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

Toggle Menu

Join PrimeGrid

Returning Participants


Leader Boards



1) Message boards : Number crunching : Pi Approximation Day Challenge (Message 156284)
Posted 79 days ago by Ken_g6Project donor
3.14 days it is lol maybe.
Thanks guys

Hm. If it's Pi Approximation Day, should the challenge be Pi days long, ending on July 26, 01:23:54?
Or should it be 22/7 days long, which is 109 seconds longer, ending at July 26, 01:25:43?

I'm looking for a second opinion. ;)
2) Message boards : Number crunching : Intel GPU (Message 155352)
Posted 142 days ago by Ken_g6Project donor
The PPS Sieve OpenCL application was successfully tested on Intel GPU a long time ago.

I think PrimeGrid forked their BOINC server before BOINC introduced Intel GPUs. So adding that may be difficult at this point.
3) Message boards : Number crunching : Tour de Primes 2022 (Message 154477)
Posted 218 days ago by Ken_g6Project donor
You should have seen my former professor roll his eyes when I told him I was crunching with PG.
He's a guy who has classes of primes named after him (Wikipedia). Looking for a new project that could use some PG love?

Little projects tend to start out manually on Mersenneforum. I searched and they've already done a little along those lines.
4) Message boards : Number crunching : Intels new cpu's with efficiency cores and performance cores. (Message 153969)
Posted 234 days ago by Ken_g6Project donor
That video is old. Looks like Intel decided to disable AVX-512 on 12th-gen completely. Although certain motherboards might re-enable it.
5) Message boards : Number crunching : Tour de Primes 2022 (Message 153744)
Posted 239 days ago by Ken_g6Project donor
I'm jealous of this guy, whoever he is, with his Apple M1 processor.

I got to thinking, M1 is ARM, and I have a Raspberry Pi 4 (the fastest they make) with ARM. So maybe I could run an emulator and get some primes that way? Well, I didn't get BOINC running, but I got mprime running on one core with Qemu...

[Work thread Feb 1 03:14] Test 1, 400 Lucas-Lehmer iterations of M2621441 using type-2 FFT length 128K, Pass1=128, Pass2=1K, clm=2.
[Work thread Feb 1 03:20] Self-test 128K passed!

For comparison, on one core of my aging Skylake...

[Work thread Jan 31 20:00] Test 1, 36000 Lucas-Lehmer iterations of M7998783 using FMA3 FFT length 400K, Pass1=320, Pass2=1280, clm=4.
[Work thread Jan 31 20:01] Test 2, 36000 Lucas-Lehmer iterations of M7798785 using FMA3 FFT length 400K, Pass1=320, Pass2=1280, clm=4.
[Work thread Jan 31 20:02] Test 3, 36000 Lucas-Lehmer iterations of M7798783 using FMA3 FFT length 400K, Pass1=320, Pass2=1280, clm=4.
[Work thread Jan 31 20:03] Test 4, 44000 Lucas-Lehmer iterations of M7471105 using FMA3 FFT length 400K, Pass1=320, Pass2=1280, clm=4.
[Work thread Jan 31 20:04] Self-test 400K passed!

Apple really pulled off a miracle with their x86 emulation on the M1!
6) Message boards : Number crunching : Tour de Primes 2022 (Message 153588)
Posted 244 days ago by Ken_g6Project donor
When PPSE drops Top5K, next project is GFN16 to post T5K primes.

The only other non-MEGA is PPS, 954k digits as of now - and closing on PPS Mega as well.

We will have GFN-16 as entry-level primes and then only MEGA in future.
Or new project needs to be introduced...

Right now there are 1,230 known mega primes. Last year we found 269. If that rate continued, only mega primes would be in the Top5K within about 14 years. I expect that rate will increase.

So, if you want to introduce a new project for entry-level primes, maybe it should be one that will take less than 14 years? Maybe much less?
7) Message boards : Number crunching : How many threads to run to get best performance? (Message 152307)
Posted 314 days ago by Ken_g6Project donor
The general rule is cash. Spend as much cash as you can to get the best performance.

Wait, I think I misspelled something. Let's try again. ;)

The general rule is cache. Run as many tasks as you can without their total memory use overflowing your CPU's cache. (Much. A little might be OK, but that's debatable.)

How much memory does each task use? Eight times the FFT size.

How do you know how much cash you have? Check with your bank.

Dangit, miss-speeled again! :P

How do you know how much cache you have? For Intel, look at
8) Message boards : Number crunching : World Emoji Day Challenge (Message 150862)
Posted 437 days ago by Ken_g6Project donor

The current leading edge (i.e., latest work unit for which work has actually been sent out to a host) is b=28069820. The leading edge was at b=26442546 at the beginning of the challenge. Since the challenge started, the leading edge has advanced 6.15% as much as it had prior to the challenge!

From sub-project life expectancy:

Generalized Fermat Number 17-Low will be tested until b reaches the starting point of 17-Mega.

And that point, from GFN status by range, is 42,597,774

Therefore, GFN-17-Low is 65.9% complete. It was 62.1% complete yesterday. So we completed 3.8% of the entire (sub-)project yesterday!
9) Message boards : Number crunching : Sier"pi"nski's Birthday Challenge (Message 149421)
Posted 563 days ago by Ken_g6Project donor
So, how many people's computers started an hour early? I know mine did. Stupid daylight saving time.
10) Message boards : Number crunching : Hash rate for RTX nerf is happening, will it affect crunching? (Message 148877)
Posted 586 days ago by Ken_g6Project donor
But what is point doing this? Solution is very simple, we all know that many rigs with 3090 crunching ETH without any problem right now. Use current driver and problem is solved, because crunchers dont need latest drivers, they dont care is few FPS more in some games...
And maybe Windows will not stay on current drivers but Linux will. So you cannot change past...

I think this is specifically about keeping the price of an RTX 3060 low. Here's a table of GPU mining efficiency. Assuming half a 3060 Ti's hash rate, a 3060 would be potentially valuable to miners at its current MSRP. But, importantly, if its price went much higher they'd shy away from it.

Next 10 posts
[Return to PrimeGrid main page]
Copyright © 2005 - 2022 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 0.74, 0.67, 0.72
Generated 29 Sep 2022 | 7:08:39 UTC