PrimeNet 5.0 Upgrade

[jinydu]

Our number 1 contributor, Curtis Cooper, has started his transition. He's currently #10 on the overall producers list with 7223.231 GHz-days. Considering he had over 252K P90 CPU years before the transition, he's uncredited by a factor of over 177. But even after that, ANONYMOUS would still be ahead by a factor of more than 10.

I guess he'll need a lot of time and patience to install v25.7 on over 850 computers.

[uigrad]

Originally Posted by mdettweiler

Hmm...that's weird. I recently set up a separate mprime v25.7 folder for PRP testing with the Five or Bust project, and chose "just stress testing" when I first ran it; it made no attempt to contact PrimeNet, as it should have.

Thus I cannot understand why yours would have tried to contact PrimeNet after you pressed N for "no, I don't want to join GIMPS, just do stress testing". Or am I reading you wrong somehow?

Actually, I think that I may have canceled stress testing right after it started (cntrl-c), and then restarted mprime. Upon restarting it immediately picked up new work for anonymous, without asking any questions.

I can't remember for sure... I just know that I was concerned about it picking up work that I didn't want, and it did it anyway. I just created a clean directory and tried to replicate what happened, and I can't. It must have been something in the original prime.txt file that convinced mprime that I wasn't stress testing, and that I needed more work.

[petrw1]

Maybe I'm lucky OR maybe my farm is simple (2 PIII, 3 PIV, one PIV dual) but I have had no real issues; not that everything is as it should be yet but I understand I need to wait for the last details. Being an IT person myself, I understand that sometimes you need to make tough choices including going forward sooner than you would like to. I am sure they did not make the decision to go to v5/25.7 lightly.

1. My first 3 upgrades (1 PIII and 2 PIV) went easier than I expected: stop Prime95; backup my folder; download the new version; unzip into the same folder replacing files; restart.
2. Three upgrades to go including the dual core ... my fingers are loosely crossed.
3. As I upgraded each they showed up on my user account summary page.
4. New completions are being reported in since Oct 27.
5. New assignments are showing up as the correct type and to the correct PC (those that are upgraded)

A couple small concerns:
1. Prior v4 credits show up on my CPU page but not my summary page ... I understand I just need to wait a few more days. I can do that.
2. All my v4 assignmes show up as LL on the v5 server; even the factoring assignments. I hope it will be corrected but if not then as they complete the next few weeks they will merely be a memory.
3. I haven't figured out how to rename my "v4-computer" names for the old assignments yet but again that will either be corrected or roll off eventually.

So I will go on record as saying: "Scott, George and crew: you are doing just fine in my books".

[ADBjester]

If you stop the program, get worktodo.txt setup how you like it, and then restart, it will process exactly what you have in the worktodo.txt, regardless of what primenet says. The only question is whether primenet will let you turn them in when they complete.

This seems to have worked for now.... We'll see how it works when it comes time to submit the results.

Given the editing I had to do to add the "N/A," section and to put each individual assignment in the proper section.... my guess is that the average/casual user is going to find this too difficult. They'll hopefully just abandon the in-progress exponents in favor of newer ones, rather than abandon GIMPS entirely.

End result: I just have a few exponents queued up well in advance of when I'll need them... and I may need to ask for manual credit on the ones that were in progress when I switched over.

Jester

[uigrad]

Originally Posted by ADBjester

This seems to have worked for now.... We'll see how it works when it comes time to submit the results.

Given the editing I had to do to add the "N/A," section and to put each individual assignment in the proper section.... my guess is that the average/casual user is going to find this too difficult. They'll hopefully just abandon the in-progress exponents in favor of newer ones, rather than abandon GIMPS entirely.

End result: I just have a few exponents queued up well in advance of when I'll need them... and I may need to ask for manual credit on the ones that were in progress when I switched over.

Jester

Did you answer my question about whether the exponents appear in your assignment page online?

If they exist on that page, and are listed as v4_computers, then you should be able to turn them in with "N/A" for the key.

If they are on that page, but have a regular computer name, then primenet will not allow you to turn them in without the key. If you pulled out the key, and replaced it with "N/A", then you'll probably need to release them (you can do it through the web interface) before you can turn them in.

I didn't really understand that "N/A" was the key until a couple people responded to my post about it:
http://www.mersenneforum.org/showthr...142#post147142

Hope that helps.

[ADBjester]

Originally Posted by uigrad

Did you answer my question about whether the exponents appear in your assignment page online?

If they exist on that page, and are listed as v4_computers, then you should be able to turn them in with "N/A" for the key.

Sorry, I must have missed that.

Yes, the old exponents exist on that page for my v5 account:

v4_computers 0 39142601 LL 34.80% 2008-10-27 09:17 2 2008-10-29 11:44 2008-10-30 11:44 2008-11-17 11:44 19
v4_computers 0 39826379 LL 46.10% 2008-10-27 02:47 2 2008-10-29 18:27 2008-10-30 18:27 2008-11-15 18:27 17
v4_computers 0 42126521 LL 74.30% 2008-10-27 02:47 2 2008-10-29 18:27 2008-10-30 18:27 2008-11-06 18:27 8
v4_computers 0 42241393 LL 52.40% 2008-10-26 23:47 3 2008-10-29 18:26 2008-10-30 18:26 2008-11-14 18:26 16
v4_computers 0 43060733 LL 89.00% 2008-10-27 23:06 2 2008-10-29 00:37 2008-10-30 00:37 2008-11-03 00:37 5
v4_computers 0 43713101 LL 19.60% 2008-10-27 17:48 2 2008-10-29 18:27 2008-10-30 18:27 2008-12-02 18:27 34
v4_computers 0 44036719 LL 90.60% 2008-10-27 04:40 2 2008-10-29 06:58 2008-10-30 06:58 2008-11-03 06:58 5
v4_computers 0 44106793 LL 82.60% 2008-10-27 10:32 2 2008-10-29 11:39 2008-10-30 11:39 2008-11-06 11:39 8
PrestONE 1 46564261 LL 0.00% 2008-10-29 18:46 0 2008-10-29 18:46 2008-10-30 18:46 2008-12-16 19:22 48
PrestONE 2 46564303 LL 0.00% 2008-10-29 18:46 0 2008-10-29 18:46 2008-10-30 18:46 2008-12-16 19:22 48
PrestONE 3 46564327 LL 0.00% 2008-10-29 18:46 0 2008-10-29 18:46 2008-10-30 18:46 2008-12-16 19:22 48

If they are on that page, but have a regular computer name, then primenet will not allow you to turn them in without the key. If you pulled out the key, and replaced it with "N/A", then you'll probably need to release them (you can do it through the web interface) before you can turn them in.

The few that do have a computer name there also have a key -- they were the ones assigned today. I'll probably release those. For the others, there was no key, and I did not delete "N/A" -- I had to insert it into the old-style .ini file lines that I re-copied over.

Jester

[garo]

Jester, my guess is that your v24 communicated via the v4 bridge with the new server and your exponents were assigned to your virtual "v4_computer". So when you installed v25 and tried to comm, it was recognized as a separate program instance and the assignment was rejected.

[cheesehead]

Originally Posted by jinydu

Anyway, it seems that the v25.7 installation process is quite different from the process for v24 and earlier. It looks ready to run as soon as the download is finished; I wasn't asked for an installation location. I also wasn't given an opportunity to overwrite the previous version.

All that sounds like you downloaded a self-unpacking .exe file rather than the zip/gz files. Self-unpacking .exe's contain both the packed file and the software to unpack it. When one is executed, _then_ it would ask about where to put the unpacked software.

OTOH, at the moment, all I see for v25.7 is zip/gz files. The only self-unpacking executables I can find are for earlier versions. Where did you download from?

Does that mean I should download directly to the Prime95 folder?

No! Download it to some other folder (I have a /download especially for that purpose), then execute it from there. If it's self-unpacking, you'll know in a few seconds. If in fact it was a real Prime95 executable, that would quickly become obvious, too. Either way, you wouldn't have clobbered anything.

[Prime95]

Originally Posted by ADBjester

I *know* I should be in the top 500 lifetime (I've been here since near the beginning and my v4 account has over 700 P90 CPU years = 3600+ Ghz-Days), I'm not listed. Did my prior stats get assigned to "ANONYMOUS", and if so, how can I get ten years of work back?

Your credit will not be lost. I don't know if Scott can fix the "top producer" reports (I wrote the stored procedures, Scott did not) to add in past v4 credit before I get back from vacation in 2 1/2 weeks. If not, that will be a top priority when I return.

[Prime95]

Originally Posted by S45653

However, when it tried to send expected completion dates for the existing assignments it received "PrimeNet error 43: Invalid assignment key", then it abandoned the current tests and got new assignments from the server. I unchecked Options | Start at Bootup and then clicked Test | Exit.
what to do next?

Version 25.6 when talking to the v4 server generated funky 128-bit assignment IDs. Try this: Manually edit worktodo.txt, add in a line for each of the exponents that got an error 43 but leave out the "0000000..." assignment ID.
That is, if the v25.6 line was

Test=000000000000000000000000000xxxxx,exponent,69,1

add to your 25.7 worktodo.txt:

Test=exponent,69,1

[ET_]

CPU Name  	Exponent  	Result Type 	Received 	age  days      Result 	                                    GHz-Days
Manual testing	  65536	   NF-ECM       2008-10-30 11:09	  0.0          10 curves, B1=44000000, B2=4400000000	    9.9159
v4_computers	20252509	C	2008-10-27 14:42	  0.0	       SIGNED]__	                           16.4270
v4_computers	20252503	C	2008-10-27 14:42	  0.0	       SIGNED]__	                           16.4270
v4_computers	35325137	C	2008-10-27 14:18	  0.0	        __	                                   46.5114

It seems that part of the resultmessage was eaten by the server (see under the result column)

Luigi