Team StarQwest's hourly information report.

[retina]

Originally Posted by Andi47

Does 4 cores / 8 threads for Bloomfield and Lynnfield processors mean, that these are hyperthreaded Quad-cores?

Yes, they brought back hyperthreading.

[Housemouse]

How will AMD's FireStream compare to the quad cores? I read it will process at up to 500 GFLOPs.

[jinydu]

The M332192831 test has been silent for nearly two weeks. The last update was at 28-Feb-08 20:16. The iteration number was 6,036,045.

[ixfd64]

Are you planning to get the 8-core "Nehalem-EX" when it comes out next year? :D

[robo_mojo]

Has there been any recent progress on this test?

[lavalamp]

His last update to Primenet was yesterday:

Originally Posted by http://mersenne.org/primenet/status.txt

******** 78 6817889 78.7 1942 60.3 18-Apr-08 15:53 31-Jan-08 14:22 StarQwest StarQwest1

He's 2.05% complete.

[StarQwest]

For those who are wondering why progress has stopped on M332192831, I have postponed further testing on this number until I get a faster computer. I am running a 2005 PC, which only contains a single Pentium D processor on this test. I will likely get a new computer later this year. To understand my reasoning for this, consider the following:

Imagine a spacecraft set to go to Alpha Centauri with an estimated travel time of 1,000 years. Over that time span, far faster spacecraft would be invented, and a craft invented 100 years later could possibly get there in 100 years, or 800 years before the original spacecraft. Similarly, my PC is simply to slow to effectively test M332192831, and any time lost in the meantime will more than be made up for when I someday get a faster computer. Eight core processors will come out early next year, which would shorten the testing time to approximately 2 years, as compared to 10 years on my current machine. So, the way I figure it, if I tested nonstop for the next 6 months on my machine, I would finish only 1.2 months sooner if I got an 8-core PC next year. It hardly seems worth it to me.

Please understand that I have NOT in any way given up on this test, and plan to resume it in the near future. It is just not practical at the current time, so I will resume testing 10-15 million digit numbers.

[starrynte]

however,

Originally Posted by retina

No. You should always start your job now. Because when new hardware comes along you can transfer your work to the new machine and continue the job.

If you do nothing now then the job will take longer later, even though the new hardware may be faster you still lose the initial computing time.

in the spacecraft example:

if you wait: it takes 1000 years to go to alpha centauri right now. in 100 years, it will take only 100 years to get to alpha centauri. if you start in 100 years, you get there in 200 years (100 years waiting, 100 years traveling)

whereas if you start now: in 100 years, you will have traveled 10% of the distance. you upgrade your spacecraft to a better one. as mentioned above, in 100 years the better spacecraft will take 100 years to get to alpha centauri from earth; however, you have already traveled 10% of the distance, leaving you with 90% left, or 90 years in the new spacecraft. so you continue in the upgraded spacecraft and arrive at alpha centauri 190 years after leaving earth now (100 years in first spacecraft, 90 years in upgraded one).

however, I do understand why you might stop testing; it is just too slow to wait months or even years for a result...

[Andi47]

Originally Posted by starrynte

however,

in the spacecraft example:

if you wait: it takes 1000 years to go to alpha centauri right now. in 100 years, it will take only 100 years to get to alpha centauri. if you start in 100 years, you get there in 200 years (100 years waiting, 100 years traveling)

whereas if you start now: in 100 years, you will have traveled 10% of the distance. you upgrade your spacecraft to a better one. as mentioned above, in 100 years the better spacecraft will take 100 years to get to alpha centauri from earth; however, you have already traveled 10% of the distance, leaving you with 90% left, or 90 years in the new spacecraft. so you continue in the upgraded spacecraft and arrive at alpha centauri 190 years after leaving earth now (100 years in first spacecraft, 90 years in upgraded one).

Hmmm... maybe a spacecraft is a bad example:

When you start now, you can't upgrade the spacecraft halfway to Alpha Centauri, because you need to be on earth to buy a new warp engine and a dock to properly mount it. So you will be faster when you wait until the docks in Utopia Planitia are building spaceships with warp drives.

But with Mersenne numbers, you can upgrade your hardware halfway through the LL-test and thus you will be faster when you start now.

[Mini-Geek]

Originally Posted by Andi47

But with Mersenne numbers, you can upgrade your hardware halfway through the LL-test and thus you will be faster when you start now.

Right, but at what cost? If you started it years ago on a 386 when it was state-of-the-art and upgraded it, say, every two years, you'd finish it some minute amount faster than if you started it on, say, state-of-the-art for 2010 hardware, but it would take a very long time, time that could be better spent searching for primes more reasonably sized for your hardware.

[retina]

Originally Posted by Mini-Geek

Right, but at what cost? If you started it years ago on a 386 when it was state-of-the-art and upgraded it, say, every two years, you'd finish it some minute amount faster than if you started it on, say, state-of-the-art for 2010 hardware, but it would take a very long time, time that could be better spent searching for primes more reasonably sized for your hardware.

But the work still needs to be done. It makes no difference what order you do it. Just run the smaller numbers later on your faster hardware. The overall finish time for all the work will be about the same, only the order of completion is different.