package. All you'll have to do is download, double click a couple of times and be quickly contributing to the project. All I need is the following:
- email=your@email.here
- userid=PG_username (Make sure it's unique. If you don't have a unique PG username, add your PG userID to it. For example, for me, John would become John_2449)
- clientid=clientID (can be anything you want...does NOT have to be PG clientID)
- teamid=teamname (see here Wieferich / Wall-Sun-Sun for teams already participating)
- OS (64bit ONLY), and how many cores you wish to run.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Task Lengths
Task lengths are ~1hr for Wieferich and ~45 minutes for Wall-Sun-Sun. If we get a good turnout for this event, it might be best to increase the cache to 2 or more. Below is an example of an 80/20 share with each project requesting 3 WU's at a time:
server=WIEFERICH:80:3:prpnet.mine.nu:13000
server=WALLSUNSUN:20:3:prpnet.mine.nu:13001
____________
|
|
|
|
|
|
Help for Mac users
If you get an error message saying that PRPNet cannot find the wwww program even though you made the suggested changes to the master_prpclient.ini, you also have to change the line about the wwww executable:
Originally it said:
// This is the name of the executable used for Wieferich and Wall-Sun-Sun searches.
// Available ONLY to 64bit OS. Uncomment if you have 64bit OS.
wwwwexe=wwww.exe
You have to change the last line to:
wwwwexe=./wwww
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
You have to change the last line to:
wwwwexe=./wwww
Thanks for the catch. It has been updated.
____________
|
|
|
|
|
|
A big thank you from me to John, Mark and Lennart for all the work to make these searches live. Absolutely awesome. |
|
|
 DoESVolunteer tester
Send message
Joined: 11 Oct 08
Posts: 784
ID: 30382
Credit: 74,878,735
RAC: 0
     
|
|
Yeah -- same from me & thanks for for the opportunity to be part of the beta test- it was a lot of fun
____________
Member of AtP
Shown here is an Australian native rat (Ratus Kickarsus) |
|
|
Sysadm@NbgVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1188
ID: 18646
Credit: 490,016,651
RAC: 0
                  
|
|
stats are still beta and can be found at http://u-g-f.de/PRPNet_beta/
I am working on combine the regular stats with the wwww-Projects, but some differences make it not easy ...
____________
Sysadm@Nbg
my current lucky number: 3749*2^1555697+1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
 |
|
|
|
|
|
Right, time to rack up some points for the cows on this one! ;)
Lets see where I can get at with my one computer!
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
 |
|
|
|
|
|
Wieferich seems to be down... (port 13000)
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
It appears that Dorais' and Klyve's definition did catch some "classical" "near" Wieferich primes but not all. So far, all |A| < = 100 that we've found were already discovered by them. However, we are finding other |A| < = 1000 that they did not search for.
We have already rediscovered the following:
3411159925463651 is a special instance (-1 -88 p)
3690728733648797 is a special instance (-1 +167 p)
3723113065138349 is a special instance (-1 +18 p)
3948546628939699 is a special instance (-1 +93 p)
4032459967159163 is a special instance (-1 +86 p) To see the full this we've disovered so far for |A| < = 1000, please see: Wieferich User Finds
Moving forward, we expect to rediscover the following:
4143792274787999 is a special instance (+1 +108 p)
4150209531584437 is a special instance (-1 -24 p)
5131427559624857 is a special instance (+1 -18 p)
5367369195612269 is a special instance (-1 -159 p)
5464249230405811 is a special instance (-1 -213 p)
5539428831517831 is a special instance (+1 +115 p)
5683778474515027 is a special instance (-1 -166 p)
5755502459289463 is a special instance (+1 +238 p)
6517506365514181 is a special instance (-1 -29 p) Note: It will be interesting to see if the gap between 4.15e15 and 5.13e15 is a valid gap for |A| < = 100.
With the way progress is moving, we could very well reach 6.7e15 within a couple of weeks. What an accomplishment that would be. :)
____________
|
|
|
|
|
|
I wish I could run it but I can't run the new wwww on my computer which is 32-bit.
____________
|
|
|
|
|
I wish I could run it but I can't run the new wwww on my computer which is 32-bit.
You can't use 64-bit Wubi: http://www.ubuntu.com/download/ubuntu/windows-installer ?
____________
Oh Bondage? Up Yours.
http://www.youtube.com/watch?v=ogypBUCb7DA
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 237,712,514
RAC: 2,203
                      
|
I wish I could run it but I can't run the new wwww on my computer which is 32-bit.
You can't use 64-bit Wubi: http://www.ubuntu.com/download/ubuntu/windows-installer ?
Unless he has a new computer, all the computers he has listed under BOINC have 32 bit CPUs. WUBI (or any other dual-boot or VM) is only helpful if you're running a 32-bit OS on a 64 bit CPU. If you're running a 32 bit CPU, you can't run the code at all.
____________
My lucky number is 75898524288+1 |
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
               
|
|
According to WubiGuide, he can use a 32bit-CPU.
Can I force Wubi to download and install a 32 bit version of Ubuntu?
Yes: either pre-download the appropriate 32 bit ISO manually and place it in the same folder as Wubi.exe or start Wubi with the "--32bit" argument.
To modify arguments, right-click Wubi.exe and select "Create Shortcut". Then right-click the shortcut, select Properties, and modify the Target line, for example: "C:\Documents and Settings\<user>\Desktop\wubi.exe" --32bit
It would be interesting to know, if the apps have a speed gain in 64bit or not. If not, then we would have the same situation like with the LLR-app.
Speed gain yes, but negligible with 5 percent.
____________
Best wishes. Knowledge is power. by jjwhalen
 |
|
|
|
|
According to WubiGuide, he can use a 32bit-CPU.
Can I force Wubi to download and install a 32 bit version of Ubuntu?
Yes: either pre-download the appropriate 32 bit ISO manually and place it in the same folder as Wubi.exe or start Wubi with the "--32bit" argument.
To modify arguments, right-click Wubi.exe and select "Create Shortcut". Then right-click the shortcut, select Properties, and modify the Target line, for example: "C:\Documents and Settings\<user>\Desktop\wubi.exe" --32bit
It would be interesting to know, if the apps have a speed gain in 64bit or not. If not, then we would have the same situation like with the LLR-app.
Speed gain yes, but negligible with 5 percent.
He can download 32 bit Ubuntu but he can't run the WSS project which requires 64 bit.
____________
Oh Bondage? Up Yours.
http://www.youtube.com/watch?v=ogypBUCb7DA
|
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
|
He can download 32 bit Ubuntu but he can't run the WSS project which requires 64 bit.
You are right. John wrotes "They are only available for 64bit OS's (Mac, Linux, Windows)."
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress has gone well today. We have now entered the void between between 4.15e15 and 5.13e15 for Wieferich where the previous Dorais and Klyve search found nothing.
Completed Thru Leading Edge Wieferichs Near Wieferichs
40099e11 41903e11 0 26 Please spread the word about "A week of Wieferich and Wall-Sun-Sun". We could come close to reaching 5.5e15 in Wieferich by the end of the week. :)
[EDIT] Here are the "near" Wieferich finds so far ranked by "nearness":
Finds Finder Team Date
3723113065138349 (-1 +18 p) Tarmo_Ilves Aggie_The_Pew 2011-12-31 15:31:52 UTC
4150209531584437 (-1 -24 p) [DPC]Pyrus Dutch_Power_Cows 2012-01-01 23:21:06 UTC
4032459967159163 (-1 +86 p) Buckeye74 SOTKE 2012-01-01 16:53:30 UTC
3411159925463651 (-1 -88 p) Tarmo_Ilves Aggie_The_Pew 2011-12-30 11:44:45 UTC
3948546628939699 (-1 +93 p) Lumiukko PrimeSearchTeam 2012-01-01 10:18:35 UTC
4143792274787999 (+1 +108 p) KD7LRJ Aggie_The_Pew 2012-01-01 23:20:13 UTC
3690728733648797 (-1 +167 p) ardo PrimeSearchTeam 2011-12-31 12:57:03 UTC
3596116634982419 (-1 +238 p) Tarmo_Ilves Aggie_The_Pew 2011-12-31 04:44:32 UTC
3522008578854793 (+1 +249 p) KD7LRJ Aggie_The_Pew 2011-12-30 22:09:21 UTC
3084683987742809 (+1 +378 p) Mark_Doom Aggie_The_Pew 2011-12-29 07:27:28 UTC
3508559613958337 (+1 -393 p) ardo PrimeSearchTeam 2011-12-30 21:16:53 UTC
4139674979655779 (-1 +500 p) Sideshow_Larry Turan@BOINC 2012-01-02 01:02:52 UTC
4095949437075647 (+1 +535 p) KD7LRJ Aggie_The_Pew 2012-01-01 20:01:58 UTC
4002406214080093 (-1 -547 p) KD7LRJ Aggie_The_Pew 2012-01-01 14:05:39 UTC
3281465623326827 (-1 +580 p) DoES Aggie_The_Pew 2011-12-30 00:04:37 UTC
3584947606208831 (+1 -593 p) KD7LRJ Aggie_The_Pew 2011-12-31 04:01:28 UTC
3808852067010139 (-1 -606 p) KD7LRJ Aggie_The_Pew 2011-12-31 22:17:50 UTC
4097229609590477 (-1 +666 p) Tom* Sicituradastra 2012-01-01 20:56:16 UTC
3783824393973131 (-1 -716 p) [SG-SPEG]BerndBrot SETI.Germany 2011-12-31 19:45:53 UTC
3615682761793517 (-1 -725 p) SysadmAtNbg SETI.Germany 2011-12-31 06:40:36 UTC
3831027317277109 (-1 -729 p) Tarmo_Ilves Aggie_The_Pew 2011-12-31 23:58:58 UTC
3956087188482973 (-1 -745 p) KD7LRJ Aggie_The_Pew 2012-01-01 10:47:32 UTC
3317543516011981 (-1 -750 p) DoES Aggie_The_Pew 2011-12-30 02:59:00 UTC
3173788761759379 (-1 +792 p) Tarmo_Ilves Aggie_The_Pew 2011-12-29 14:46:56 UTC
3113440825206487 (+1 +887 p) DoES Aggie_The_Pew 2011-12-29 10:14:59 UTC
3606350362143511 (+1 -985 p) DoES Aggie_The_Pew 2011-12-31 05:51:03 UTC
Here are the "near" Wall-Sun-Sun finds so far ranked by "nearness":
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
106411e10 108297e10 0 6
Finds Finder Team Date
989130429463493 (0 -185 p) DoES Aggie_The_Pew 2011-12-28 08:40:17 UTC
1009216602431657 (0 -205 p) Tarmo_Ilves Aggie_The_Pew 2011-12-28 16:24:38 UTC
1047632207771407 (0 +541 p) SysadmAtNbg SETI.Germany 2011-12-30 10:42:05 UTC
1001953024782971 (0 +633 p) [SG-SPEG]BerndBrot SETI.Germany 2011-12-28 13:20:35 UTC
1070876995179289 (0 -689 p) DoES Aggie_The_Pew 2012-01-01 19:30:01 UTC
976925681690507 (0 +815 p) BarczykW GPUForce 2011-12-28 03:13:04 UTC
____________
|
|
|
|
|
|
So I've read the description of the projects, but my brain is not even close to that of a mathematicians. Could anyone explain in plain English what on earth this near prime business means and why it's interesting?
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
|
|
|
|
|
|
Lol cows are big but brains?
____________
 |
|
|
|
|
So I've read the description of the projects, but my brain is not even close to that of a mathematicians. Could anyone explain in plain English what on earth this near prime business means and why it's interesting?
Maybe the wikipedia articles will help a little:
http://en.wikipedia.org/wiki/Wieferich_prime
http://en.wikipedia.org/wiki/Wall%E2%80%93Sun%E2%80%93Sun_prime
Although I'm sure there are equally good descriptions on other websites if you do some searching on google.
____________
 |
|
|
|
|
So I've read the description of the projects, but my brain is not even close to that of a mathematicians. Could anyone explain in plain English what on earth this near prime business means and why it's interesting?
Those near W... primes are being recorded for statistical purposes and allow for some kind of error checking. For example, it is possible to calculate the "expected" number of near Wieferich primes in a given interval, say eg. from 3000000000000000 to 7000000000000000. For that interval, one can expect to find circa 47 near Wieferich primes with |A| <= 1000. Thus one can compare the collected data (ie. the number of near Wieferich primes found) with the expected number and those two values should ideally be close together. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Progress has gone well today. We have now entered the void between between 4.15e15 and 5.13e15 for Wieferich where the previous Dorais and Klyve search found nothing.
Well, the search is not void of |A| <= 1000 finds between 4.15e15 and 5.13e15 for Wieferich:
Finds Finder Team Date
4268733356999057 (+1 +601 p) mackerel Aggie_The_Pew 2012-01-02 06:53:05 UTC
4389970931341169 (+1 -252 p) ardo PrimeSearchTeam 2012-01-02 14:12:31 UTC
The quest now is to find anything |A| <= 100.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
Progress really picked up the pace today. Thank you! If we are able to sustain this, 6.0e15 is well within reach. We could even make a run for 6.7e15 and reach first pass territory for Wieferich. ;)
Project stats are available here: Wieferich | Wall-Sun-Sun
For those increasing there cache, please monitor the Pending Work stats page for expiring tasks. Current deadline is 24 hours.
Other than the two mentions in the previous post, no other "near" Wieferich primes have been found. This barren expanse so far is proving to be true. The good news is that we are already half way through it.
Completed Thru Leading Edge Wieferichs Near Wieferichs
42494e11 46995e11 0 28
Wall-Sun-Sun continues to make good progress as well. Another "near" Wall-Sun-Sun |A| <= 1000 was discovered today.
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
109892e10 113212e10 0 7
Finds Finder Team Date
1096692588113113 (0 -138 p) Tarmo_Ilves Aggie_The_Pew 2012-01-02 10:30:31 UTC
Please spread the word about "A week of Wieferich and Wall-Sun-Sun". The home warming party ends 7 January 2012.
____________
|
|
|
DoESVolunteer tester
Send message
Joined: 11 Oct 08
Posts: 784
ID: 30382
Credit: 74,878,735
RAC: 0
|
|
xp x64 still reporting a bit odd????
67457593 (0 -27 p) --------------Aggie_The_Pew-- Little_Gekko------2012-01-03 02:59:26 UTC
989130429463493 (0 -185 p)-- Aggie_The_Pew-- Little_Gekko------ 2011-12-28 08:40:17 UTC
____________
Member of AtP
Shown here is an Australian native rat (Ratus Kickarsus) |
|
|
|
|
I wish I could run it but I can't run the new wwww on my computer which is 32-bit.
You can't use 64-bit Wubi: http://www.ubuntu.com/download/ubuntu/windows-installer ?
Unless he has a new computer, all the computers he has listed under BOINC have 32 bit CPUs. WUBI (or any other dual-boot or VM) is only helpful if you're running a 32-bit OS on a 64 bit CPU. If you're running a 32 bit CPU, you can't run the code at all.
Wabbit98 may be able to "participate" with a 32-bit CPU using qemu-system-x86_64. Here is a claim of someone running 64-bit Debian Linux on a 32-bit CPU. http://lists.nongnu.org/archive/html/qemu-discuss/2011-11/msg00001.html
Might be fun to try, but not recommended for speed or energy efficiency. You can't make a silk purse out of a rat's ear. (No offence to rodents. Heh, for entertainment I began this crunch with teamid=Aggie_The_Pew but cleared that entry before any work units were completed.)
Actually, I take that back. If this really works, I could put an old AGP ATI 3650 to work in a Pentium III system. Currently the CAL driver fails because it demands a 64-bit CPU. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 237,712,514
RAC: 2,203
|
|
Hmmmm.
So you're going to run a CPU emulator (that's gunna be really fast, huh?) on an almost 10 year old 32 bit CPU so you can run 64 bit code for a project with 24 hour deadlines?
Ok.
Lemme know how that works out for ya. ;-)
The idea definitely gets a +1 for ingenuity!
/serious:on
Why does this software need 64 bits? Most prime finding programs use floating point math because the heavy lifting is done by Fourier transforms, so the size of the integer math hardware doesn't matter. Or at least doesn't matter much.
Sieves work with much smaller numbers, and do their work with integer math, so they benefit from having 64 bit ALUs. Is this software more like a sieve than your typical Proth-type program?
I'm curious.
____________
My lucky number is 75898524288+1 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
xp x64 still reporting a bit odd????
67457593 (0 -27 p) --------------Aggie_The_Pew-- Little_Gekko------2012-01-03 02:59:26 UTC
989130429463493 (0 -185 p)-- Aggie_The_Pew-- Little_Gekko------ 2011-12-28 08:40:17 UTC
That's because I haven't fixed it to work on 64-bit WinXP. |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
Why does this software need 64 bits? Most prime finding programs use floating point math because the heavy lifting is done by Fourier transforms, so the size of the integer math hardware doesn't matter. Or at least doesn't matter much.
Sieves work with much smaller numbers, and do their work with integer math, so they benefit from having 64 bit ALUs. Is this software more like a sieve than your typical Proth-type program?
It requires 64 bits because there are some hand-written assembler modules that require 64-bit operations, which can only be done by a 64-bit OS. It could be modified to work on a 32-bit OS, but that would be a significant amount of work, i.e. a lot more asm code, and it would be much slower than the 64-bit version.
The math involved is more like a sieve than an FFT (used in prime finding). Given a range, the software extracts all prime numbers from that range, then performs a Wieferich or Wall-Sun-Sun test on each prime.
Although the FPU was an option, previous projects stopped because FPU rounding errors became an issue. I'm fairly certain that is why the previous Wall-Sun-Sun search stopped around 1e15. Since wwww does not rely on the FPU, it doesn't have any rounding issues and will have no issues up to 2^62, where it could experience overflows. If that happens, a completely new algorithm will be required.
P.S. My last comment is a distinct possibility. I'm working on an OpenCL flavor of wwww. I expect it to be anywhere from 4x to 30x faster than the CPU-only version AND it should run on 32-bit OSes. With such a speed increase and more participation 1e18 should be a reasonable target for the project, whereas 1e17 is a reasonable target today. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 237,712,514
RAC: 2,203
|
|
Thanks, Mark. That all makes sense.
____________
My lucky number is 75898524288+1 |
|
|
|
|
Why does this software need 64 bits? Most prime finding programs use floating point math because the heavy lifting is done by Fourier transforms, so the size of the integer math hardware doesn't matter. Or at least doesn't matter much.
Sieves work with much smaller numbers, and do their work with integer math, so they benefit from having 64 bit ALUs. Is this software more like a sieve than your typical Proth-type program?
It requires 64 bits because there are some hand-written assembler modules that require 64-bit operations, which can only be done by a 64-bit OS. It could be modified to work on a 32-bit OS, but that would be a significant amount of work, i.e. a lot more asm code, and it would be much slower than the 64-bit version.
The math involved is more like a sieve than an FFT (used in prime finding). Given a range, the software extracts all prime numbers from that range, then performs a Wieferich or Wall-Sun-Sun test on each prime.
Although the FPU was an option, previous projects stopped because FPU rounding errors became an issue. I'm fairly certain that is why the previous Wall-Sun-Sun search stopped around 1e15. Since wwww does not rely on the FPU, it doesn't have any rounding issues and will have no issues up to 2^62, where it could experience overflows. If that happens, a completely new algorithm will be required.
P.S. My last comment is a distinct possibility. I'm working on an OpenCL flavor of wwww. I expect it to be anywhere from 4x to 30x faster than the CPU-only version AND it should run on 32-bit OSes. With such a speed increase and more participation 1e18 should be a reasonable target for the project, whereas 1e17 is a reasonable target today.
If I may I ask, does the PG Wieferich project render Wieferich at elmath.org obsolete? |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
If I may I ask, does the PG Wieferich project render Wieferich at elmath.org obsolete?
This search is purely a search for the next Wieferich and any near-Wieferich's it finds along the way. If that is all that they are doing, then I see little benefit to both projects co-existing. So one might ask the question of why PrimeGrid is taking this on. there are a number of reasons, including;
1) Faster client software. If there software was anywhere near as fast as wwww, we wouldn't be having this discussion.
2) Even faster client software on the way with an OpenCL implementation.
3) More uniform control over stats and workunits. There's look prettier, but it is difficult to make sense out of them.
4) Open source. As far as I know they have never released a line of code. All of my code (both PRPNet and wwww) are freely available under GPL.
5) More interest. PrimeGrid users have shown a lot of interest in the search. The size of PrimeGrid alone will drive heavy participation in the project. From what I can tell, interest in their project has waned considerably. |
|
|
|
|
If I may I ask, does the PG Wieferich project render Wieferich at elmath.org obsolete?
At least the PG search seems to be a lot more efficient (I am still amazed by the progress the PG search has made since its start). The Wieferich@Home client seems to be much slower than the PRP client and I am also still sceptical of the 'periodic test' algorithm W@H is using, because that algorithm is based on a yet to be proven conjecture, ie the observation that the two known Wieferich primes minus one (1092 and 3510) have a repetitive appearance when written as binary numbers (1092 is 010001000100 in binary and 3510 is 110110110110 in binary). However, it is possible that Wieferich primes without this property exist, which their periodic test could not detect (after all, it is even possible that this phenomenon is just a curious coincidence only holding for 1093 and 3511 but not for any other Wieferich prime). Given the probable relative scarcity of Wieferich primes and the uncertaincy regarding the validity of the conjecture, I am sceptical whether their periodic test is useful at all. It might be just a waste of computing resources. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
If I may I ask, does the PG Wieferich project render Wieferich at elmath.org obsolete?
This search is purely a search for the next Wieferich and any near-Wieferich's it finds along the way. If that is all that they are doing, then I see little benefit to both projects co-existing. So one might ask the question of why PrimeGrid is taking this on. there are a number of reasons, including;
1) Faster client software. If there software was anywhere near as fast as wwww, we wouldn't be having this discussion.
2) Even faster client software on the way with an OpenCL implementation.
3) More uniform control over stats and workunits. There's look prettier, but it is difficult to make sense out of them.
4) Open source. As far as I know they have never released a line of code. All of my code (both PRPNet and wwww) are freely available under GPL.
5) More interest. PrimeGrid users have shown a lot of interest in the search. The size of PrimeGrid alone will drive heavy participation in the project. From what I can tell, interest in their project has waned considerably.
Back in March 2009, PrimeGrid reviewed this search after a flurry of requests by PG users. There was concern that the application being used was inefficient and people interested in a Wieferich effort wanted a place to put their resources that used more efficient code. At the time, PrimeGrid had limited resources and were pursuing other prime searches.
Again in March 2010 there were more requests that PrimeGrid investigate taking up the search. As you can see from this post, PrimeGrid continued to have limited resources and were pursuing other prime searches.
Jump to summer 2011, there were more requests. At that time, we had the resources and could add more projects but the timing was off. Finally, in December 2011, through the endless efforts of Mark and Lennart, PrimeGrid was able to implement the Wieferich and Wall-Sun-Sun prime searches in PrimeGrid's PRPNet using a more efficient wwww search application.
This has been a slow process for PrimeGrid to eventually adopt these projects. Please note that PrimeGrid will coordinate with Wieferich@Home so as not to impede upon their active ranges. Emails were sent last month so that we could establish search ranges (no response yet).
Currently, we are doing a double check effort (new effort for "near" Wieferichs at |A| <= 1000) up to 6.7e15. There's no documented completion of work beyond that limit. Wieferich@Home is searching 9e15+ range. We are presuming that all has been checked up to that point. Regardless if it was documented or not, we'll do a complete search up to 9e15. Hopefully we'll be able to coordinate beyond that.
At the current rate, PrimeGrid's Wieferich prime search could reach 6.7e15 by the end of the week and 9e15 by the end of next week. However, we don't expect to sustain this rate after this first week is complete...although it sure would be nice. :)
If PrimeGrid is able to achieve 9e15 by the end of next week, that will mean 6e15 amount of work was completed in 2 1/2 weeks. That's a nice demonstration on how efficient Mark's (and Geoff's) code is.
If you wish to join the effort, please see the Welcome to a week of Wieferich and Wall-Sun-Sun post at the beginning of this thread.
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
Jump to summer 2011, there were more requests. At that time, we had the resources and could add more projects but the timing was off. Finally, in December 2011, through the endless efforts of Mark and Lennart, PrimeGrid was able to implement the Wieferich and Wall-Sun-Sun prime searches in PrimeGrid's PRPNet using a more efficient wwww search application.
I had been sitting on the (Wieferich) code for a few years. It originated as a 64-bit PPC program. Geoff wrote the x86-asm routine that allowed it to be ported to Linux. Since Geoff is incognito, I had to make further modifications to get it work on Win64 and add some performance enhancements. The Wall-Sun-Sun code was originally FPU only and was designed to stop working at about 2^50 (1e15). I tried numerous things to address FPU rounding errors (such as using long double), but going to integer only was the only way to get around it. I had to write a single asm routine to eke out the best performance. That asm routine is x86 only. Since nobody is clamoring for a PPC version (and my PPC is pretty much dead), I have had no reason to make it work on that platform.
My thoughts for wwww originated earlier this year and I was hesitant to integrate into PRPNet due to the changes needed to PRPNet to support the projects and because I didn't want to poach from wieferich@home. I know that they don't have ownership of the search, yet I have been hesitant WRT competing with them. I offered my code to them years ago, but since my code is GPL, they refused to use it. I raised a stir over there at the time (because I said their code was slow) and a number of people left their project. Part of me is surprised that the project hasn't died on its own. This is the way I see it:
wieferich@home --> model t
wwww --> porsche
wwwwCL --> rocket
If PrimeGrid is able to achieve 9e15 by the end of next week, that will mean 6e15 amount of work was completed in 2 1/2 weeks. That's a nice demonstration on how efficient Mark's (and Geoff's) code is.
I'm very happy with the progress. I hope to see the Wall-Sun-Sun project get as much love as Wieferich. |
|
|
|
|
|
rogue said... my code is GPL
That's wonderful! How can we receive copies of your source code? |
|
|
|
|
|
Are there any plans for a CUDA implementation?
The math involved is more like a sieve than an FFT (used in prime finding). Given a range, the software extracts all prime numbers from that range, then performs a Wieferich or Wall-Sun-Sun test on each prime.
Does FFT refer to Fast Fourier Transform, or is it an initialism for something else?
____________
 |
|
|
|
|
|
Thanks for the answers Toshio Yamaguchi, rogue and John. It is not like I prefer one project over the other. But I am concerned if one project becomes obsolete or useless because another one does it lots more efficiently. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 237,712,514
RAC: 2,203
|
Are there any plans for a CUDA implementation?]
Mark will probably answer this for himself, but it looks like the answer is "yes", if what you meant was "GPU" rather than "CUDA":
2) Even faster client software on the way with an OpenCL implementation.
OpenCL runs on both Nvidia and ATI/AMD GPUs.
Does FFT refer to Fast Fourier Transform, or is it an initialism for something else?
Fast Fourier Transform.
The bonus question, of course, is why do we use FFTs to hunt for prime numbers? The answer is the numbers we use are so large that it's actually faster to use a forward FFT, do some math on the transformed number, and do an inverse FFT than it is to do the math directly on the original number.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
I didn't know OpenCL ran on Nvidia. Learn something new every day.
I figured that if it did stand for fast fourier then something long those lines was going on.
____________
|
|
|
|
|
rogue said... my code is GPL
That's wonderful! How can we receive copies of your source code?
(update)
OK, I found most of it at your home page. All C++. A bit out of date, no wwww, and no GPL notice. That's understandable given your time constraints.
Is the generated asm code for uint64 on some platforms really so bad that it must be hand-written? I have an integer-based crunching project in plain C whose speed approximately doubles with judicious use of static variables and maxing the optimization in GCC 4.6.1 with -Ofast -march=corei7 (not unsafe because not using IEEE math). On the other hand, speed improved a little when I didn't inline a hot subroutine, showing that GCC doesn't guess well about the runtime profile for register allocation.
I guess a Wieferich work unit would take up to 19 hours compiled for i386 and run on 866 MHz Pentium 3. These old things should be retired because they consume at least 25 times the energy to compute the same result as a modern 64-bit CPU. |
|
|
|
|
P.S. My last comment is a distinct possibility. I'm working on an OpenCL flavor of wwww. I expect it to be anywhere from 4x to 30x faster than the CPU-only version AND it should run on 32-bit OSes. With such a speed increase and more participation 1e18 should be a reasonable target for the project, whereas 1e17 is a reasonable target today.
Can you please tell when approximately the OpenCL application wiil be available?
Cheers
|
|
|
|
|
If PrimeGrid is able to achieve 9e15 by the end of next week, that will mean 6e15 amount of work was completed in 2 1/2 weeks. That's a nice demonstration on how efficient Mark's (and Geoff's) code is.
how long it takes to make a double-check from the 1 to 6e15 with this amazing speed?
____________
wbr, Me. Dead J. Dona
 |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 2,436
                          
|
how long it takes to make a double-check from the 1 to 6e15 with this amazing speed?
Couple of weeks, that's without GPU version. Considering Wieferich is running at least from 2008 (using it's slow app)...
It sounds applealing to go from scratch using GPUs :-)
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
Can you please tell when approximately the OpenCL application wiil be available?
I can't say right now. I have most of it written and working, but there is much to be done yet before I can release the code. |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
rogue said... my code is GPL
That's wonderful! How can we receive copies of your source code?
http://home.roadrunner.com/~mrodenkirch/wwww_1.3.zip |
|
|
valtercVolunteer tester
Send message
Joined: 30 May 07
Posts: 119
ID: 8810
Credit: 5,673,154,322
RAC: 0
            
|
|
tried wwww on CentOS
./wwww
./wwww: /lib64/libc.so.6: version `GLIBC_2.7' not found (required by ./wwww)
that is a know problem with CentOS releases
is it possible to build a statically linked executable?
thank you all
... or a makefile for the source code.... (as and gcc opt switches...) |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
OK, I found most of it at your home page. All C++. A bit out of date, no wwww, and no GPL notice. That's understandable given your time constraints.
Is the generated asm code for uint64 on some platforms really so bad that it must be hand-written? I have an integer-based crunching project in plain C whose speed approximately doubles with judicious use of static variables and maxing the optimization in GCC 4.6.1 with -Ofast -march=corei7 (not unsafe because not using IEEE math). On the other hand, speed improved a little when I didn't inline a hot subroutine, showing that GCC doesn't guess well about the runtime profile for register allocation.
I guess a Wieferich work unit would take up to 19 hours compiled for i386 and run on 866 MHz Pentium 3. These old things should be retired because they consume at least 25 times the energy to compute the same result as a modern 64-bit CPU.
I have been busy and haven't kept my site up to date. Technically PRPNet isn't GPL right now. I just need to add the GPL header to the sources.
As for hand-written ASM, the math involved requires the multiplication of two 64-bit values to create a 128-bit value, which is then multiplied by another 64-bit value to create a 196-bit value. Technically, this could be done in plain C, by using gcc's int128_t datatype. I didn't become aware of the int128_t datatype until after the asm was written. The asm is probably twice as fast as plain C would be. |
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
|
tried wwww on CentOS
./wwww
./wwww: /lib64/libc.so.6: version `GLIBC_2.7' not found (required by ./wwww)
that is a know problem with CentOS releases
is it possible to build a statically linked executable?
thank you all
... or a makefile for the source code.... (as and gcc opt switches...)
Maybe you can use my makefile but you have to check the paths:
NVIDIA_SDK = $(HOME)/NVIDIA_GPU_Computing_SDK
NVIDIA_INC = -I$(NVIDIA_SDK)/C/common/inc
NV = $(CUDA_PATH)/bin/nvcc
# NV = $(CUDA_PATH)/open64/bin/nvopencc
CC = g++
CUDA_PATH = /usr/local/cuda
CUDA_LIB += $(CUDA_PATH)/lib64
#CUDA_LIB += -L$(CUDA_PATH)/lib
BOINC_PATH = /usr/local/boinc
BOINC_LIB += $(BOINC_PATH)/lib
BOINC_API += $(BOINC_PATH)/api
# -I$(BOINC_PATH) -I$(BOINC_LIB) -I$(BOINC_API) -I$(CUDA_PATH)/include
# -lboinc_api -lboinc
#
INCL += -I/usr/local/include
INCL2 += -I$(BOINC_PATH)
INCL2 += -I$(BOINC_PATH)/lib
INCL2 += -I$(BOINC_PATH)/api
INCL2 += -I$(BOINC_PATH)/include
#USE_BOINC = -DUSE_BOINC
#USE_BOINC =
ARCH += -gencode arch=compute_13,code=sm_13
ARCH += -gencode arch=compute_20,code=sm_20
ARCH += -gencode arch=compute_20,code=sm_21
#ARCH = -arch=sm_21
NVCCFLAGS += --ptxas-options=-v
NVCCFLAGS += -DMERS_PACKAGE -DBIT_SIEVE -DTESTING_SMALL_EXPONENTS -DSIEVE_SIZE_IN_BYTES=32 -DNUM_SMALL_PRIMES=32768 -DDO_NOT_USE_LONG_DOUBLE
# NVCCFLAGS += -use_fast_math -maxrregcount=32
TUNE = -mtune=core2 -DX86_64 -DGPUONLY
TUNE2 = -m64
# DEBUG = -g -G
DEBUG =
# DEBUG2 = -Wall -Wextra -g -ggdb3
all: wwww_GPU wwww_CPU
gpu: wwww_GPU
cpu: wwww_CPU
# --with-pic libstdc++.a -lm
# --with-pic /usr/lib/gcc/x86_64-linux-gnu/4.3/libstdc++.a
#LIB = -fPIC -s -lstdc++
LIB = -fPIC -s
# LIB = -fPIC -DSHARED
wwww_GPU: sieve_GPU.o wallsunsun_GPU.o wieferich_GPU.o do_boinc_init_GPU.o
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) -o wwww_GPU sieve_GPU.o wallsunsun_GPU.o wieferich_GPU.o do_boinc_init_GPU.o -L$(CUDA_LIB) -lcudart -lcufft -lm -L$(BOINC_API)/libboinc_api.a -L$(BOINC_LIB)/libboinc.a
sieve_GPU.o:
$(NV) $(TUNE2) -O3 $(USE_BOINC) $(DEBUG) $(INCL) $(NVIDIA_INC) -I. sieve.cu $(NVCCFLAGS) $(ARCH) -c -o $@
wallsunsun_GPU.o:
$(NV) $(TUNE2) -O3 $(USE_BOINC) $(DEBUG) $(INCL) $(NVIDIA_INC) -I. wallsunsun.cu $(NVCCFLAGS) $(ARCH) -c -o $@
wieferich_GPU.o:
$(NV) $(TUNE2) -O3 $(USE_BOINC) $(DEBUG) $(INCL) $(NVIDIA_INC) -I. wieferich.cu $(NVCCFLAGS) $(ARCH) -c -o $@
#boinc_interface64.o:
# $(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL2) -I. boinc_interface.cpp -o $@
do_boinc_init_GPU.o:
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL2) -I. do_boinc_init.cpp -c -o $@
wwww_CPU: sieve_CPU.o wallsunsun_CPU.o wieferich_CPU.o do_boinc_init_CPU.o
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(DEBUG2) -o wwww_CPU sieve_CPU.o wallsunsun_CPU.o wieferich_CPU.o do_boinc_init_CPU.o -lm -L$(BOINC_API)/libboinc_api.a -L$(BOINC_LIB)/libboinc.a
sieve_CPU.o:
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL) -I. sieve.c -c -o $@
wallsunsun_CPU.o:
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL) -I. wallsunsun.c -c -o $@
wieferich_CPU.o:
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL) -I. wieferich.c -c -o $@
do_boinc_init_CPU.o:
$(CC) $(LIB) $(TUNE) -O3 $(DEBUG2) $(USE_BOINC) $(INCL2) -I. do_boinc_init.cpp -c -o $@
clean:
-rm *.o wwww_GPU wwww_CPU
clean_gpu:
-rm *_GPU.o wwww_GPU
clean_cpu:
-rm *_CPU.o wwww_CPU
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
tried wwww on CentOS
./wwww
./wwww: /lib64/libc.so.6: version `GLIBC_2.7' not found (required by ./wwww)
that is a know problem with CentOS releases
is it possible to build a statically linked executable?
thank you all
... or a makefile for the source code.... (as and gcc opt switches...)
Simple make:
gcc *.c *nix*.s -O3 -m64 -o wwww |
|
|
valtercVolunteer tester
Send message
Joined: 30 May 07
Posts: 119
ID: 8810
Credit: 5,673,154,322
RAC: 0
|
tried wwww on CentOS
./wwww
./wwww: /lib64/libc.so.6: version `GLIBC_2.7' not found (required by ./wwww)
that is a know problem with CentOS releases
is it possible to build a statically linked executable?
thank you all
... or a makefile for the source code.... (as and gcc opt switches...)
Simple make:
gcc *.c *nix*.s -O3 -m64 -o wwww
just need to add -lm
thank you. |
|
|
|
|
Wieferich@Home is searching 9e15+ range. We are presuming that all has been checked up to that point.
I don't think so. From this Wayback Machine snapshot from 2008 it seems they only did very limited work above 1.17e15, 1.27e15, 1.37e15 and 6.17e15, with the performed work in each range being around 1e12. After that they seem to have jumped around in the 9e15 range, where they still remain. Thus I believe the assumption that everything up to 9e15 has been checked by them to be incorrect. Also given the speed of their client and the progress they have made since I first followed their project (since end of 2009) is not very encouraging (they are still hanging around in the 9e15 range and this for probably 2-3 years). Not very promising in my opinion. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Wieferich@Home is searching 9e15+ range. We are presuming that all has been checked up to that point.
I don't think so. From this Wayback Machine snapshot from 2008 it seems they only did very limited work above 1.17e15, 1.27e15, 1.37e15 and 6.17e15, with the performed work in each range being around 1e12. After that they seem to have jumped around in the 9e15 range, where they still remain. Thus I believe the assumption that everything up to 9e15 has been checked by them to be incorrect.
Thanks for the research, Toshio. This just means that PrimeGrid's search through the 6.7e15 to 9e15 range will be a first pass. Maybe the next Wieferich prime is within there? ;) At the very least, I'm sure we'll find additional information on "near" Wieferich primes.
____________
|
|
|
|
|
At least the PG search seems to be a lot more efficient (I am still amazed by the progress the PG search has made since its start). The Wieferich@Home client seems to be much slower than the PRP client and I am also still sceptical of the 'periodic test' algorithm W@H is using, because that algorithm is based on a yet to be proven conjecture, ie the observation that the two known Wieferich primes minus one (1092 and 3510) have a repetitive appearance when written as binary numbers (1092 is 010001000100 in binary and 3510 is 110110110110 in binary). However, it is possible that Wieferich primes without this property exist, which their periodic test could not detect (after all, it is even possible that this phenomenon is just a curious coincidence only holding for 1093 and 3511 but not for any other Wieferich prime). Given the probable relative scarcity of Wieferich primes and the uncertaincy regarding the validity of the conjecture, I am sceptical whether their periodic test is useful at all. It might be just a waste of computing resources.
I'd have put it a tad more strongly.
It is the wrong way round isn't it? Or maybe the scientific method has moved on since I was a lad...
The hypothesis is that Wieferich primes are periodic in binary. The test is not to see if more periodic binary numbers are Wieferich primes but to see if more Wieferich primes are periodic binary numbers since finding a Wieferich prime which is not a periodic binary number would dispose of the hypothesis.
____________
Oh Bondage? Up Yours.
http://www.youtube.com/watch?v=ogypBUCb7DA
|
|
|
|
|
At least the PG search seems to be a lot more efficient (I am still amazed by the progress the PG search has made since its start). The Wieferich@Home client seems to be much slower than the PRP client and I am also still sceptical of the 'periodic test' algorithm W@H is using, because that algorithm is based on a yet to be proven conjecture, ie the observation that the two known Wieferich primes minus one (1092 and 3510) have a repetitive appearance when written as binary numbers (1092 is 010001000100 in binary and 3510 is 110110110110 in binary). However, it is possible that Wieferich primes without this property exist, which their periodic test could not detect (after all, it is even possible that this phenomenon is just a curious coincidence only holding for 1093 and 3511 but not for any other Wieferich prime). Given the probable relative scarcity of Wieferich primes and the uncertaincy regarding the validity of the conjecture, I am sceptical whether their periodic test is useful at all. It might be just a waste of computing resources.
I'd have put it a tad more strongly.
It is the wrong way round isn't it? Or maybe the scientific method has moved on since I was a lad...
The hypothesis is that Wieferich primes are periodic in binary. The test is not to see if more periodic binary numbers are Wieferich primes but to see if more Wieferich primes are periodic binary numbers since finding a Wieferich prime which is not a periodic binary number would dispose of the hypothesis.
Their periodic test algorithm cannot find Wieferich primes not having a periodic binary representation. In early 2010, a preprint of their paper "Search for Wieferich primes through the use of periodic binary strings", Serdica J. Comp. Vol. 4 No. 3 was available for download from their project page as a PDF file and I still have a copy of that preprint on my computer. From this paper it is clear that their periodic test algorithm in fact tests whether more binary periodic numbers plus one are Wieferich. |
|
|
|
|
Their periodic test algorithm cannot find Wieferich primes not having a periodic binary representation. In early 2010, a preprint of their paper "Search for Wieferich primes through the use of periodic binary strings", Serdica J. Comp. Vol. 4 No. 3 was available for download from their project page as a PDF file and I still have a copy of that preprint on my computer. From this paper it is clear that their periodic test algorithm in fact tests whether more binary periodic numbers plus one are Wieferich.
Alas the paper is no longer available from their website. (I asked you in a PM if you could email me a copy.) I am not sure that my maths skills are strong enough to appreciate it anyway. It does sound to me that the underlying concept may be lacking some intellectual rigour.
A very reasonable response to that would be to point out that my maths education stopped with maths for scientists and some quantum chemistry as an undergraduate in the 70s and that perhaps it'd be better if I confined my opinions to organic chemistry, chemical engineering, law and rats!
____________
Oh Bondage? Up Yours.
http://www.youtube.com/watch?v=ogypBUCb7DA
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
WOW!!! The past two days have seen sustained progress. 6.7e15 is well within reach over the next 2 days. Tremendous effort everyone. Thank you to all involved.
Project stats are available here: Wieferich | Wall-Sun-Sun
The barren expanse from 4.15e15 and 5.13e15 for Wieferich is valid. There's no |A| <= 100 near Wieferich's although we did discover several |A| <= 1000 ones. Since the last update, here are the 19 additional finds sorted by nearness:
Completed Thru Leading Edge Wieferichs Near Wieferichs
53532e11 58578e11 0 47
Finds Finder Team Date
5131427559624857 (+1 -18 p) JAMC My_Way 2012-01-03 21:51:51 UTC
5539428831517831 (+1 +115 p) ardo PrimeSearchTeam 2012-01-04 15:43:53 UTC
5367369195612269 (-1 -159 p) ardo PrimeSearchTeam 2012-01-04 08:41:19 UTC
5683778474515027 (-1 -166 p) composite 2012-01-04 22:16:58 UTC
5464249230405811 (-1 -213 p) gomeyer US_Navy 2012-01-04 13:04:23 UTC
5755502459289463 (+1 +238 p) Arjant2 Dutch_Power_Cows 2012-01-05 00:27:57 UTC
4716670255419061 (-1 -247 p) KD7LRJ Aggie_The_Pew 2012-01-03 03:51:46 UTC
4686562925594831 (+1 -371 p) ardo PrimeSearchTeam 2012-01-03 02:23:25 UTC
5793958190595593 (+1 +425 p) ardo PrimeSearchTeam 2012-01-05 01:13:30 UTC
4965931158433309 (-1 +430 p) KD7LRJ Aggie_The_Pew 2012-01-03 14:38:43 UTC
5333734094863417 (+1 -467 p) Tom* Sicituradastra 2012-01-04 05:46:22 UTC
5541846220990307 (-1 +479 p) ardo PrimeSearchTeam 2012-01-04 15:45:49 UTC
5423711968552697 (+1 -488 p) samuel7 PrimeSearchTeam 2012-01-04 10:41:00 UTC
5258715693439951 (+1 -609 p) ardo PrimeSearchTeam 2012-01-04 04:02:30 UTC
5544680687900581 (-1 -871 p) KD7LRJ Aggie_The_Pew 2012-01-04 14:35:13 UTC
5078844401054113 (+1 +906 p) Lumiukko PrimeSearchTeam 2012-01-03 21:13:22 UTC
4901556577749379 (-1 -927 p) KD7LRJ Aggie_The_Pew 2012-01-03 12:04:26 UTC
5083735428486929 (+1 -939 p) KD7LRJ Aggie_The_Pew 2012-01-03 19:36:45 UTC
4690605285430003 (-1 +966 p) ardo PrimeSearchTeam 2012-01-03 02:32:38 UTC
Wall-Sun-Sun continues to make good progress as well. 3 more "near" Wall-Sun-Sun primes were found with 2 of them being |A| <= 100.
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
121851e10 123371e10 0 10
Finds Finder Team Date
1131092589761081 (0 -27 p) DoES Aggie_The_Pew 2012-01-03 02:59:26 UTC
1217727803528521 (0 -49 p) Sashixi 2012-01-04 20:03:58 UTC
1148677955963719 (0 +800 p) DoES Aggie_The_Pew 2012-01-03 11:21:46 UTC
Please spread the word about "A week of Wieferich and Wall-Sun-Sun". The home warming party ends 7 January 2012.
____________
|
|
|
|
|
how long it takes to make a double-check from the 1 to 6e15 with this amazing speed?
Couple of weeks, that's without GPU version. Considering Wieferich is running at least from 2008 (using it's slow app)...
It sounds applealing to go from scratch using GPUs :-)
I suppose it's a good idea to do second-pass from scratch. We can confirm some findings or find something new!
Will wait for admins to make a decision.
Just two more weeks...
____________
wbr, Me. Dead J. Dona
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
how long it takes to make a double-check from the 1 to 6e15 with this amazing speed?
Couple of weeks, that's without GPU version. Considering Wieferich is running at least from 2008 (using it's slow app)...
It sounds applealing to go from scratch using GPUs :-)
I suppose it's a good idea to do second-pass from scratch. We can confirm some findings or find something new!
p < 3e15 has been double checked at least once by a previous effort. 3e15 to 6.7e15 is currently being double checked by PrimeGrid's current effort. Therefore, I don't think it's necessary to go back and triple check. However, when testing the wwwwCL version, we'll definitely go back and manually check known ranges.
____________
|
|
|
|
|
3e15 to 6.7e15 is currently being double checked by PrimeGrid's current effort.
any differences with previous results?
____________
wbr, Me. Dead J. Dona
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
3e15 to 6.7e15 is currently being double checked by PrimeGrid's current effort.
any differences with previous results?
All confirmed and nothing new with |A| <= 100.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
We are knocking on the door of 6.7e15. What an amazing effort effort by all those involved. Thank you, thank you, thank you!!!
Project stats are available here: Wieferich | Wall-Sun-Sun
The last known |A| <= 100 near Wieferich has been confirmed. We will soon enter first pass ranges. Since the last update, here are the 4 additional finds sorted by nearness:
Completed Thru Leading Edge Wieferichs Near Wieferichs
57181e11 66394e11 0 51
Finds Finder Team Date
6517506365514181 (-1 -29 p) sm5ymt PrimeSearchTeam 2012-01-06 02:11:19 UTC
6218549260130329 (+1 +381 p) sm5ymt PrimeSearchTeam 2012-01-05 16:43:59 UTC
5994441921772859 (-1 +631 p) KD7LRJ Aggie_The_Pew 2012-01-05 09:15:39 UTC
5942481157969717 (-1 -864 p) sm5ymt PrimeSearchTeam 2012-01-05 07:43:08 UTC
Wall-Sun-Sun continues to make good progress as well. 3 more "near" Wall-Sun-Sun primes were found.
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
123359e10 129107e10 0 13
Finds Finder Team Date
1259881749210013 (0 +259 p) JAMC My_Way 2012-01-05 17:18:52 UTC
1248554712359503 (0 -576 p) Bas_Jansen 2012-01-05 11:32:30 UTC
1244785266374651 (0 +852 p) DoES Aggie_The_Pew 2012-01-05 10:15:00 UTC
The home warming party ends 7 January 2012.
____________
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 2,436
|
I don't think so. From this Wayback Machine snapshot from 2008 it seems they only did very limited work above 1.17e15, 1.27e15, 1.37e15 and 6.17e15, with the performed work in each range being around 1e12. After that they seem to have jumped around in the 9e15 range, where they still remain. Thus I believe the assumption that everything up to 9e15 has been checked by them to be incorrect. Also given the speed of their client and the progress they have made since I first followed their project (since end of 2009) is not very encouraging (they are still hanging around in the 9e15 range and this for probably 2-3 years). Not very promising in my opinion.
ElMath is a sitting duck. Project staff provides almost no feedback to contacs, apllication developement and efficiency is slow or nowhere.
I've provided recents information in czech boinc forum and that PG/PRPNet is approcing fast :-)
Feedback from boincers (that are/were active in Wieferich@home) is that it would be no harm that W@home will go off and PG is going large.
Let's roll and push forward!
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
|
weiferich server is 503
wss is 502
???
____________
wbr, Me. Dead J. Dona
|
|
|
|
|
|
I just upgraded to 5.0.3, I use to let it run on one server first to see that everything works :)
Lennart |
|
|
|
|
|
okay. is there a client 503?
almighty google
http://home.roadrunner.com/~mrodenkirch/prpnet_5.0.3.zip
there's my suggestion here!! OLOLO
____________
wbr, Me. Dead J. Dona
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Progress Update
We are knock on the door of 6.7e15.
We did more than just knock on the door...we smashed right through it. Well done everyone...and with a day to spare. :D
Leading edge is at 7e15.
Onwards and upwards!
____________
|
|
|
|
|
|
Is there already soms new about Boinc credit awarded for this?
Haven't seen any yet.
____________
 |
|
|
|
|
Is there already soms new about Boinc credit awarded for this?
Haven't seen any yet.
Oh yes, AtP has garnered hundred of thousands of credits already. Not sure why the cows are being left out :(
____________
@AggieThePew
 |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
As we start this last day of the home warming party, it's quite amazing to look back on what has been accomplished in just a week. By end of day, the leading edge will be approaching 8e15. On 1 January, the search was at ~3.7e15 so that's an astonishing 4.3e15 in 7 days. We could not have asked for a better start to the search...well, finding the next Wieferich prime would have been better, but we know how difficult that is.
Most of the work was double checking 3e15 to 6.7e15 while finding "classically" defined near Wieferich primes along the way. However, we are finally in first pass territory. If the current rate is sustained, we'll enter 1e16 by mid next week...far beyond our original expectations. Well done everyone.
Project stats are available here: Wieferich | Wall-Sun-Sun
No new |A| <= 100 near Wieferich primes have been found. Since the last update, here are 5 additional finds sorted by nearness:
Completed Thru Leading Edge Wieferichs Near Wieferichs
66684e11 73224e11 0 56
Finds Finder Team Date
7191640994464429 (-1 +406 p) Sideshow_Larry Turan@BOINC 2012-01-07 03:35:37 UTC
6941091684197179 (-1 +516 p) sm5ymt PrimeSearchTeam 2012-01-06 16:41:25 UTC
6811483557714481 (+1 +518 p) sm5ymt PrimeSearchTeam 2012-01-06 12:25:41 UTC
6657180771913949 (-1 +698 p) sm5ymt PrimeSearchTeam 2012-01-06 07:39:53 UTC
7284911154276809 (+1 +991 p) sm5ymt PrimeSearchTeam 2012-01-07 04:43:27 UTC
For this update, the Wall-Sun-Sun project upstages the Wieferich project by finding one |A| <= 10 and one |A| <= 100 near Wall-Sun-Sun prime. JAMC finds the first |A| <= 10 of both searches...congratulations. 5 more "near" Wall-Sun-Sun primes were found sorted by nearness:
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
130681e10 134533e10 0 18
Finds Finder Team Date
1326606479972297 (0 +9 p) JAMC My_Way 2012-01-06 21:45:03 UTC
1323740075583811 (0 -23 p) brinktastee Minnesota_Crunchers 2012-01-06 20:23:19 UTC
1341966257090429 (0 +197 p) MiHost AMD_Users 2012-01-07 04:37:35 UTC
1335604373209987 (0 +244 p) Aggie_The_Pew meilijo 2012-01-07 01:16:02 UTC
1287686692872079 (0 +464 p) samuel7 PrimeSearchTeam 2012-01-06 06:23:15 UTC
The home warming party over at the end of today.
____________
|
|
|
|
|
The home warming party over at the end of today.
should we stop crunching?
____________
wbr, Me. Dead J. Dona
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
The home warming party over at the end of today.
should we stop crunching?
Only if you want to. The welcome push is over, but Wieferich and Wall-Sun-Sun will continue to be available for anyone wishing to contribute. The goal, after all, is to find another Wieferich prime and the first Wall-Sun-Sun.
The next milestone is 1e16 for Wieferich and 15e14 for Wall-Sun-Sun. Hopefully that will happen in the next few days.
____________
|
|
|
|
|
|
I just got back from a week+ long vacation only to see that my computers sadly froze right after I left and failed to continue crunching on the two ports :-(
But I just got done reading on all the progress, and congrats everyone! Sounds like it has gone really well.
I'm home now and ready to see if I can't help push these projects just a bit further along with my crunching power. Doominator has been turned on full force. :-)
____________
 |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
The welcoming party is now over. Wieferich and Wall-Sun-Sun have been well received at PrimeGrid and are settling in for the long haul. Thank you to everyone involved. The first week of 2012 has been outstanding. Let's hope it sets the tone for the rest of the year. If so, then we should expect some significant progress this year.
In just one week's time, over 4e15 was completed in the Wieferich search and 3.2e14 was completed in the Wall-Sun-Sun search. Both levels of progress are amazing.
Project stats are available here: Wieferich | Wall-Sun-Sun
On this last day, here are the remaining finds:
Completed Thru Leading Edge Wieferichs Near Wieferichs
72323e11 79435e11 0 62
Finds Finder Team Date
7517031824422381 (-1 -163 p) samuel7 PrimeSearchTeam 2012-01-07 14:59:50 UTC
7653489434573371 (-1 -455 p) sm5ymt PrimeSearchTeam 2012-01-07 18:31:40 UTC
7795104211956571 (-1 -615 p) pabliedung USA 2012-01-07 23:43:48 UTC
7428722701584719 (+1 -689 p) xorrbit Aggie_The_Pew 2012-01-07 09:43:56 UTC
7399615392955711 (+1 +772 p) sm5ymt PrimeSearchTeam 2012-01-07 08:56:55 UTC
7589182338215419 (-1 +909 p) sm5ymt PrimeSearchTeam 2012-01-07 16:05:44 UTC
Completed Thru Leading Edge Wall-Sun-Suns Near Wall-Sun-Suns
136800e10 140073e10 0 19
Finds Finder Team Date
1347577483826921 (0 +611 p) Scott_Brown Duke_University 2012-01-07 06:49:20 UTC
Thanks again for all your support and computing power!!!
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
Both servers look like they will run out of work today. |
|
|
|
|
|
It seems I can no longer receive Wieferich WUs. |
|
|
|
|
|
And I'm also hitting a brick Wall-Sun-Sun. No units to be found there either.
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
|
|
|
|
|
It seems I can no longer receive Wieferich WUs.
...same here!
____________
http://stats.free-dc.org/pgridtag.php?id=51592&theme=2 |
|
|
|
|
|
WSS
Where's the beef?! |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
PRPNet Update 5.0.4
5.0.3 was short lived. A bug was introduced that caused a crash when a wwww near miss was found.
Please see this post to download 5.0.4.
____________
|
|
|
|
|
PRPNet Update 5.0.4
5.0.3 was short lived. A bug was introduced that caused a crash when a wwww near miss was found.
Please see this post to download 5.0.4.
John, will you resubmit all numbers checked with 503 version?
Now i understand why i found one client hanged at the morning.
____________
wbr, Me. Dead J. Dona
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
John, will you resubmit all numbers checked with 503 version?
Now i understand why i found one client hanged at the morning.
The server is fine, so nothing on the server is invalid. The client (besides the crash) was reporting all near misses to the server, so nothing was lost. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Progress Update
A +4 near Wall-Sun-Sun prime was found by user alan_buxey. Below is the Top 10 "nearest" Wall-Sun-Sun primes found at PrimeGrid:
User Team Client Prime Date
alan_buxey Loughborough_University big_mac 1570636036156699 (0 +4 p) Wednesday 11th of January 2012 02:00:55 PM
JAMC My_Way 5 1326606479972297 (0 +9 p) Friday 06th of January 2012 10:45:03 PM
brinktastee Minnesota_Crunchers asus-i7-950 1323740075583811 (0 -23 p) Friday 06th of January 2012 09:23:19 PM
DoES Aggie_The_Pew Little_Gekko 1131092589761081 (0 -27 p) Tuesday 03rd of January 2012 03:59:26 AM
Sashixi srv-sascha-lin 1217727803528521 (0 -49 p) Wednesday 04th of January 2012 09:03:58 PM
Mark_Doom Aggie_The_Pew AMD6-2 1511064984846883 (0 -101 p) Tuesday 10th of January 2012 02:23:21 AM
Tarmo_Ilves Aggie_The_Pew server3 1096692588113113 (0 -138 p) Monday 02nd of January 2012 11:30:31 AM
DoES Aggie_The_Pew Little_Gekko 989130429463493 (0 -185 p) Wednesday 28th of December 2011 09:40:17 AM
MiHost AMD_Users Farscape2 1341966257090429 (0 +197 p) Saturday 07th of January 2012 05:37:35 AM
Tarmo_Ilves Aggie_The_Pew massin 1009216602431657 (0 -205 p) Wednesday 28th of December 2011 05:24:38 PM
____________
|
|
|
|
|
|
A +3 Wall Sun-Sun was found!
JAMC / My_Way / Mint4 / 1955981616534559 (0 +3 p) / Friday 20th of January 2012 10:52:55 PM
We are getting closer to the first WSS...
____________
ESP: Eliminated k=94373
SR5: Eliminated k=97366 and k=325918 |
|
|
|
|
|
may we have some... ugh... like... graph?
____________
wbr, Me. Dead J. Dona
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
may we have some... ugh... like... graph?
no graph, but here are sortable lists:
Wall-Sun-Sun
Wieferich
Also, here are extended lists in pdf format:
Wall-Sun-Sun
Wieferich
____________
|
|
|
|
|
may we have some... ugh... like... graph?
no graph, but here are sortable lists:
Wall-Sun-Sun
Wieferich
Also, here are extended lists in pdf format:
Wall-Sun-Sun
Wieferich
Thanks John. That is some nice data. A few quick notes:
For the two Wieferich primes there is given
p +-1 A
1093 0 0
3511 0 0
Shouldn't it read
p +-1 A
1093 +1 0
3511 +1 0
Also I think the near Wieferich prime
2276306935816523 -1 - 3 p
is missing. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Shouldn't it read
p +-1 A
1093 +1 0
3511 +1 0
Also I think the near Wieferich prime
2276306935816523 -1 - 3 p
is missing.
I updated the +1's. However, 2276306935816523 -1 - 3 p is already listed. Are you referring to someplace else?
____________
|
|
|
|
|
Shouldn't it read
p +-1 A
1093 +1 0
3511 +1 0
Also I think the near Wieferich prime
2276306935816523 -1 - 3 p
is missing.
I updated the +1's. However, 2276306935816523 -1 - 3 p is already listed. Are you referring to someplace else?
Sorry, my fault. It is listed under p > 10 with |A| <= 10. Somehow I missed it. |
|
|
|
|
|
excel is complete waste
____________
wbr, Me. Dead J. Dona
|
|
|
