Continuing problems with 130.237.232.237

[bruce]

This server had a problem in the January-March timeframe. After considerable work, the problem seemed to have been resolved. Merged problems with projects 6903/6904, Part 1

Unfortunately after a month or two, similar problems started cropping up again.
Merged problems with projects 6903/6904, Part 2 (April-June)

I have been asked not to merge the topics so here's a list of some more topics that are apparently part of the same issue but have not been merged:
No credit for 6903 (Run 6, Clone 17, Gen 88)
Same WU come down on different machines at different times!
No points on 6904 (Run 0, Clone 27, Gen 93)
No Credit Project: 6901 (Run 19, Clone 8, Gen 188)
No credit for Project: 6901 (Run 13, Clone 14, Gen 77)
No credit for Project: 6904 (Run 0, Clone 20, Gen 95)
Project: 6901 (Run 4, Clone 0, Gen 264)
Project: 6903 (Run 7, Clone 10, Gen 71)

[kasson]

There are different issues going on here. There was a stats problem on June 5 that we are investigating; the stats problem affected work units returned to this server but occurred on the stats server rather than on the work server. We will post more on that when we have more data and a resolution. The work units in question were logged and transmitted to the stats server, but credit was not entered. The reason for this is not yet clear.

[kasson]

Update: we found a single set of stats updates had not been processed correctly on June 5 for three servers: folding-2, vsp08f, and vspg9b. We are manually processing those now.
For server .237, this includes a set of returns on the following work units (P,R,G,C; note the inversion of the last two entries)
6901,13,77,14
6901,16,115,0
6901,3,81,9
6901,9,49,11
6903,11,114,8
6903,3,111,6
6903,5,92,1
6903,6,120,7
6903,6,88,17
6903,9,106,0
6904,0,93,27
6904,0,95,20

[bruce]

Confirmed: credit for (P6904 R0 C27 G93) was added to the stats database on 2012-06-16 15:35:04.

[highfiver]

I am having problems uploading the following completedWU's for the last three days

6903 (Run 9, Clone 19, Gen 69)
6901 (Run 9, Clone 19, Gen 211)
6903 (Run 5, Clone 12, Gen 127)
6903 (Run 4, Clone 16, Gen 125)
6904 (Run 2, Clone 33, Gen 97)

Can anyone help?

[Grandpa_01]

hi highfiver

Have you tried rebooting and or re setting your router / and or modem and computers, I have returned 10 bigadv WU's in the last three days to this server with no problem the last 1 was just an hour or so ago so my guess would be it is a problem with your ISP or equipment, do you have any logs from the rigs in question they may or may not be of some help.

[-alias-]

I can no longer get the P690x work units from server 130.237.232.237 and all my rigs is now linked to the server 128.143.231.201 where I only can get down the P8101 WUs. It's pretty boring to just fold P8101 on all four rigs day in and day out. My thought, to help the purpose, my team and myself, was to buy one Supermicro X9QRi-F motherboard http://www.supermicro.com/products/moth ... QRi-F_.cfm and 4 x E5-4650 http://ark.intel.com/products/64622/Int ... tel-QPI%29 and combine it in a fifth and last rig, but if only the WU P8101 is available in the future I might just have to forget about that plan.

Or is it just that the server 130.237.232.237 still has problems to serve all requests?

At the moment my rigs look like this, picture from last night:

The SR2 rig, I had to clock down from 3.6 to 3.5GHz in order to be able to fold the P8101. If I let it run at 3.6GHz, which is normal for all other WUer, then crash it every 2%.

[Grandpa_01]

Sorry but that is the way PG has chosen to go for now. You need less than 16 cores to get the 69xx WU's my Gulftowns are still getting them, you could take those 2 5645's out of the sr2 and put them in single processor boards OC them and make double the PPD running 69xx WU's pretty sad situation isn't it, a single processor rig can make more PPD than a MP.

[bollix47]

For the sr2 rig you could turn off Hyper Threading in the bios and show only 12 cores to the servers. That way you'll get the 690x projects and probably as much or more PPD than you're currently getting with 24 threads on the 8101 project.

[bruce]

Until the 690x projects end. No matter what you prefer, the Pande Group plans to benchmark future projects more like the 8101. Do not base future hardware plans on projects 690x. Deadlines have been tightened to be consistent with a 16 physical core requirement.

[bollix47]

Unfortunately, my dual hex xeon @2.66GHz can't reliably finish the 8101 project before the preferred deadline so I didn't have much choice but to do things that way. I could have switched to regular SMP and lost 50K PPD(I could have spent thousands less for that kind of PPD) but since I purchased that computer specifically for folding bigadv I decided to try continuing by dropping HT.

I realize it won't last forever but as the saying goes "make hay while the sun shines".

[Nathan_P]

Unfortunately, my dual hex xeon @2.66GHz can't reliably finish the 8101 project before the preferred deadline so I didn't have much choice but to do things that way. I could have switched to regular SMP and lost 50K PPD(I could have spent thousands less for that kind of PPD) but since I purchased that computer specifically for folding bigadv I decided to try continuing by dropping HT.

I realize it won't last forever but as the saying goes "make hay while the sun shines".

Just switched my slower dual hex xeon (2.4ghz) back to SMP, yes i have lost 50k PPD but its not just about the points. especially since the rig cost me less than £650, you cant get 53k PPD for that money - specially not pulling just 219watts from the wall

[-alias-]

Until the 690x projects end. No matter what you prefer, the Pande Group considers their benchmarking to have been an error and future projects will be benchmarked more like the 8101. Be sure to take that into account before you revise your hardware specifically to get projects 690x.

When do the 690x projects end then?
I have 2 x 980X rigs left from the last generation rigs, and one 2600K, but because the PPD/Watt for 980X is significantly lower then for G34 and the SR2 rigs I do not use them anymore for folding. I could stop the SR2 and fold on the 2600K with 690x, but my total PPD will go down so that is not an option.

I know that the discussion about 690x generation is over, but I can not help but wonder why they can not be allowed to run on the 4P rigs as long as they are here? Who would be harmed by such an arrangement? And when the new generation was ready, then changed one over just like that, in one big bang!

[Grandpa_01]

Until the 690x projects end. No matter what you prefer, the Pande Group considers their benchmarking to have been an error and future projects will be benchmarked more like the 8101. Be sure to take that into account before you revise your hardware specifically to get projects 690x.

When do the 690x projects end then?
I have 2 x 980X rigs left from the last generation rigs, and one 2600K, but because the PPD/Watt for 980X is significantly lower then for G34 and the SR2 rigs I do not use them anymore for folding. I could stop the SR2 and fold on the 2600K with 690x, but my total PPD will go down so that is not an option.

I know that the discussion about 690x generation is over, but I can not help but wonder why they can not be allowed to run on the 4P rigs as long as they are here? Who would be harmed by such an arrangement? And when the new generation was ready, then changed one over just like that, in one big bang!

I think bruce may have hit the nail on the head. Perhaps PG does not need any more MP rigs to run the bigadv projects, I very seriously doubt you will see 2 many more of them being built at this point, the incentive to build and run them has been removed. I know at this time I would not recommend to anybody that they build one.

[7im]

But yet I continue to read about them being built every day.

Add new donors who were never spoiled on the cake WUs like 6903 and 6904 will build them going forward.

Gratitude.


Now let's get back on topic of any issues with server .237.