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
1) Message boards : Number crunching : 2021 Challenge Schedule (Message 147292)
Posted 64 days ago by rogue
Results for each year can also be found in Current Overall Standings.
In recent years, around 800 users and 200 teams scored points in the challenges.
It's a lot of information, since 2008.

What info exactly would you like to see summarized in a single page?


Look at my post from December 21st. That is a summary across all years, but the spreadsheet I used to generate that data needs more data such as:


    total number users validated results
    total results
    leading edge prior
    leading edge post
    primes found during search



This type of information helps one see how much growth there is in participate year over year. This is the type of summary I would like to see on a single web page.

Since I cannot attach a file to my posts, you cannot see the spreadsheet that I created. I'll post it on discord if case anyone is interested.

2) Message boards : Project Staging Area : Current n-limit of factorials & primorials? (Message 146992)
Posted 73 days ago by rogue
I think that I have worked through the bugs. I hadn't realized that the PRPNet project is testing to n=1e6, not n=1e7. I just ran a test with -p10001e9 -P10002e9 -n2e5 -N1e6 and it completed in around 30 minutes. This means that my GPU should be able to test a range of close to 5e10 in about 24 hours.

I estimate that at 1e13 that one candidate will be removed every 40 minutes, which is much faster than a PRP test. I don't know how deeply this project has been sieved.
3) Message boards : Project Staging Area : Current n-limit of factorials & primorials? (Message 146925)
Posted 75 days ago by rogue
Okay, I have some preliminary numbers for mfsieve. On the CPU, the speed of the new code compared to AVX is a wash, but I have noticed that the AVX code sometimes misses factors. Note that it doesn't mean that found factors are invalid, just that some of the expected factors are not found.

The big boost is on the GPU. I discovered that mfsievecl hasn't worked in a long time. It just crashes. I actually don't know the last time it worked correctly. The version I am testing now works. It is about 28x faster than the CPU on my laptop. Specifically speaking comparing an Intel i7-8850H to an NVIDIA Quadro P3200. I haven't run any tests on other systems at this time.

Anyways I estimate a range of 1e9 to complete (up to n=1e7) in about 50 hours on this laptop.
4) Message boards : Number crunching : 2021 Challenge Schedule (Message 146904)
Posted 75 days ago by rogue
I put all challenge history into a spreadsheet and generated the table below. There are too many different GFN challenges to track them all separately and as all GFN stats are on a single subproject status page, I combined them as a single "Project". This includes the 2021 challenges. It does not include any Tour de Primes challenges since those challenges are too general.

Project Type Challenges Total Days Last Challenge Year AP26 2 4 2009 AP27 4 16 2021 WW 1 3 2021 321 Sieving 2 3 2020 Cullen / Woodall Sieving 2 2 2012 GCW Sieving 1 10 2017 PPS Sieving 6 18 2020 PPSE Sieving 1 3 2010 PSP Sieving 5 9 2010 SoB, PSP, ESP Sieving 1 3 2015 TRP Sieving 6 16 2016 321 PRPing 11 73 2019 Cullen PRPing 3 20 2018 Cullen / Woodall PRPing 2 19 2020 ESP PRPing 4 22 2021 GCW PRPing 3 15 2021 GFN PRPing 12 108 2021 PPS PRPing 12 35 2019 PPS Div PRPing 3 15 2021 PPS Mega PRPing 1 5 2016 PPSE PRPing 2 4 2020 PSP PRPing 6 72 2021 SGS PRPing 8 14 2020 SoB, PSP, ESP PRPing 1 8 2015 SR5 PRPing 6 24 2020 SoB PRPing 3 40 2021 Twin PRPing 1 1 2008 TRP PRPing 8 47 2020 Woodall PRPing 5 47 2018


I would post the spreadsheet with the details behind these numbers, but I don't see a way to attach a file to a post.
5) Message boards : Project Staging Area : Current n-limit of factorials & primorials? (Message 146658)
Posted 82 days ago by rogue
Yves and I are looking into speed bumps for factorial sieving. Stay tuned.

Intriguing. Any hints on what to expect?


Possibly a doubling of the speed on CPUs. Uncertain on GPUs until it is implemented.
6) Message boards : Project Staging Area : Current n-limit of factorials & primorials? (Message 146644)
Posted 83 days ago by rogue
How many candidates have been sieved already? PRPnet says up to 280000 for factorials and 3200000 for primorials. It'll probably be half a year or more until this is reached for both projects. Or are there more candidates sieved but not prepared as tasks yet?

If not, what will be done when the limit is reached? Unless the project is abandoned sieving will be required. rogue said it's possible to modify mfsieve to work with P < 2^62 instead of the current 2^52 limit.

Are there any plans yet?


Yves and I are looking into speed bumps for factorial sieving. Stay tuned.
7) Message boards : Number crunching : Start discussing new goals for 2021 (Message 146607)
Posted 84 days ago by rogue
I can modify the main loop (which is in asm) to use extended floating point, (which is x86 only) to support sieving p up to 2^62.


Actually I do have extended floating point code, but am not using it, probably because mmx is faster, but the mmx code is only used for p < maxn.

As I think about this, there is an optimization to save on some mulmods. I'll think about that some more.
8) Message boards : Project Staging Area : FFT size for factorials and primorials (Message 146601)
Posted 85 days ago by rogue
Note that the FFT sizes are dependent upon the CPU as not all FFT routines are available on all CPUs.

PRPNet computes the decimal length of primorials and factorials when loaded into the database, so if a prime is found the length on the User Primes page will be the actual decimal length.

But you can use the number of bits to estimate the decimal length. I don't recall if it is the exact number of bits.
9) Message boards : Number crunching : Start discussing new goals for 2021 (Message 146600)
Posted 85 days ago by rogue
I can modify the main loop (which is in asm) to use extended floating point, (which is x86 only) to support sieving p up to 2^62.
10) Message boards : Number crunching : Start discussing new goals for 2021 (Message 146564)
Posted 85 days ago by rogue
I was searching for mfsieve, didn't know it was part of mtsieve, thanks.

The prime limit of 2^52 is probably the reason why sieving came to an end.

Sorry for OT...


It shouldn't be too hard to support up to 2^62, although AVX code cannot be used between 2^52 and 2^62.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2021 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 2.46, 2.78, 2.90
Generated 7 Mar 2021 | 15:08:41 UTC