| Author |
Message |
|---|
 Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
                           
|
|
UPDATE: Wall-Sun-Sun has been turned back on. For more details, see this post.
Due to a problem with the GPU version of wwww, the Wall-Sun-Sun PRPNet server has been shut down until the problem is corrected. We do not expect the outage to last more than a couple of days.
The GPU version of wwww was producing wrong results, both false negatives and false positives. Therefore, all results previously computed by wwwwcl (the GPU version of wwww) need to be crunched again. That's roughly 87% of all work completed on Wall-Sun-Sun.
In addition, approximately one third of all WSS "finds" were found by wwwwcl, and are likely incorrect.
Before we can restart the server, we not only need to distribute new, fixed versions of wwwwcl, but we also need to update the server software so that it can reject incorrect results coming from the older wwwwcl applications. Not everyone reads the forums, so there's undoubtedly going to be situations where people keep running the older wwwwcl application.
Right now, the server is technically up so that work currently being processed can be returned and the web pages can be viewed, but no new work is being sent out.
If you are currently processing any Wall-Sun-Sun tasks with a GPU, please consider aborting them. The results that will be produced are invalid and will be discarded. If you return the tasks before the new server software is installed, you will get credit, but the results are useless. After the new server software is installed, results from old versions of wwwwcl will be rejected and you'll be forced to upgrade.
CPU versions of wwww are not affected, and continue to produce correct results. However, until we upgrade the server, no new work is being sent out.
There's no exact timetable yet for releasing the fixed wwwwcl, or for restarting the server, but I don't expect it to take very long.
As for the stats, the current thinking is that we're not going to remove the existing statistics. If you crunched the tasks, it's not your fault there was a bug in the software, so we're not going to penalize anyone. The GPU-derived near-WSS finds, however, once we verify they're false, will have to be removed, unfortunately. This is a fluid situation, so we may need to modify this policy if the need arises.
We apologize for any inconvenience this may cause.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Thanks for staying on top of things like this. The most important thing is that the results we produce are valid and true, and it's good to know that the science is always being vetted for problems like this! |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
|
I just turned Wall-Sun-Sun back on.
If you're using the GPU app wwwwcl or wwwwcl64, you MUST upgrade to the new 2.2.5 version of wwwwcl/wwwwcl64.
Results computed with older versions of wwwwcl/wwwwcl64 will be ignored and you'll be wasting your time and money.
This is a mandatory upgrade.
Information about the latest PRPNet software, including the updated wwwwcl, can be found here.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Let me say that this new package is AWESOME. With this package, for the first time, I can run wwwwcl64 with the 4-quad scripts WITHOUT three of the instances crashing immediately.
Good work. |
|
|
RogerVolunteer developer
Volunteer tester
Send message
Joined: 27 Nov 11
Posts: 1137
ID: 120786
Credit: 268,346,718
RAC: 8,270
            
|
|
Is this new version of wwwwcl mandatory for Wieferich? |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
Is this new version of wwwwcl mandatory for Wieferich?
No.
However, if at some point you switch to WSS and forget to upgrade, you'll find yourself crunching and crunching and crunching, and getting no credit for it. So I'd recommend upgrading anyway.
There was, unfortunately, no way to prevent sending work to older versions of wwwwcl, so if you run WSS with the old software you'll be needlessly crunching for nothing.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
The Linux 64 bit prpclient file does not have the execute bit set. You need to manually set it until it's fixed.
____________
|
|
|
|
|
|
Sorry about that, I've fixed it an uploaded a new package now.
The Linux 64 bit prpclient file does not have the execute bit set. You need to manually set it until it's fixed.
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
[list][list=][img][/img][url][color=red][/color][size=12][/size] ενταξει θα προσπαθησω![/url][/list][/list]
____________
|
|
|
|
|
|
WSS is running again but i think at the moment we continue crunching at the leading edge. When do we start redoing the work that was incorrect?
____________
ESP: Eliminated k=94373
SR5: Eliminated k=97366 and k=325918 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
WSS is running again but i think at the moment we continue crunching at the leading edge. When do we start redoing the work that was incorrect?
When we have enough free time to go and erase a year's worth of work. Things have been sort of hectic, and since there's no real rush to erase everyone's data, we're going to take the time to make sure we do it right. No time has been set, but I don't think it will be a lot longer.
____________
My lucky number is 75898524288+1 |
|
|
|
|
WSS is running again but i think at the moment we continue crunching at the leading edge. When do we start redoing the work that was incorrect?
When we have enough free time to go and erase a year's worth of work. Things have been sort of hectic, and since there's no real rush to erase everyone's data, we're going to take the time to make sure we do it right. No time has been set, but I don't think it will be a lot longer.
Great, thanks for the answer! I'm looking forward to that. In the meantime: full power on the current range...
____________
ESP: Eliminated k=94373
SR5: Eliminated k=97366 and k=325918 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
WSS is running again but i think at the moment we continue crunching at the leading edge. When do we start redoing the work that was incorrect?
When we have enough free time to go and erase a year's worth of work. Things have been sort of hectic, and since there's no real rush to erase everyone's data, we're going to take the time to make sure we do it right. No time has been set, but I don't think it will be a lot longer.
Great, thanks for the answer! I'm looking forward to that. In the meantime: full power on the current range...
All faulty calculations have been re-inserted into the queue to be processed again. That's about 1.7 million tests.
All faulty WSS 'finds' have been erased. 109 of the 177 near-WSS primes had to be voided.
Credit for faulty work will NOT be rescinded. You did the work, you get to keep the credit.
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
|
We are aware of a problem with WSS where it is deleting old test results as new results are being run for the old, faulty tests originally processed with the earlier version of wwwwcl.
This is not the intended behavior.
Erasing the old results causes the PRPNet user totals to drop. That is also not the intended behavior.
This will probably be fixed within a couple of days.
____________
My lucky number is 75898524288+1 |
|
|
pschoeferVolunteer developer
Volunteer tester
Send message
Joined: 20 Sep 05
Posts: 664
ID: 845
Credit: 2,304,468,412
RAC: 840,732
                   
|
|
Some ranges seem to be re-tested more than once. The latest special instance 2930875090703129 (0 -192 p) was reported by MiHost and me.
____________
  |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
Some ranges seem to be re-tested more than once. The latest special instance 2930875090703129 (0 -192 p) was reported by MiHost and me.
Lovely.
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
|
New work generation has been disabled until this latest problem can be understood and corrected.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Is the port down as well? I not able to get work and I read the post on no new work generated but wasn't sure if that meant work ready to send or actual sends.
Thanks |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
Is the port down as well? I not able to get work and I read the post on no new work generated but wasn't sure if that meant work ready to send or actual sends.
Thanks
It means the port is up but not sending new work. To the client, it will appear as if the port has run out of work, but it will respond to the client, accept completed results, and serve webpages. You just can't get new work.
It's like setting "no new work" on a BOINC client, except this is set on the PRPNet server.
____________
My lucky number is 75898524288+1 |
|
|
|
|
Is the port down as well? I not able to get work and I read the post on no new work generated but wasn't sure if that meant work ready to send or actual sends.
Thanks
It means the port is up but not sending new work. To the client, it will appear as if the port has run out of work, but it will respond to the client, accept completed results, and serve webpages. You just can't get new work.
It's like setting "no new work" on a BOINC client, except this is set on the PRPNet server.
Gotcha, thanks for the info.
Rick |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
|
New work generation is enabled again, and work is flowing once more. The problems with old results (and credit) being erased, and duplicate tasks being assigned, should be fixed now.
I'll be replacing the missing results in the near future, which will restore the missing credit.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
So, it's been a while since I've seen an update and was wondering how we stood on finishing the "rework". Based on the stats page it looks like we are getting fairly close to the end of the recycled work units?
____________
@AggieThePew
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13620
ID: 53948
Credit: 266,248,132
RAC: 296,587
|
So, it's been a while since I've seen an update and was wondering how we stood on finishing the "rework". Based on the stats page it looks like we are getting fairly close to the end of the recycled work units?
We started with 1.7 million tasks that needed to be repeated. There's currently 548 thousand remaining, so we're about 68% done.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Any update? )))
____________
wbr, Me. Dead J. Dona
 |
|
|
|
|
Any update? )))
Check this thread:
http://www.primegrid.com/forum_thread.php?id=4196&nowrap=true#69906
____________
676754^262144+1 is prime |
|
|
|
|
Any update? )))
See also this message. The trailing edge already passed 2138636e10 and is (as of this post) at 2423724e10, so everything that had to be redone has been redone AFAICT. |
|
|
