LLRnet/PRPnet rally August 12th-19th

[mdettweiler]

Originally Posted by vaughan

Where's MyDogBuster? With the rally running now his #2 rank is looking shaky.

I ran some numbers to see how much ROLP needs to increase its output in order to have a chance at beating PrimeSearchTeam in this rally, and it looks like we need to increase by ~150 pairs/hour in order to approximately break even with them. Hey Ian, I don't suppose you know anyone with that kind of firepower?

[vaughan]

Gary you don't need to read this. :)

Hey Max, while you have your calculator out ... how many more cores do I need to add to equal Gary's output?

[mdettweiler]

Originally Posted by vaughan

Gary you don't need to read this. :)

Hey Max, while you have your calculator out ... how many more cores do I need to add to equal Gary's output?

Just kidding--you're actually pretty close as it is. You'd need to add enough to produce about 34 pairs/hour--8 cores (i.e. 2 quads) should be enough to match Gary's output and go just a wee bit over it. (7 cores would just about exactly do it, but since most modern machines have cores in multiples of 2 I expect 8 would be more convenient. )

Of course, if you've got more to add, pour 'em on--if you add enough AMD Users might be able to actually gain on ROLP and upset their 2nd place for the rally. (That's the nice thing about 7-day rallies: there's actually some room to stage an upset or two! ) I haven't actually run the numbers on this one (it would take a bit more time which I don't have tonight), but if I had to take a rough guess I'd say adding 6 or 7 more quads might do the trick.

[gd_barnes]

Originally Posted by vaughan

Gary you don't need to read this. :)

Hey Max, while you have your calculator out ... how many more cores do I need to add to equal Gary's output?

I guess it all depends on whether you want to equal my TOTAL output for the entire week or just my hourly output -or- if you want an amount that will allow AMD users to overtake ROLP.

Bring it on!

[kar_bon]

Checking the result-page for G9000 is still/again giving "ERROR: Server cannot handle more connections"!

[Vato]

it's been out for 143 mins by the time i type this.
i might come back later in the week, as i don't have time for babysitting my machines at the moment.

[kar_bon]

Originally Posted by Vato

it's been out for 143 mins by the time i type this.
i might come back later in the week, as i don't have time for babysitting my machines at the moment.

Perhaps you can switch to Port 3000, LLRnet is more stable!

[gd_barnes]

Sorry guys. It's the middle of night here when PRPnet went haywire again and I just now saw this. I've now stopped and restarted it again. Karsten, if there's ever a reason to look at being able to VNC into my machine, now would be the time to look at what Max sent you regarding that.

We'll add whatever time is lost at the end of the rally to the PRPnet side of things.

Max and Mark, it's in your hands to see why this keeps happening. This is extremely frustrating in the middle of an otherwise successful rally.

[Vato]

Originally Posted by kar_bon

Perhaps you can switch to Port 3000, LLRnet is more stable!

LLRnet is showing it's old age and I won't touch it again.

[vaughan]

Aw come on Vato its not that bad! I'm running it on at least two dozen cores because I just don't understand how to setup the newfangled PRPnet thingy with its do.bat and so on.

[kar_bon]

Originally Posted by Vato

LLRnet is showing it's old age and I won't touch it again.

I don't know which LLRnet version you're using. See here for the new LLRnet which supports any new LLR.

Copy all files in a folder, edit "llrconfig.txt" to

server = "www.noprimeleftbehind.net"
port = 3000
username = "Vato"
WUCacheSize=1

and start "do.bat". That's all.
If you need any help, PM me.