It's awful quiet around here these days.
Hello enderak and spongetim , I see I have been left in the dust
I'm just running on two cores of a 2500k, it's a shame so much efficiency is lost when you try to run on 3 or 4 cores, it's just not worth it in my opinion.
Would be nice to see a new prime found, we have gone ages without one.
Oh well, keep on chugging along!
Yes, I haven't posted in years, busy with life and all that. But for early Christmas, I got five AMD quad cores, and boom: much higher production. So much better than the Athlon 1800-3000s I've been using all these years. So, tell me more about losing efficiency with multi cores. I'm super pleased with my new rigs, I don't see how it could not be worth it. Please explain, Nitrousine.
For example:Originally Posted by tim
With 2 cores, I get around 60 M
With 3 cores, it should be more like 90 M if it was as efficient, but it's more like maybe 80 M
With 4 cores, it should be about 120 M, but rather it is more like 90 M
It's been like this for years, and it would be nice to see efficiency the same no matter how many cores you use.
I don't think it slows down as much on AMD systems when using more cores.
Even though I said the above, I'm thinking about running on all cores on everything I have for a few weeks or so to try to catch up a little bit.
With the windows client, I had been getting .17-.25 seconds per iteration with Athlon 1800-3000 single cores. I'm now getting .036-.046 second per iteration per core with Athlon II x4 3.1 ghz.
Much of the fun for me in this is keeping the machines up and running, networked and maintained. I haven't done anything with optimizing the cpus, or with overclocking.
Where do you get the 60 M measurement? Is that from the jEMs/s metric from the seventeenorbust pending tests page?
Yeah, it's the jEMs/s from http://www.seventeenorbust.com/stats...html?teamID=60
Missed this thread. Not much going on with me, just have the client chugging away on a couple machines, one's an original i7 965 and the other is a Xeon L5410, both are running 2 workers with 2 threads each. I believe when I was testing I found this configuration to be just about as productive overall as 4 workers with 1 thread each, but I did it this way to prevent the individual tests from taking too long.
I took a break for a while to assist with the Five or Bust project. Since that project came to a completion, I switched back to SoB but haven't been doing much besides running the client in the background. Kind of stuck where I'm at in the rankings without the resources or much desire to increase production. A little disheartening to run the project for 5 years (just noticed I'll be hitting exactly 5 years sometime this week) and not see any primes crop up in that time and no activity in the forums in the past year. Just waiting for a prime to surface to breathe some new life into the project.
Posting Permissions
Reply With Quote