The Scottish Boinc Team

http://boincstats.com/en/stats/challenge/team/chat/815

Name Need Faster Ships, Pirates Ahoy part deux NFS@Home Challenge
Status Upcoming
Project NFS@Home
Issued by Anguillan Pirates
Start time 2016-10-07 00:00 UTC
End time 2016-10-14 00:00 UTC
Late entrants allowed? Yes
Number of teams participating » 7
Number of users participating 0

Called on my AP account as per Carlos's request but I expect most of us will crunch for TSBT.
Carlos, will you post on the NFS forum as I am offshore just now and the internet here is seriously bad ?

Can someone also post a suitable message on the bam challenge site? Feel free to wind up gridcoin on behalf of Bryan :)

I say we crunch for the Anguillan Pirates team. TSBT doesn't need the FB points and no other team is close to us on NFS.

scole250 wrote:I say we crunch for the Anguillan Pirates team. TSBT doesn't need the FB points and no other team is close to us on NFS.

Yeah, but there is a team above us that we could challenge See if they can hold on to their 2.5M lead ... give us some incentive.

I will be crunching for ap. I can't call a challenge and then not crunch it. That's bad form. Others are of course free to choose whoever they want to crunch for. Ultimately it makes no difference to the project as long as the units are crunched

I've already posted at NFS@Home, Greg will move it shortly to news.
Thank you all.

You're welcome. Going to start trying to move systems over to NFS remotely on the worst connection I've been on outside of Africa. This might not go well.

Nice of Sici to drop another comp on top of our one. Very nice, cheers.

It's a GPU challenge. Shouldn't affect this one much.

Reminder that NFS IP address changed. If you want to bunker and have a firewall rule using IP address make sure it's using...
137.151.45.0 - 137.151.45.255

Thanks Scole. I'll give it a go........

Can anyone guess which Muppet forgot that he had set his account to crunch only 14e units a few weeks ago ? Now I'm swamped with the damn things. Not a happy pirate.

Oops!

If you wanted to donate 200 hours crunching them to me that would hit my next wuprop level and save me having to do it

Alez wrote:Can anyone guess which Muppet forgot that he had set his account to crunch only 14e units a few weeks ago ? Now I'm swamped with the damn things. Not a happy pirate.

Can you spell "abort" -or- "project reset"?

Pete, if you have the top 2 checked in preferences then you are screwed. The 1st one lasieved pays squat ... 1/4 what the "f" WU pay and they take the same amount of time. The "e" WU pay about 1/2.

You want to select "5f" as the only one. "f" no longer has any WU as it was replaced by the 5F application. Running those will pay 2-4X what the others pay for the same number of cpu cycles.

Please select ''lasieve5f - used for huge factorizations, uses up to 1 GB memory''

No it's me and yes I can spell abort etc. But not so easy when you are controlling stuff on crap connection with my phone. I got stuff moved I think and this close not going to risk breaking it all.
For best credit Pete you want the 16e units. Last option only but they do use a lot more memory.

Ap number one with a bullet. :)

Hey Pete, it's a trap! Scole, the bunker master, is sucking us in! Don't believe for a minute that he is maxed out at 7k/hour I've been down this road before :banghead:

Gridcoin has arrived and has taken 2nd place out on nowhere...

They better be ready to add some more WU. We're going through them faster than an elephant in a peanut patch.

Bryan wrote:Hey Pete, it's a trap! Scole, the bunker master, is sucking us in! Don't believe for a minute that he is maxed out at 7k/hour I've been down this road before :banghead:

That's all I got.

Dirk Broer wrote:Gridcoin has arrived and has taken 2nd place out on nowhere...

Thanks for the heads up Dirk. I put out a call for help ... maybe some folks will be available. Then again, maybe not.

It's hard to compete against a University system and a datacenter!!!!

scole250 wrote:They better be ready to add some more WU. We're going through them faster than an elephant in a peanut patch.

Bryan wrote:Hey Pete, it's a trap! Scole, the bunker master, is sucking us in! Don't believe for a minute that he is maxed out at 7k/hour I've been down this road before :banghead:

That's all I got.

Well if that's true then I'll go live ... I was holding back a little bit waiting for you to dump again. We need to do it anyway to see where we are against Gridcoin.

NFS still run best under linux?

I have the 2 I7 machines that were running the GPUs still on Windows and there doesn't seem to be any difference.

I just heard from Cineon_Lut (Crannog) and he is redirecting his fleet our way. Said it would take him a couple of hours. That will be a BIG boost.

Mumps is coming but he'll be crunching under someone else's account

Bryan wrote:I just heard from Cineon_Lut (Crannog) and he is redirecting his fleet our way. Said it would take him a couple of hours. That will be a BIG boost.

Remind Crannog to set his team to Anguillan Pirates.

scole250 wrote:

Bryan wrote:I just heard from Cineon_Lut (Crannog) and he is redirecting his fleet our way. Said it would take him a couple of hours. That will be a BIG boost.

Remind Crannog to set his team to Anguillan Pirates.

Yeah, I had told him in the PM asking for help and he acknowledged it. I think he will be bringing his whole fleet and not the partial he was running on CSG for us.

I'm going to be soaring with the eagles for a while The afterburners have been lit!

Tasks ready to send: 84801
Tasks in proces: 68127

Will there be a lasieveg for the 17e lattice in the future?

Dirk Broer wrote:Tasks ready to send: 84801
Tasks in proces: 68127

Will there be a lasieveg for the 17e lattice in the future?

There's a discussion at mersenneforum about doing a GNFS(251) which not yet fully in agreement if it will needs a new application. The experts say it can be run with lasieve5f application but in case not someone needs to code a new application. This would be a world factorization GNFS record if NFS decides to do it and it would need more than 5000 cores running for 3-4 months straight without taking into consideration the time to spend on GPU's for the polynomial and access to a big cluster to run the Linear Algebra (data more or less sized 10Terabytes).

More tasks will be added shortly, there's a lot of integers to sieve: 6,460+, 11,671M, 7,373- and 3,703+. Each one needs more or less 300k wu's.

Carlos

pinhodecarlos wrote:it would need more than 5000 cores running for 3-4 months straight without taking into consideration the time to spend on GPU's for the polynomial and access to a big cluster to run the Linear Algebra (data more or less sized 10Terabytes). Carlos

Well, in some months AMD Zen systems may be let loose that have (hopefully) motherboards for four 32-core, 64-thread CPUs.
That would be 256 threads(!) on such a system alone. I daresay Intel won't sit still on this one too, and will come with new Xeons with 32 cores -and an added Xeon Phi?- in the near future too.

Dirk Broer wrote:

pinhodecarlos wrote:it would need more than 5000 cores running for 3-4 months straight without taking into consideration the time to spend on GPU's for the polynomial and access to a big cluster to run the Linear Algebra (data more or less sized 10Terabytes). Carlos

Well, in some months AMD Zen systems may be let loose that have (hopefully) motherboards for four 32-core, 64-thread CPUs.
That would be 256 threads(!) on such a system alone. I daresay Intel won't sit still on this one too, and will come with new Xeons with 32 cores -and an added Xeon Phi?- in the near future too.

For big factorizations you need more than 256 threads. The usual cluster used has more than 512 cores.
Also let me share it is scheduled to be tested some applications to the newest server bought by the mersenneforum community which includes one of the latest Xeon phi.

Bryan was just trying to make me feel better before he blows by me

Da Seeker wrote:

Alez wrote:Ap number one with a bullet. :)

It seems we sturred up a hornest nest with that bullet.

No sweat. Anguillan Pirates wrote the book on the Kobayashi Maru test.

BTW...the 16e WUs run faster under linux on the Z600/T5500 systems but faster under Win64 on the newer rigs.

Da Seeker wrote:

Bryan wrote:

Dirk Broer wrote:Gridcoin has arrived and has taken 2nd place out on nowhere...

Thanks for the heads up Dirk. I put out a call for help ... maybe some folks will be available. Then again, maybe not.

It's hard to compete against a University system and a datacenter!!!!

Not only that. 41 users against 9.

It's not quantity that counts, it is purely quality. Every Pirate is a quality person ... even if I have to say so myself

Same as last time. Quality against quantity and the Borg were vanquished. Sent orders home via carrier pigeon to kill all gpu tasks and get everything cracking on NFS. Servers were only running 12 hours a day. That's being fixed, all sails being set.

So far so good. GC is reducing the Pirate's lead but unless something changes they will run out of time before they can overtake the Pirates.

SUSA should drop them to 3rd place tonight!

SUSA is #2 and it appears that NeuraMiner has pulled off so the Pirates should be safe at #1 .... all-in-all a very successful challenge

Come on NeuralMiner. You can't back off half way through a challenge. NFS still needs more lasieve5f WUs completed.

It looks like TSBT got into some good wind ... those sails must have been billowing!

Seems the Borg loose again :) might even be safe to jump some systems over to tsbt.

I have the 7 day challenge blues I did earn my 5th WUProp purple app star on NFS 16e Lattice Sieve V5 though :)

scole250 wrote:I have the 7 day challenge blues I did earn my 5th WUProp purple app star on NFS 16e Lattice Sieve V5 though :)

I'm shocked!!
Congrats on the badge upgrade though

I like a Team Challenge, but they are like company and fish. After a few days, they start to stink

scole250 wrote:I like a Team Challenge, but they are like company and fish. After a few days, they start to stink

I agree .. 3 days is enough especially if it is one where you bunker. I like a sprint, not a marathon

Da Seeker wrote:

Bryan wrote:

scole250 wrote:I like a Team Challenge, but they are like company and fish. After a few days, they start to stink

I agree .. 3 days is enough especially if it is one where you bunker. I like a sprint, not a marathon

I incline to think similarly. Four days is max. No bunkering and WU's to count only after the start. Like a good pirate, I would only participate if I had a chance of winning. :pirate:

I don't know Pete, a bunker is like coming out of a fog bank and nailing them. They don't see ya coming

I agree with Bryan, it's more exciting and challenging to run after a team.
Anyway, thank you all for the crunching time on all applications. Right now we only have two persons post-processing the 14e queue, myself with one quad-core machine, and Sean. 15e completed ones are with Tom and the 16e ones are with administrator Greg on two dual 10 core machines.

2,1285- is still on standby because we need access to a big cluster (512 cores or more).

A very good showing for Anguillan Pirates who finished in 1st and a good finish for TSBT who finished in 4th. SETI.USA did ok too finishing in a close 2nd.

This was a friends and family effort so I'll include individual finishes for all who I think many have contributed to either Anguillan Pirates or TSBT...
Bryan 1st
Crannog 2nd
Da Seeker 5th
scole of TSBT 6th
somanyroads 7th
Dave GPU 11th
STMahlberg 12th
alez [ TSBTs Pirate ] 13th
Sphynx 17th
MindCrime 18th
Paul of TSBT 27th
Dirk Broer 35th
Carlos Pinho [TSBTs Pirate] 45th
denim 47th
Skywalker TSBT 48th
dcushing 62nd
Anguillan Pirate 67th
Steve 112th
phildemeur 116th

Well done all :pirate:

Great results everyone, a hearty ration of rum all round was well earned I think