PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
11) Message boards : Aggie The Pew message board (Message 131341)
Posted 3 days ago by Profile Michael GoetzProject donor
I'm throwing around the idea of getting a system built and wondering if I go AMD? The 24 core threadripper 2970WX has become more reasonably priced recently. Does anyone have any experience with 16, 24 or 32 core threadrippers? The other option is another intel 99xx-X system with AVX512 such as the 9920-X which has half the cores of the 2970WX. Any ideas or other suggestions would be very much appreciated. Not sure yet if I will get a powerful graphics card too - I'm kind of thinking that 1 is enough - but I might change my mind.


Simplifying things a lot:

Tier 1: If you want to run LLR, you can't beat the Skylake-X CPUs (or the Xeon's that also have dual AVX-512 units.) Consider these CPUs to run at speed 1.0.

Tier 2: Next best would be any other Intel CPUs or the brand new Ryzen 3 CPUs. They run at speed 0.5.

Tier 3: The bottom tier is any AMD CPU that isn't a Ryzen 3. These run at speed 0.25.

If you can afford a Skylake-X system, I'd go with that. That's the king of the hill. If not, I'd probably go with Ryzen 3 as it's now equivalent to the non-AVX-512 Intel CPUs and is more affordable. I'd stay away from any AMD CPU that's more than a week old, however. They have more cores, but they run at half the speed of Intel. Ryzen 3 is the first AMD CPU that runs at full speed.

Ryzen 3 might be the best bang for the buck. You can put the money you save into a good AIO water cooler and a better GPU.

Thanks Michael. I can dump the threadripper idea. I didn't know they were so terrible at LLR. I will investigate Ryzen 3 and compare costs with a Skylake-X system.
Cheers, Nick


There's some recent threads on our forums where our early adopters are talking about their experiences with Ryzen 3.
12) Message boards : Seventeen or Bust : The SoB Double Check has begun (Message 131340)
Posted 3 days ago by Profile Michael GoetzProject donor
Just thinking aloud: Could the server send out a third task to a user after some sort of time limit, for example 10 days? I am guessing that you've all discussed this sort of thing for other projects, but I couldn't help spewing the idea out of my big fat mouth.


It's possible, yes.
13) Message boards : Aggie The Pew message board (Message 131332)
Posted 3 days ago by Profile Michael GoetzProject donor
I'm throwing around the idea of getting a system built and wondering if I go AMD? The 24 core threadripper 2970WX has become more reasonably priced recently. Does anyone have any experience with 16, 24 or 32 core threadrippers? The other option is another intel 99xx-X system with AVX512 such as the 9920-X which has half the cores of the 2970WX. Any ideas or other suggestions would be very much appreciated. Not sure yet if I will get a powerful graphics card too - I'm kind of thinking that 1 is enough - but I might change my mind.


Simplifying things a lot:

Tier 1: If you want to run LLR, you can't beat the Skylake-X CPUs (or the Xeon's that also have dual AVX-512 units.) Consider these CPUs to run at speed 1.0.

Tier 2: Next best would be any other Intel CPUs or the brand new AMD Zen 2 architecture CPUs. They run at speed 0.5.

Tier 3: The bottom tier is any AMD CPU that isn't a Zen 2. These run at speed 0.25.

If you can afford a Skylake-X system, I'd go with that. That's the king of the hill. If not, I'd probably go with Zen 2 as it's now equivalent to the non-AVX-512 Intel CPUs and is more affordable. I'd stay away from any AMD CPU that's more than a week old, however. They have more cores, but they run at half the speed of Intel. Ryzen 3 is the first AMD CPU that runs at full speed.

Zen 2 might be the best bang for the buck. You can put the money you save into a good AIO water cooler and a better GPU.

EDIT: Changed "Ryzen 3" to "Zen 2".
14) Message boards : Seventeen or Bust : The SoB Double Check has begun (Message 131331)
Posted 3 days ago by Profile Michael GoetzProject donor
Considering how rare a SOB find would be, please tell me there is a special badge? :)


You would get another "K" badge, and another "M" batch as well. See account for SyP. /JeppeSN


Yeah but there should also be a special and unique badge for a SOB find IMHO. Just how like GFN even has "worlds firsts badges"


It would be the largest known Proth prime, but we find "the largest known <some type of> prime" all the time without giving out special badges for it.

All conjectures already give out a special K badge.

It wouldn't be a world first, or even a PrimeGrid first, so that doesn't make sense.

But you bring up a good point. While, in this context, it makes no sense that there should be a special badge for finding the first prime -- since it wouldn't be the first prime -- perhaps there should be one for finding the last prime, i.e., for finally solving the conjecture.
15) Message boards : Aggie The Pew message board (Message 131320)
Posted 3 days ago by Profile Michael GoetzProject donor
Don't relax too much!

The next challenge is in less that two weeks!

There's a 7 day ESP challenge to honor Lennart starting on August 3rd at midnight.
16) Message boards : Number crunching : 50th Anniversary of the Moon Landing Challenge (Message 131319)
Posted 3 days ago by Profile Michael GoetzProject donor
Did the challenge have any significant impact on the PPS sub-project "life" expectancy ?
... In about 2 years, when PPS catches up to PPS-MEGA ...


I've already given you enough information to figure out the answer to that question.

Edit: If you need more of a hint:

977780 (86%) have returned a successful result. [959880 (85%) / 0 (0%) / 17900 (2%)]


The other important information was on Discord: we normally do about 16000 tasks per day.
17) Message boards : Number crunching : 50th Anniversary of the Moon Landing Challenge (Message 131314)
Posted 3 days ago by Profile Michael GoetzProject donor
The cleanup begins! It should about 2 to 4 weeks.

Cleanup status:
Jul 20: 50th Anniversary of the Moon Landing: 78226 tasks outstanding; 60522 affecting individual (283) scoring positions; 23305 affecting team (60) scoring positions.
18) Message boards : Number crunching : 50th Anniversary of the Moon Landing Challenge (Message 131312)
Posted 3 days ago by Profile Michael GoetzProject donor
And we're done! Final statistics:


Challenge: 50th Anniversary of the Moon Landing
App: 10 (PPS-LLR)
(As of 2019-07-20 20:28:47 UTC)

1134613 tasks have been sent out. [CPU/GPU/anonymous_platform: 1116663 (98%) / 0 (0%) / 17950 (2%)]

Of those tasks that have been sent out:

58186 (5%) were aborted. [58186 (5%) / 0 (0%) / 0 (0%)]
1413 (0%) came back with some kind of an error. [1413 (0%) / 0 (0%) / 0 (0%)]
977780 (86%) have returned a successful result. [959880 (85%) / 0 (0%) / 17900 (2%)]
92884 (8%) are still in progress. [92856 (8%) / 0 (0%) / 23 (0%)]

Of the tasks that have been returned successfully:

78697 (8%) are pending validation. [77164 (8%) / 0 (0%) / 1533 (0%)]
898498 (92%) have been successfully validated. [882131 (90%) / 0 (0%) / 16367 (2%)]
502 (0%) were invalid. [502 (0%) / 0 (0%) / 0 (0%)]
83 (0%) are inconclusive. [83 (0%) / 0 (0%) / 0 (0%)]

The current leading edge (i.e., latest work unit for which work has actually been sent out to a host) is n=2790454. The leading edge was at n=2749370 at the beginning of the challenge. Since the challenge started, the leading edge has advanced 1.49% as much as it had prior to the challenge!


A total of 16 PPS primes were found during the challenge. Great job everyone!

Now it's time to enjoy the A/C for a couple of days. Cleanup information coming soon, but this will be a quick cleanup.
19) Message boards : Number crunching : 50th Anniversary of the Moon Landing Challenge (Message 131305)
Posted 3 days ago by Profile Michael GoetzProject donor
Following up on the aborted tasks thing...

There's about 50K aborted tasks.

7 people have about 1K to 2K aborted tasks, 2 people have about 5K, and one person has about 20K. That person aborted 13K on the first day, and the rest on the second day.

Both of the people with 5K also aborted them all on the first two days.

Here's the breakdown, by date.

+------------+-------+ | d | cnt | +------------+-------+ | 2019-07-15 | 25411 | | 2019-07-16 | 21685 | | 2019-07-17 | 1805 | | 2019-07-18 | 1746 | | 2019-07-19 | 912 | | 2019-07-20 | 576 | +------------+-------+


As you can see, 46K of the 50K tasks all happened in the first two days as people were figuring it out.

Why the interest in the number of aborted tasks? (The reason I put them in the statisics, by the way, is not because I actually care how many were aborted. They used to be lumped in with "errors", and they're not really errors, and it made it look like there were a lot more errors than there actually were.)

20) Message boards : Number crunching : 50th Anniversary of the Moon Landing Challenge (Message 131303)
Posted 3 days ago by Profile Michael GoetzProject donor
1082957 tasks have been sent out.

The last time we had a challenge that sent out more than a million tasks was in November of 2017, when the Pierre de Fermat challenge sent out a combined 1159921 GFN-15, GFN-16, and GFN-17-Low tasks.

Of the 1082957 Tasks how many where Aborted by user?


5%, and most of those were from the very beginning of the challenge.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2019 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.02, 2.07, 1.71
Generated 24 Jul 2019 | 6:53:21 UTC