FormulaBOINC

[TennesseeTony]

Runtimes have updated to 3+ hours at 3.5GHz to 4.5+ hours at 2.5GHz. (Haswell-E) But only 90 minutes into it. RAM and pagefile usage is still low.

As long as no more sleeping giants wake up, we should place well enough, I think, even with diminished output. The sleeping giants are already running this one apparently, but their scores don't look too good, only in the millions. We have ZERO at the moment for 2017. I currently already have 1.08M points, and I think 750k of that was from one of the sub-projects from a BOINC Pentathlon, so only a few days to a week of processing got me 750k.

I think we'll be alright, and take Bronze. The Sottish BOINC team seem to lead the pack by a VERY large margin, so if they are still running it, OCN might have to settle for Silver again.

 

[4thKor]

You guys able to get any work? I've got two DL-580's with quad E7-4870's fired up and foaming at the mouth, but no tasks available.

 

[TennesseeTony]

Sorry guys, that's my fault. WUSS isn't known as the "Work Unit Stealing System" for nothing. I only had 1850 2300+ tasks last I checked.

EDIT: And now the moral dilemma... release some as I know I can't possibly finish them all in 4 days, or hold them in hopes of a drought, and win by being the only one with tasks........lol. But....I still don't really know how long they will take.... Maybe I DON'T have enough. Grrr.

 

[4thKor]

Sorry guys, that's my fault. WUSS isn't known as the "Work Unit Stealing System" for nothing. I only had 1850 tasks last I checked.

I learned something in school today. Sharing.

 

[TennesseeTony]

Yeah ok. So how long they run for on your E7-4870's?

Dumping some in a minute or two.

 

[4thKor]

I've never run this project. Hopefully about 3 minutes.

 

[TennesseeTony]

Lol. Alright, I released at least 450 494 birdies from their "network suspended" cages. They jumped onto the internet and it's the last I've seen of them. Catch them if you can!

 

[StefanR5R]

And now the moral dilemma...

...whether or not to take the gold medal with this queue of 2300, because no more WUs will be made available until after Easter. (WAG)

 

[4thKor]

I got one whole task.

 

[Ken g6]

I got one whole task.

I got two! I wasn't sure whether I should bunker more or not at the time; I guess I should have.

 

[StefanR5R]

So, some OCN users do have more or less moderate bunkers at the moment. Also, nf@h tasks do indeed have a quorum of one. (I looked at other users' tasks; I still got none on IVB-E and Phenom II.)

 

[StefanR5R]

Now I am getting tasks.

 

[TennesseeTony]

Well pooh. Looks like OCN has about 300 tasks more than I do, plus I seriously doubt I can make too much of a dent in the quantity I do have left, even after sharing some.

Alright boys...back to PrimeGrid for you I guess. Time for me to shut down the GPU crunching, so I can go 100% CPU for 4 days. At least my air conditioning system will be happy.

EDIT: Ah, good to see some more tasks coming in from somewhere.

EDIT 2: Looks like the lead project scientist is busy putting out more tasks, according to a pic over on OCN of a message/reply.

 

[StefanR5R]

Brought a Merom online. This will give 4thKor's Westmeres a run for their money for sure.

 

[Orange Kid]

I had pm'ed Eric on the boards over there and he was not aware of the race. Had he known there would have been more work at the start. Lots there now.

This project has been selected for a sprint. Now.
Your out of work and missing a huge chance to complete tons of work.

New work should be there now. How long does this sprint last?

How can I get advance warning of this next time? If I had known I could have had some extra work ready to go.

Eric

 

[StefanR5R]

Thanks Orange Kid. And respect to Eric for resolving this so quickly.

 

[Orange Kid]

We are not even on the list anywhere
No where to go but up .

1 The Scottish Boinc Team_25___17,941,203
2 US NAVY_____________18____5,574,791
3 Overclock.net_________ 15____2,845,160
4 AMD Users____________12____2,144,559
5 meisterkuehler.de Team__10____1,712,788
6 SETI@Netherlands______8_____1,389,942
7 Sicituradastra.____.______6_____1,321,187
8 Ars Technica___________4_____1,040,995
9 BOINC.SK_____________2______952,330
10 Crunching@EVGA______1_____435,821

 

[TennesseeTony]

Uh oh, I dusted off a few cores and the new downloaded tasks are completing in only 2 seconds. Either the new workunits have issues, or I need to put the dust back on those cores.

 

[Kiska]

Hrm, maybe I can add some cores into this race....

EDIT: OK I am adding cores....

So I've got about ~300 tasks between all my servers, don't know how many will be complete by the time it starts

 

[TennesseeTony]

Brought a Merom online. This will give 4thKor's Westmeres a run for their money for sure.

Lol! I had to come back and do some research here. My memory served me correctly, it would seem. Yeah, I'm betting that Core2 will put a hurting on his setup.

 

[crashtech]

So... I seem to have missed the starting gun on this one. I have 12 Westmere cores on the job now, and I am trying to figure out how to be of help here. Oddly enough, I have a Core2 Quad rig sitting quietly next to me, but my guess is that Tony's comment is in jest...?

 

[TennesseeTony]

The race hasn't started necessarily, we are just building up a bunker right now, so fill your queue then disable your network activity so they don't get uploaded yet. I presumed Stefan to be joking, as the Core2 is several generations prior to Westmere, but I don't know, the Westmere is putting up a decent fight against my Haswell-E systems!

I don't think anyone's ran this project on the TeAm in some time, not in full force at least. The C2D-Quad might actually be an effective weapon in this battle. It would be an interesting experiment!

 

[crashtech]

Yeah, the bunkering is what I meant. I have every machine with four or more cores working on this now, 30 Westmere cores, 4 Haswells, and 4 Skylakes. Not sure if the C2Q is worth the trouble, haha.

 

[TennesseeTony]

Very strange, for me the old Westmere is finishing tasks in half the time of the much newer Haswell-E systems. Simple variation in the batches of tasks, or some sort of architectural difference that's causing an advantage?

 

[StefanR5R]

Some task run times:

Merom (mobile Core 2 Duo),* 2.33 GHz, 32-bit Linux

3h30m, 3h20m (one of each completed when I checked several hours ago)​

Phenom II X4, 2.5 GHz, 64-bit Linux

2h40m, 2h30m, 1h10m (several of each completed by now)​

Ivy Bridge-E 6C/12T, 4.5 GHz, 64-bit Windows

2h15m, 1h30m, 1h10m, 55m, 2s** (several of each completed by now)​

*) The Westmere-EX destroyer. @crashtech, your Core 2 Quad might fare better because cooling of my Merom is borderline, and the 32 bit OS may be a handicap.

**) So far I got 8 WUs of the 2-seconds type which Tony mentioned.

Very strange, for me the old Westmere is finishing tasks in half the time of the much newer Haswell-E systems. Simple variation in the batches of tasks, or some sort of architectural difference that's causing an advantage?

My first bet would be variation of tasks. See above, there seem to be at least three types: ~2h tasks, ~1h tasks, and those ominous 2s tasks. Also, if you look at the credits of tasks which have been uploaded yesterday, they look like about two types of tasks credit-wise too.