Primegrid

Post Reply
Message
Author
User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#41 Post by rebirther » 16.10.2010, 10:17

PPS Sieve für ATI-Karten, folgende Anleitung beachten:

app_info.xml file erstellen und in den Projektordner von Primegrid reinkopieren, Berechnungen auf CPU deaktivieren und GPU aktivieren.

Code: Select all

<app_info>
<app>
<name>pps_sr2sieve</name>
<user_friendly_name>Proth Prime Search &#40;Sieve&#41;</user_friendly_name>
</app>
<file_info>
<name>ppsieve-cl-boinc-x86-windows.exe</name>
<executable>
</file_info>
<app_version>
<app_name>pps_sr2sieve</app_name>
<version_num>124</version_num>
<plan_class>ati13ati</plan_class>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>-m 16</cmdline>
<file_ref>
<file_name>ppsieve-cl-boinc-x86-windows.exe</file_name>
<main_program>
</file_ref>
</app_version>
</app_info>
Die Zeit auf einer HD5850 bewegt sich um die 5min, leider ist die CUDA-app auf einer GTX460 viel schneller aber sicherlich wird daran noch gearbeitet, dennoch sind die Zeiten gegenüber einer CPU erstaunlich.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#42 Post by rebirther » 09.11.2010, 10:04

PPS Sieve geht in Pause...
Link

Durch die GPU App ist das Ziel schneller erreicht worden als man gedacht hatte.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#43 Post by rebirther » 20.11.2010, 20:40

Nun wurde auf TPSieve umgestellt, PPS Sieve bleibt gleich, hier muss man noch die app_info.xml Datei anpassen:

Code: Select all

<app_info>
<app>
<name>pps_sr2sieve</name>
<user_friendly_name>Proth Prime Search &#40;Sieve&#41;</user_friendly_name>
</app>
<file_info>
<name>tpsieve-cl-boinc-x86-windows.exe</name>
<executable>
</file_info>
<app_version>
<app_name>pps_sr2sieve</app_name>
<version_num>130</version_num>
<plan_class>ati13ati</plan_class>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>tpsieve-cl-boinc-x86-windows.exe</file_name>
<main_program>
</file_ref>
</app_version>
</app_info>
Apps Download

Werde ich demnächst noch testen, ob die funktioniert, die Zeiten werden sich aber ca. um den Faktor 4 erhöhen.

Update:
1,722.60s | 1,800.00cr. HD5850, GPU-Auslastung liegt bei 80%, da geht sicherlich noch was...

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#44 Post by rebirther » 24.11.2010, 17:15

The first prime in almost 2 1/2 years has been discovered for The Riesel Problem. More details to come after the primary finder has responded.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#45 Post by rebirther » 04.12.2010, 14:37

On 21 Nov 2010, 11:14:00 UTC, PrimeGrid’s The Riesel Problem project eliminated k=191249 by finding the Mega prime: 191249*2^3417696-1

The prime is 1,028,835 digits long and enters Chris Caldwell's The Largest Known Primes Database ranked 29th overall. This is the 2nd largest prime found in The Riesel Problem and the first k elimination in over 2 1/2 years. 63 k's now remain.

The discovery was made by Jonathan Pritchard of the United Kingdom using an AMD Phenom 9950 with 8GB RAM, running Windows 7. This computer took 7 hours and 31 minutes to complete the prime test using LLR. Jonathan is a member of the Turan@BOINC team.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#46 Post by rebirther » 05.12.2010, 19:25

Laut Rytis ist es jetzt möglich, ATI GPU Wus ohne app_info.xml file zu bekommen.

Link Forum

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#47 Post by rebirther » 20.12.2010, 17:07

Primorial prime found in PRPNet

Tis the season to be merry!!! :) With last week's Factorial prime and today's Primorial prime, this holiday is shaping up to be quite generous.

The prime is only the 19th known p#-1 form and the 1st discovered in 18 years. It is the 1st found of both forms (p#+1, p#-1) in over 9 years. Needless to say, this is exceptionally rare. It almost doubles the size of the previous record of 392113#+1 (169966 digits).

More details to come once the double check is complete.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#48 Post by rebirther » 28.12.2010, 14:56

SR5 PRPNet Challenge - second prime!

Murphy Babb of the United States has found the second prime of the SR5 PRPNet Challenge. And what a nice one it is...coming in ranked 4th for SR5 primes and 265th overall in Top 5000 primes database!

183916*5^519597-1

This prime marks a nice milestone for the SR5 project. Removing k=183916 brings the sieve file to less than 3M terms remaining (to n=2M). The current sieve file has 2,982,533 terms in 181 sequences.

Only one more prime is needed to make 2010 the best year of the SR5 project in regards to prime score. If you are interested in participating, please see the Last 10 Days of 2010 PRPNet Challenge post.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#49 Post by rebirther » 31.12.2010, 07:45

SR5 PRPNet Challenge - fourth and fifth prime!

The stars must have aligned today as two SR5 primes were found in the Challenge. One by Murphy Babb (his second) of the United States and one by Sascha Beat Dinkel of Switzerland. Murphy's prime (3938*5^558032-1) is 390,052 digits and ranks 3rd for SR5 primes and 218th overall. Sascha's prime (151026*5^559670-1) is 391,198 digits and ranks 2nd for SR5 primes and 216th overall.

This is a GREAT way to lead into the final 24 hours of the Challenge. Looks like we've even hit the 5 primes request. What more is there to ask for now??? :D How about an SR5 prime for everyone participating? Is that too lofty of a goal?

SR5 PRPNet Challenge - third prime!

Matthew Poulter of the United Kingdom has found the third prime of the SR5 PRPNet Challenge. It is 385,673 digits and ranks 2nd for SR5 primes and 224th overall in the Top 5000 primes database!

105782*5^551766-1

Now that 2010 has become the best year of the project, let's see how far we can push it. With just over 2 days left in the Challenge, we have an opportunity to find a couple more primes and make 2010 head and shoulders above the rest.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#50 Post by rebirther » 05.01.2011, 17:53

CWPSieve für CUDA im Test:

Link

OpenCL für ATI kommt auch noch!

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

GPWSieve CUDA

#51 Post by rebirther » 15.01.2011, 16:32

Für GPWSieve CUDA

In den Primegrid-Ordner kopieren:

Benötigte Files

app_info.xml file für Windows:

Code: Select all

<app_info>
<app>
<name>gcwsieve</name>
<user_friendly_name>Cullen/Woodall Prime Search &#40;Sieve&#41;</user_friendly_name>
</app>
<file_info>
<name>gcwsieve_1250180098.sieveinput_in</name>
<status>1</status>
<sticky>
</file_info>
<file_info>
<name>gcwsieve_20110101.sieveinput_in</name>
<status>1</status>
<sticky>
</file_info>
<file_info>
<name>cwpsieve-cuda-boinc-x86-windows.exe</name>
<executable>
</file_info>
<file_info>
<name>cudart.dll</name>
</file_info>
<app_version>
<app_name>gcwsieve</app_name>
<version_num>101</version_num>
<file_ref>
<file_name>cwpsieve-cuda-boinc-x86-windows.exe</file_name>
<main_program>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
<open_name>cudart.dll</open_name>
</file_ref>
<plan_class>cuda23</plan_class>
<avg_ncpus>0.500000</avg_ncpus>
<max_ncpus>0.500000</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
</app_version>
</app_info>
Die 0.5 Einstellungen muss man beibehalten um invalide Results unter 3sek. CPU-Zeit zu vermeiden.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#52 Post by rebirther » 18.01.2011, 17:35

Mega Prime found for The Riesel Problem

On 14 Jan 2011 9:03:07 UTC, PrimeGrid’s The Riesel Problem project eliminated k=428639 by finding the Mega prime: 428639*2^3506452-1

The prime is 1,055,553 digits long and enters Chris Caldwell's The Largest Known Primes Database ranked 29th overall. This is the 2nd largest prime found in The Riesel Problem and PrimeGrid's second elimination in 2 months. 62 k's now remain.

The discovery was made by Brett Melvold of Australia using an Intel i7 920 @ 2.67GHz with 6GB RAM, running Windows 7. This computer took 11 hours and 13 minutes to complete the primality test using LLR. Brett is a member of the BOINC@AUSTRALIA team.

More details to come. 18 Jan 2011 17:31:53 UTC
Wirklich Klasse!

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#53 Post by rebirther » 25.02.2011, 17:30

Es gibt jetzt endlich 3 Einstellungen für die GPU Sieve Teile, keine Ahnung ob es wirklich funktioniert, um gleichzeitig bei einem Projekt für CPU und das andere für GPU Arbeit zu bekommen, muß man lediglich das Häkchen setzen.

User avatar
grmbl
Admin
Posts: 274
Joined: 18.12.2005, 21:57
Location: Potsdam
Contact:

#54 Post by grmbl » 25.02.2011, 21:21

@reb Kannste dich erinnern, vor einen Monat hab ich mir die Funktion gewünscht. Nun gibt es sie.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#55 Post by rebirther » 25.02.2011, 21:29

grmbl wrote:@reb Kannste dich erinnern, vor einen Monat hab ich mir die Funktion gewünscht. Nun gibt es sie.
Das hatte ich schon im Forum mit diskutiert, ist aber alles noch buggy.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#56 Post by rebirther » 19.03.2011, 18:12

Und wieder ein Sieve Projekt dem Ende nahe:
4-6 weeks
321 Prime Search (Sieve) will be sieved to optimal depth.
Vielleicht mal Luft für neue Projekte. Wer noch ein Badge sammeln will, sollte sich ranhalten.

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#57 Post by rebirther » 19.04.2011, 19:20

Mal eine kleine Info zu prpnet dank pschoefer:

Leider ist die CUDA-App doch nicht so effizient bei gfn wie ich dachte, gegenüber einem Q9450@3,2Ghz braucht eine WU 3h vs. 12h, rechnet man 4 Kerne aus, so ist das Verhältnis 1:1, also eher ineffizient was den Stromverbrauch angeht.

Werde das Ding wohl mal am Wochenende testen und einen Guide drüber schreiben, wie man was einzustellen hat. Als Belohnung gibt es natürlich den PSA badge. Einzigstes unter etlichen ist immer noch das sr5 Projekt, wo noch die verbleibenden k-Werte gesucht werden, aber leider nicht über BOINC!

Das Programm ist im Downloadbereich unter BOINC Addons zu finden, 1x für CPU und 1x für GPU, da der Server bei Primegrid zu langsam ist für den Download (um die 15kb/s).

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#58 Post by rebirther » 23.04.2011, 19:32

Available: 321 Prime Search (Sieve) 45398

Nach dem Queue ist definitiv Schluss!

User avatar
rebirther
Admin
Posts: 2901
Joined: 19.12.2005, 00:59
Location: Germany

#59 Post by rebirther » 24.04.2011, 18:25

Für PRPnet Five or Bust double check werden unbedingt mehr Leute/Cores benötigt.
Please keep spreading the word. There are ~60 active participants. We are still shooting for 120 active participants...tell a teammate or two or three. :)

---------------------------------

For users reading this for the first time, please consider donating a core or two to the effort. If you are in search for your PSA badge (or upgrade the one you have), now is a good opportunity to get that done as well as give to a worthy cause. :) PSA credit is available.

To participate, please add the following server to your prpclient.ini file:

server=5oB:100:1:pgllr.mine.nu:13000
Man muß allerdings beachten, das die Laufzeit 10-15h dauert und je weiter man kommt, desto länger es werden kann. Checkpoints sollten aber vorhanden sein.

Natürlich kann man mittendrin auch mal stoppen (CTRL+C drücken). Wiederaufnahme mit Hilfe von Checkpoints sollte jederzeit möglich sein (unbedingt die master_prpclient.ini vorher bearbeiten und update drücken).

User avatar
odicin
Moderator
Posts: 390
Joined: 25.11.2010, 22:03
Location: Potsdam
Contact:

#60 Post by odicin » 24.04.2011, 19:49

Ich habe für 2 wochen einen 4c-server zur verfügung, bis er zum Einsatz kommt. Den werde ich mal mit 2cores rechnen lassen, die anderen beiden gehen auf PPS...

Gruß Odi

Post Reply