Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Number crunching :
World Maths Day Challenge
Author |
Message |
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Welcome to the World Maths Day Challenge
World Maths Day - a day to promote numeracy through games and competition. It has been created for students aged 4 – 18 years with the addition for the first time this year of an adult category, ages 19 to 119. Only students in the 4 to 18 age categories are eligible for prizes. Adults just get to join in the fun! Compete against others around the world in live games of mental arithmetic.
This year's goal is 1 billion correct answers...a very ambitious goal considering last year's total was just under 480 million correct.
Come join the fun and register...there's no charge. Participate and compare your scores to others. :) NOTE: Actual start time of World Maths Day is 28 Feb 11:00 UTC and will last for 48 hours til 2 Mar 11:00 UTC.
In honor of World Maths Day, PrimeGrid is offering a 24 hour Challenge on The Riesel Problem (Sieve).
To participate in the Challenge, please select only the TRP (Sieve) project in your PrimeGrid preferences section. The Challenge will begin 1 March 2011 18:00 UTC and end 2 March 2011 18:00 UTC. Application builds are available for MacIntel, Linux 32 & 64 bit and Windows 32 & 64 bit. Note: 64 bit builds benefit from a 1.7X speed advantage over 32 bit...so 1.7X the credit. :)
NOTE: In your PrimeGrid preferences section, set "Send work from any subproject if selected projects have no work" to no to guarantee that no other work will be sent.
Recommendation: The TRP (Sieve) application requires a one time download of a 32.4MB sieve file. Please consider running a few TRP (Sieve) WU's before the Challenge begins so your client will already have the sieve file. This way, bandwidth can be used to deliver WU's instead of the sieve file. :) Thank you!
Time zone converter:
The World Clock - Time Zone Converter
NOTE: The countdown clock on the front page uses the host computer time. Therefore, if your computer time is off, so will the countdown clock. For precise timing, use the UTC Time in the data section to the left of the countdown clock.
Scoring Information
Scores will be kept for individuals and teams. Only work units issued AFTER 1 March 2011 18:00 UTC and received BEFORE 2 March 2011 18:00 UTC will be considered for credit. Since this is a fixed credit project, we'll be using cobblestones for scoring.
About The Riesel Problem
Hans Ivar Riesel (born 1929 in Stockholm) is a Swedish mathematician. In 1956, he showed that there are an infinite number of positive odd integer k's such that k*2^n-1 is composite (not prime) for every integer n>=1. These numbers are now called Riesel numbers. He further showed that k=509203 was such one.
It is conjectured that 509203 is the smallest Riesel number. The Riesel problem consists in determining that 509203 is the smallest Riesel number. To show that it is the smallest, a prime of the form k*2^n-1 must be found for each of the positive integer k's less than 509203. As of December 2010, there remains 62 k's for which no primes have been found. They are as follows:
2293, 9221, 23669, 31859, 38473, 40597, 46663, 65531, 67117, 74699, 81041, 93839, 97139, 107347, 121889, 123547, 129007, 141941, 143047, 146561, 161669, 162941, 192971, 206039, 206231, 215443, 226153, 234343, 245561, 250027, 252191, 273809, 304207, 315929, 319511, 324011, 325123, 327671, 336839, 342847, 344759, 353159, 362609, 363343, 364903, 365159, 368411, 371893, 384539, 386801, 397027, 398023, 402539, 409753, 415267, 444637, 470173, 474491, 477583, 485557, 494743, 502573
For a more detailed history and status of the Riesel problem, please visit Wilfrid Keller's The Riesel Problem: Definition and Status.
Additional Information
The Riesel problem is to k*2^n-1 as the Sierpinski problem is to k*2^n+1. There is no equivalent to the 'prime' Sierpinski problem since k=509203, the conjectured smallest Riesel number, is prime.
What is sieving?
Sieving is the first step to prime finding. In general, a sieve separates wanted/desired elements from unwanted material using a tool such as a mesh, net or other filtration or distillation methods. The word "sift" derives from this term. (Wikipedia - Sieve)
In PrimeGrid's case, the desired elements ultimately are prime numbers and the unwanted material are composite numbers. Our tool of choice for PSP/SoB sieve is Geoff Reynolds' sr2sieve program. It eliminates possible candidates by removing numbers that have small factors. As this process is much faster than primality testing, it is good to thoroughly sieve a data set before primality testing.
Sieving removes many candidates at the beginning. However, the deeper the sieve goes, the slower the rate of removal, till eventually sieving removes candidates at the same rate as primality testing. This is sometimes referred to as "optimal depth". Primality testing is recommended at this point.
There are many factors that determine how much time and how deep to sieve. After sieving, all the remaining candidates must be primality tested to determine their "prime" status.
32 bit OS with 64 bit CPU
This is for those who are "driving with the hand brake active" (running a 32 bit OS on a 64 bit machine) ;) Wubi is a very nice tool that installs Ubuntu as a dual boot to your 64 bit machine. It's as simple as adding a program to Windows. You can even uninstall it like you would any other program. :)
Wubi - Ubuntu Installer
After getting 64 bit OS running on your machine, here's a link to ALL BOINC clients (including Linux x64): http://boinc.berkeley.edu/download_all.php If you are new to Linux and need help getting set up, let us know.
It really is simple. :) With that said, check what installations WUBI will NOT work on: Unsupported set-ups
Ubuntu has a power save feature that throttles the CPU frequency down to save power. It does not treat BOINC as an active program so will throttle down thus causing under-performance. For those downloading the latest Ubuntu 10.04 (http://www.ubuntu.com/), please follow these directions to make sure the frequency is not throttled down while BOINC is running:
1. In Ubuntu, right click the top menu bar
2. Select add to panel
3. Select "CPU Frequency Scaling Monitor"
4. Click add
This will add a processor icon to the top menu. Click on the icon and set to "Performance". This also allows you to see the CPU frequency.
____________
| |
|
mackerel Volunteer tester
 Send message
Joined: 2 Oct 08 Posts: 2579 ID: 29980 Credit: 548,943,830 RAC: 30,613
                             
|
Sounds good to me :) Enough time for some SoB units to go through and for me to get some 64-bit OSes set up... | |
|
|
Too bad i only have one 64-bit system available.
But i will be there!
____________
| |
|
Honza Volunteer moderator Volunteer tester Project scientist Send message
Joined: 15 Aug 05 Posts: 1930 ID: 352 Credit: 5,473,123,160 RAC: 5,279,173
                                   
|
If it goes well, it will be my final push toward ruby badge during challenge...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 | |
|
|
If it goes well, it will be my final push toward ruby badge during challenge...
850K points in 1 day? :)
____________
Polish National Team | |
|
Honza Volunteer moderator Volunteer tester Project scientist Send message
Joined: 15 Aug 05 Posts: 1930 ID: 352 Credit: 5,473,123,160 RAC: 5,279,173
                                   
|
850K points in 1 day? :) That's from PSA/GPUs where RAC goes crazy.
I might be doing *only* ~100K daily on CPUs dozens servers...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 | |
|
|
This page: http://www.primegrid.com/challenge/challenge.php
still shows the 2010 challenge serie. And that long after the first
of the 2011 challenges is run, and now we are getting ready for the second one. ;-p
When will it be changed?
____________
| |
|
|
850K points in 1 day? :) That's from PSA/GPUs where RAC goes crazy.
I might be doing *only* ~100K daily on CPUs dozens servers...
It's pretty impressive too.
____________
Polish National Team | |
|
|
So what appropriate scoring system will be used?
Presumably one based on Reisel primes and whatever World Maths Day is about. (I refuse sites that just require flash just to even see it.) | |
|
|
So what appropriate scoring system will be used?
Presumably one based on Reisel primes and whatever World Maths Day is about. (I refuse sites that just require flash just to even see it.)
first post in this thread:
"Since this is a fixed credit project, we'll be using cobblestones for scoring." | |
|
|
So what appropriate scoring system will be used?
Presumably one based on Reisel primes and whatever World Maths Day is about. (I refuse sites that just require flash just to even see it.)
first post in this thread:
"Since this is a fixed credit project, we'll be using cobblestones for scoring."
Obviously you missed the point as I am trying to suggest that the scoring system should match the 'world maths day' not just the usual boring cobblestones. For example, '298.32' doesn't match (as it should be at least an odd natural number). But '357' (composed of three digits that come from at least two Reisel sources) could as I wanted a number near the current value. | |
|
|
Obviously you missed the point as I am trying to suggest that the scoring system should match the 'world maths day' not just the usual boring cobblestones. For example, '298.32' doesn't match (as it should be at least an odd natural number). But '357' (composed of three digits that come from at least two Reisel sources) could as I wanted a number near the current value.
aahh - now i get it!
357 is a nice number anyway.. ;)
| |
|
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
____________
Thanks,
Jim
| |
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
If you've done one WU since 28 January 2011 then you're OK. You can also look in the PrimeGrid folder to see if you have the current sieve file: TRP_20110128.sieveinput
____________
| |
|
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
If you've done one WU since 28 January 2011 then you're OK. You can also look in the PrimeGrid folder to see if you have the current sieve file: TRP_20110128.sieveinput
____________
Thanks,
Jim
| |
|
|
the server-clock seems to be about 3 minutes ahead.. | |
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 86,409,424 RAC: 12,023
                     
|
the server-clock seems to be about 3 minutes ahead..
Not anymore :) Thanks.
____________
| |
|
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
If you've done one WU since 28 January 2011 then you're OK. You can also look in the PrimeGrid folder to see if you have the current sieve file: TRP_20110128.sieveinput
If I delete all WUs just before the race, or detach and re-attach, will I need to download the 35MB file (again) ?
____________
| |
|
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
If you've done one WU since 28 January 2011 then you're OK. You can also look in the PrimeGrid folder to see if you have the current sieve file: TRP_20110128.sieveinput
If I delete all WUs just before the race, or detach and re-attach, will I need to download the 35MB file (again) ?
I hope someone more expert than I will answer, but I think if you just delete WUs, you will retain the sieve file. Detaching, I think, will delete everything associated with PG.
Jim
| |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
If I have run TRP Sieve in the past, do I already have that 35 MB file?
If you've done one WU since 28 January 2011 then you're OK. You can also look in the PrimeGrid folder to see if you have the current sieve file: TRP_20110128.sieveinput
If I delete all WUs just before the race, or detach and re-attach, will I need to download the 35MB file (again) ?
I hope someone more expert than I will answer, but I think if you just delete WUs, you will retain the sieve file. Detaching, I think, will delete everything associated with PG.
Jim
I believe that is correct. The sieve file *should* not be deleted when a WU completes, and aborting the WU is one form of completion.
Detaching from the project, however, completely erases all traces of the project on your computer, so the sieve file (and everything else related to Primegrid) gets erased.
____________
My lucky number is 75898524288+1 | |
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
How to Join the Challenge
To participate in the Challenge, please select only the TRP (Sieve) project in your PrimeGrid preferences section. Also, set "Send work from any subproject if selected projects have no work" to no to guarantee that no other work will be sent. The challenge will begin 1 March 2011 18:00 UTC and end 2 March 2011 18:00 UTC.
Application builds are available for the following:
- Microsoft Windows (98 or later) running on an Intel x86-compatible CPU
- Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU
- Linux running on an Intel x86-compatible CPU
- Linux running on an AMD x86_64 or Intel EM64T CPU
- Mac OS 10.4 or later running on Intel
- Mac OS 10.5+ running on an Intel 64-bit CPU
Note: 64 bit builds benefit from a 1.7X speed advantage over 32 bit...so 1.7X the credit. :)
Recommendation: The TRP (Sieve) application requires a one time download of a 32.4MB sieve file. Please consider running a few TRP (Sieve) WU's before the Challenge begins so your client will already have the sieve file. This way, bandwidth can be used to deliver WU's instead of the sieve file. :) Thank you!
Time zone converter:
The World Clock - Time Zone Converter
NOTE: The countdown clock on the front page uses the host computer time. Therefore, if your computer time is off, so will the countdown clock. For precise timing, use the UTC Time in the data section to the left of the countdown clock.
Scoring Information
Scores will be kept for individuals and teams. Only work units issued AFTER 1 March 2011 18:00 UTC and received BEFORE 2 March 2011 18:00 UTC will be considered for credit. Since this is a fixed credit project, we'll be using cobblestones for scoring.
Stats will be available on the front page next to the clock within 30 minutes after the start.
What to expect at Challenge start
The server always gets hit hard at the beginning of the Challenge. However, we will be using the same procedure that we've used for the past three years which has been very successful. A few hours before the start (1 March 2011 18:00 UTC) "max cache" will be dropped. Additionally, "max to send" will be dropped. Anyone attempting to request more will be met with "max cache" messages.
We'll raise the front page buffer before the start of the Challenge. These settings will be adjusted as necessary. Afterwards, as conditions allow, we'll raise "max cache". We'll continue to increase "max cache" when the server can handle it.
This method allows the greatest opportunity for the most clients to get work. Even if it's just 1 task for the first few minutes, at least the client is crunching. Past Challenges have shown that clients have an excellent chance to reach "max cache" before their first task is complete. We expect the same this time.
Strategies for starting the Challenge
Depending on a variety of factors, different strategies work for different users. Here are just a few to consider:
- large farm, user can be present at start
-Set Computer is connected to the Internet about every to 0 days
-Set Maintain enough work for an additional to 0 days
-Change PrimeGrid preferences to a fast WU project such as PPS (LLR)
-At Challenge start, update PrimeGrid preferences and only select TRP (Sieve) project
-At worse, you'll only be 1 PPS (LLR) WU late on each core starting the Challenge.
-After all machines have work, increase Maintain enough work for an additional
- large farm, user NOT able to be present at start
-Same settings as above.
-At worse, you'll only be 1 WU late on each core starting the Challenge.
- a few computers, user can be present at start
-Change PrimeGrid preferences and only select TRP (Sieve) project
-Set computers to "No New Tasks" for PrimeGrid
-At Challenge Start, update computers to "Allow New Tasks"
- a few computers, user NOT able to be present at start
-Set Computer is connected to the Internet about every to 0 days
-Set Maintain enough work for an additional to 0 days
-Change PrimeGrid preferences and only select TRP (Sieve) project
-At worse, you'll only be 1 WU late on each core starting the Challenge.
NOTE: This presumes you have all other BOINC projects suspended.
Best of Luck to everyone!
____________
| |
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 673 ID: 845 Credit: 2,496,300,451 RAC: 512,602
                           
|
the server-clock seems to be about 3 minutes ahead..
It's some seconds ahead again.
____________
| |
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
A few hours to go!!!
Please once again review the "Welcome" post as well as the "How to Join" post so you can be informed and make any last minute adjustments that are necessary.
An hour before the Challenge, we begin to monitor the Meebo chat room at the bottom of the forum page. If at any time PrimeGrid becomes unresponsive, you can still access the Meebo chat room externally here: http://www.meebo.com/room/primegrid/
Before the start, the Subproject stats on the Server Status page will be disabled as will the 24 hour stats on the bottom of the front page. Also, the Top Primes Finder page will be disabled. This is all done to free up as much resources as possible to allow the server to send and receive work. With these short WU's and the increased competition, we expect a very high volume at start.
What to expect at Challenge start
The server always gets hit hard at the beginning of the Challenge. However, we will be using the same procedure that we've used for the past three years which has been very successful. A few hours before the start (1 March 2011 18:00 UTC) "max cache" will be dropped. Additionally, "max to send" will be dropped. Anyone attempting to request more will be met with "max cache" messages.
We'll raise the front page buffer before the start of the Challenge. These settings will be adjusted as necessary. Afterwards, as conditions allow, we'll raise "max cache". We'll continue to increase "max cache" when the server can handle it.
This method allows the greatest opportunity for the most clients to get work. Even if it's just 1 task for the first few minutes, at least the client is crunching. Past Challenges have shown that clients have an excellent chance to reach "max cache" before their first task is complete. We expect the same this time.
Stats will be available on the front page next to the clock within 30 minutes after the start.
BOINCstats provides hourly updates of Project Challenges. There you can view user and team credit and ranking trends.
____________
| |
|
|
Stats will be available on the front page next to the clock within 30 minutes after the start.
THAT's really sportive.. ;)
| |
|
|
looks like a pretty smooth start - well done! | |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
Stats will be available on the front page next to the clock within 30 minutes after the start.
THAT's really sportive.. ;)
LOL
You could have correct stats available 30 milliseconds after the start.
Since these WUs take a few hours even on the fastest CPUs, the stats are going to be 100% accurate -- and blank -- for a while yet. :)
I suspect the fastest anyone could start to return challenge WUs is around 2 hours for a very heavily overclocked i7 with HT turned off. However, since this is an application that does generally benefit from HT, I suspect most people with HT CPUs will have HT turned on, so the first results will probably be coming back from Phenom IIs and Core2s in about 4 hours, or maybe 3 if the OC is cranked up really high.
Mine will start showing up around the 6 hour mark.
____________
My lucky number is 75898524288+1 | |
|
mackerel Volunteer tester
 Send message
Joined: 2 Oct 08 Posts: 2579 ID: 29980 Credit: 548,943,830 RAC: 30,613
                             
|
I fell for that too! Went to look at 30 mins and saw... nothing. Then I remember the WUs take hours...
My fastest box is an overclocked X6 at 3.5 GHz, doing a unit in just over 3 hours per core. But that was late starting from finishing some other work... I had set it to 4 GHz before, which if it was linear might do a unit in 2h40m. But I don't run that due to excessive power requirements to get there. | |
|
Honza Volunteer moderator Volunteer tester Project scientist Send message
Joined: 15 Aug 05 Posts: 1930 ID: 352 Credit: 5,473,123,160 RAC: 5,279,173
                                   
|
New Sandy Bridge i5 (model 2500 at stock 3,3Ghz) does 4 WUs in 2,5 hours. OC at 4Ghz would squeeze it around 2 hours.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 | |
|
|
i wonder who will be the first to report.. | |
|
Honza Volunteer moderator Volunteer tester Project scientist Send message
Joined: 15 Aug 05 Posts: 1930 ID: 352 Credit: 5,473,123,160 RAC: 5,279,173
                                   
|
i wonder who will be the first to report..
Wow, we have some heavy artillery doing Challenge with 12-cores hosts...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 | |
|
|
i wonder who will be the first to report..
Wow, we have some heavy artillery doing Challenge with 12-cores hosts...
this looks much more like the blitz.. ;)
| |
|
|
Well I know my output will be reduced due to not being able to download any new work on one computer even with a 10 day cache set, it keeps crunching 4 SoBs and a few Yoyo's. So I will do Proth Prime on it and let my other computers do the challenge bit with TRP..
Conan
EDIT (Well will wonders never cease, BOINC Manager has done what was asked of it and downloaded some TRP work units, oh Happy Day, so now I am off to crunch some numbers).
____________
| |
|
|
Hi John...
I want to tell you,
that we are still again with our SG-Booster...
best RealSanta | |
|
|
Well I know my output will be reduced due to not being able to download any new work on one computer even with a 10 day cache set...
I had the same problem Conan, with just the one laptop with an i7 that can happily crunch 8 at a time. Couldn't get any TRP at first, but sometime later as I was working out at the gym it worked out how to get hold of some.
____________
| |
|
|
...
I suspect the fastest anyone could start to return challenge WUs is around 2 hours for a very heavily overclocked i7 with HT turned off. However, since this is an application that does generally benefit from HT, I suspect most people with HT CPUs will have HT turned on, so the first results will probably be coming back from Phenom IIs and Core2s in about 4 hours, or maybe 3 if the OC is cranked up really high.
Mine will start showing up around the 6 hour mark.
I have a Phenom II 1090T moderately o/c'd to 3.7 GHz. It's taking a hair under 3 hours per w/u on each of 6 cores. Theoretically I could have submitted 8 rounds of w/u's during the 24 hour challenge, but due to my personal incompetence, I was about 30 minutes late getting started, so it will be only 7. Sigh. My other box is a Mac (core 2 duo, 2.9GHz - stock) and is taking about 3:35 per w/u.
--Gary | |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
...
I suspect the fastest anyone could start to return challenge WUs is around 2 hours for a very heavily overclocked i7 with HT turned off. However, since this is an application that does generally benefit from HT, I suspect most people with HT CPUs will have HT turned on, so the first results will probably be coming back from Phenom IIs and Core2s in about 4 hours, or maybe 3 if the OC is cranked up really high.
Mine will start showing up around the 6 hour mark.
I have a Phenom II 1090T moderately o/c'd to 3.7 GHz. It's taking a hair under 3 hours per w/u on each of 6 cores. Theoretically I could have submitted 8 rounds of w/u's during the 24 hour challenge, but due to my personal incompetence, I was about 30 minutes late getting started, so it will be only 7. Sigh. My other box is a Mac (core 2 duo, 2.9GHz - stock) and is taking about 3:35 per w/u.
--Gary
My estimates were a little high. My C2Q @2.4GHz ended up taking about 4:15 or so per WU.
____________
My lucky number is 75898524288+1 | |
|
|
Now this is funny, we apparently broke BoincStats with the TRP Sieve but not with the PPS llr. Go figure :)
____________
@AggieThePew
| |
|
|
Now this is funny, we apparently broke BoincStats with the TRP Sieve but not with the PPS llr. Go figure :)
not very likely - it's probably the really aged server again.. | |
|
|
Now this is funny, we apparently broke BoincStats with the TRP Sieve but not with the PPS llr. Go figure :)
not very likely - it's probably the really aged server again..
LOL Frank, I was just joking :)
____________
@AggieThePew
| |
|
|
Now this is funny, we apparently broke BoincStats with the TRP Sieve but not with the PPS llr. Go figure :)
not very likely - it's probably the really aged server again..
LOL Frank, I was just joking :)
i know - send some plaster to willy... ;) | |
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 14,760,451 RAC: 38,969
                     
|
Last night 3 of my TRP sieves on the same machine error-ed out with Memory Exceptions -- the problem was the wrapper, the actual sieve task kept on running. Normal CPU no OC. Any ideas, anyone?
http://www.primegrid.com/result.php?resultid=238614102
Moving this post to Problems and Help...
____________
Murphy (AtP)
| |
|
|
Someone forgot to program the countdown clock to stop at zero.
It's counting up now.
D'oh!
____________
www.chris-kent.co.uk aka Chief.com | |
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 14,760,451 RAC: 38,969
                     
|
Someone forgot to program the countdown clock to stop at zero.
It's counting up now.
D'oh!
That is so you know by how much you missed the deadline by...
____________
Murphy (AtP)
| |
|
|
Can you please not stop just yet, I have some WUs unfinished. My intel atom needs another 23 hours to finish the only two it started.
____________
| |
|
|
Can you please not stop just yet, I have some WUs unfinished. My intel atom needs another 23 hours to finish the only two it started.
Now that's a truly funny statement :)
____________
@AggieThePew
| |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
Last night 3 of my TRP sieves on the same machine error-ed out with Memory Exceptions -- the problem was the wrapper, the actual sieve task kept on running. Normal CPU no OC. Any ideas, anyone?
http://www.primegrid.com/result.php?resultid=238614102
Moving this post to Problems and Help...
I don't *know* exactly what happened, but perhaps I can give you some insight into the type of thing that happens.
When C programs need some memory to store data, they call the library routine malloc(), or some variation of it. Malloc then goes and gets memory from the operating system. (That's very simplified, of course, but that's sufficient for this explanation.)
Modern operating systema, including Windows, Linux, and Mac OS, all use what's called "Virtual Memory". Without going into details, the important thing to know is that there's a LOT of virtual memory. A ridiculously large amount for most programs. More than they could every use.
So what's happening when you get an 'out of memory' error? Most likely , it's one of three possibilities:
1) The program is requesting (probably due to a bug) a ridiculously large amount of memory -- more than the operating system can provide. That's not likely to have happened here because it happened to three WUs at once. The same bug wouldn't have struck three WUs simultaneously, so we can rule out a bug in the BOINC and PrimeGrid software.
2) Your computer really did run out of virtual memory. You would have noticed a lot of other symptoms of this, however. For this to happen, you would need to have a lot of applications open, and you would notice them all running very sluggishly, and a lot of other failures because every application would encounter failures trying to allocate memory.
3) Something else crashed, for some reason, and the side effects of that affected other parts of the computer. This is the most likely scenario. Something happened -- I couldn't even begin to speculate as to what -- which caused a problem for Windows. I'm saying Windows for simplicity's sake, but it applies to all computers. This other problem broke something in Windows, which affected its ability to allocate memory. Sometimes a problem may be temporary, and the problem with memory allocations go away. Sometimes it's not, and the whole computer crashes.
The bottom line is that the most likely scenario is that something else on the computer crashed in such a way that it affected the operating system's ability to allocate memory, perhaps temporarily. This, in turn, caused the errors in the WUs. If this is correct, then there's nothing in BOINC for you to fix. The problem lies elsewhere.
That's my $0.02, anyway. Take it with a grain of salt. Maybe someone else has a better answer.
____________
My lucky number is 75898524288+1 | |
|
|
The same bug wouldn't have struck three WUs simultaneously, so we can rule out a bug in the BOINC and PrimeGrid software.
Not exactly since these appear to be at different time because the WUs were sent and received at different times:
167030150 received at 2 Mar 2011 | 11:58:18 UTC
167030149 received at 2 Mar 2011 | 7:24:01 UTC
167023468 received at 2 Mar 2011 | 3:46:38 UTC
You only have 2GB with 2 threads under 32-bit XP so not a lot of 'free' memory. Really if the host requires more than about 0.5GB per WU then you may not have sufficient memory. So you may have gotten some sort of memory corruption or shared memory allocation that was incorrectly freed.
(Anyhow why 32-bit XP when the T7300 is 64-bit.) | |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
The same bug wouldn't have struck three WUs simultaneously, so we can rule out a bug in the BOINC and PrimeGrid software.
Not exactly since these appear to be at different time because the WUs were sent and received at different times:
167030150 received at 2 Mar 2011 | 11:58:18 UTC
167030149 received at 2 Mar 2011 | 7:24:01 UTC
167023468 received at 2 Mar 2011 | 3:46:38 UTC
You only have 2GB with 2 threads under 32-bit XP so not a lot of 'free' memory. Really if the host requires more than about 0.5GB per WU then you may not have sufficient memory. So you may have gotten some sort of memory corruption or shared memory allocation that was incorrectly freed.
(Anyhow why 32-bit XP when the T7300 is 64-bit.)
That's possible, especially if a lot of other stuff was running, but I still think it's unlikely that this is what happened. To start with, the WUs take no where near that much memory. I've run 4 sieves simultaneously in a VM with only 1.2 gig of ram allocated to the VM. They briefly use a lot of ram when initiallizing, but then drop a lot of it. I don't have one running right now, but I think they only use somewhere between 100 and 250 megs each.
More importantly, the amount of physical memory is only indirectly related to the "out of memory" problem. It's the amount of virtual memory that's important, and the default under XP (and most other versions of Windows) is for the virtual memory size to be twice the physical memory size. So there's actually 4 gigs of virtual memory available, which is way more than enough. To be sure, performance would suffer, but you shouldn't see "out of memory" errors when you exceed the physical memory size. That error should only happen when virtual memory is exhausted.
____________
My lucky number is 75898524288+1 | |
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 14,760,451 RAC: 38,969
                     
|
Why 32-bit XP -- Because I use that computer for work also.
There is currently less than 700 MB disk available to Boinc, and at the time of the problem, there was less than 70 MB. I may have been hit with an auto update of the OS (or other program) which took out the remaining available disk memory. I will try to clean up some disk before running big stuff again.
____________
Murphy (AtP)
| |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13781 ID: 53948 Credit: 344,062,692 RAC: 17,329
                              
|
Why 32-bit XP -- Because I use that computer for work also.
There is currently less than 700 MB disk available to Boinc, and at the time of the problem, there was less than 70 MB. I may have been hit with an auto update of the OS (or other program) which took out the remaining available disk memory. I will try to clean up some disk before running big stuff again.
OH.
How much TOTAL free space is there on the disk? If the amount of disk space available to the operating system is too low, all sorts of bad things can happen, including the "out of memory" errors that you saw.
The errors refer to RAM, not disk, but because virtual memory uses disk space as an extension of main memory (RAM), being low on disk space can cause this type of "out of memory" problem.
Also, if your computer was being updated while BOINC was running, and if the computer did an auto-restart, that could have also caused the problem you experienced. It's a good idea to shut down BOINC before restarting the computer.
____________
My lucky number is 75898524288+1 | |
|
|
The same bug wouldn't have struck three WUs simultaneously, so we can rule out a bug in the BOINC and PrimeGrid software.
Not exactly since these appear to be at different time because the WUs were sent and received at different times:
167030150 received at 2 Mar 2011 | 11:58:18 UTC
167030149 received at 2 Mar 2011 | 7:24:01 UTC
167023468 received at 2 Mar 2011 | 3:46:38 UTC
You only have 2GB with 2 threads under 32-bit XP so not a lot of 'free' memory. Really if the host requires more than about 0.5GB per WU then you may not have sufficient memory. So you may have gotten some sort of memory corruption or shared memory allocation that was incorrectly freed.
(Anyhow why 32-bit XP when the T7300 is 64-bit.)
That's possible, especially if a lot of other stuff was running, but I still think it's unlikely that this is what happened. To start with, the WUs take no where near that much memory. I've run 4 sieves simultaneously in a VM with only 1.2 gig of ram allocated to the VM. They briefly use a lot of ram when initiallizing, but then drop a lot of it. I don't have one running right now, but I think they only use somewhere between 100 and 250 megs each.
More importantly, the amount of physical memory is only indirectly related to the "out of memory" problem. It's the amount of virtual memory that's important, and the default under XP (and most other versions of Windows) is for the virtual memory size to be twice the physical memory size. So there's actually 4 gigs of virtual memory available, which is way more than enough. To be sure, performance would suffer, but you shouldn't see "out of memory" errors when you exceed the physical memory size. That error should only happen when virtual memory is exhausted.
I don't know the code nor windows but it could be the same one that gets other subprojects especially SoBs. Assuming no errors there and that this is windows xp so this applies:
This behavior can occur if the desktop heap in the WIN32 subsystem is depleted. Similarly for Vista and windows 7.
A less likely one that was for XP service packs 1a and 2 (not sure if applies to service pack 3) third-party program has to reserve lots of contiguous memory. | |
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The results are Final!
(belated closeout)
This Challenge was finalized on 2 March but the thread had yet to be closed out until now. Apologies for the delay in housekeeping.
A very respectable showing for the TRP (Sieve) and nice preparation for the Cinco de Mayo Challenge 5-13 May on the TRP (LLR) application. As for the World Maths Day challenge, it did not reach 1 billion correct answers. In fact, it failed to surpass last year's total. Although official results are not yet available, here's how this year compared to previous years:
- 2007 - 287,000 students from 98 countries answered 38,904,275 questions correctly
- 2008 - 1 million students registered from 150 countries answered 82,455,169 questions correctly
- 2009 - 1,952,879 students from 204 countries answered 452,681,681 questions correctly
- 2010 - 2,403,526 students from 236 countries answered 479,732,613 questions correctly
- 2011 - 5.3 million students registered from 218 countries answered 428,598,214 questions correctly (will be updated once final results are published)
Let's hope this down year doesn't turn into a trend. Thank you to everyone who participated.
As for the sieve, it was pushed to over 10P. Incredible progress from last March when PrimeGrid started the TRP (Sieve) effort from scratch and added 20M<n<50M. Almost 25K tasks were completed resulting in over 7.4M cobblestones awarded.
125 teams and 708 individuals participated in the Challenge.
Congratulations to Sicituradastra. as the top team and Mr. Hankey as the top individual. Remember, the top 100 participants and top 50 teams earn points for this Challenge.
Thank you to everyone. We appreciate your participation and hope you had fun!
Stats: Participants | Teams
Top 10 Teams
Rank Name Score
1 Sicituradastra. 806954.71
2 SETI.Germany 674202.46
3 SETI.USA 669131.02
4 Polish National Team 461202.21
5 JISATORA@W-ASCII 453147.58
6 Czech National Team 452550.94
7 Team 2ch 420034.10
8 BOINCstats 310550.78
9 PrimeSearchTeam 273559.14
10 The Knights Who Say Ni! 255659.96
Top 10 Individuals
Rank Name Team Score
1 Mr. Hankey SETI.USA 354702.09
2 halu JISATORA@W-ASCII 230004.47
3 Snf* Sicituradastra. 224038.07
4 nyabe Team 2ch 178395.16
5 lennart SM5YMT PrimeSearchTeam 160495.98
6 Scott Brown Duke University 144983.36
7 stevmcn 115449.71
8 3Rni Polish National Team 106201.80
9 Honza BOINC.SK 102025.33
10 Ross* Sicituradastra. 90390.86
____________
| |
|
Message boards :
Number crunching :
World Maths Day Challenge |