mersenneforum.org  

Go Back   mersenneforum.org > Extra Stuff > Programming

Reply
 
Thread Tools
Old 2023-01-03, 04:27   #12
Dr Autonomy
 
Jan 2023
Alberta, Canada

11002 Posts
Default

If you would like to try the program, feel free; it's right there.
If not, that's great too.

You want screen shots? Here's an idea: run the program and take your own damn screen shots.
Revolutionary, huh?
Dr Autonomy is offline   Reply With Quote
Old 2023-01-03, 04:33   #13
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

8,069 Posts
Default

Quote:
Originally Posted by Dr Autonomy View Post
If you would like to try the program, feel free; it's right there.
Sure, just zip up the source code, with some documentation, include a copyright notice with your real name and contact info. Like George did (prime95), Ernst did (Mlucas), Mihai did (Gpuowl) etc.

Or keep up the secrecy, and it will go nowhere. Except perhaps into a sandbox jail, VM, with network sniffers, string searchers, debuggers, disassemblers, etc, sleuthing out what it might really be doing. Not by me, most likely, but there are folks here who claim to have spent years working at MS Research, or similarly selective employers.

"I'm just looking for feedback."
Take a look at Mihai's first post and gpuowl development thread. And the reference info collection I've been building for >4.5 years. No charge, except good manners in the future. Or you too can be on my ignore list. I had other things planned for this evening.

Oh, and those folks whose forum usernames appear in red? Make nice with them. They can move, modify, or remove your posts, remove attachments, and scold, warn or ban you or almost anyone else here for what they consider good cause.
Attached Files
File Type: pdf primality testing run time scalings compared.pdf (62.7 KB, 47 views)

Last fiddled with by kriesel on 2023-01-03 at 05:07
kriesel is offline   Reply With Quote
Old 2023-01-03, 04:37   #14
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

202458 Posts
Default

Huge kudos on going to the effort of writing an assembly language LLT program! Hopefully, you've learned a lot along the way. The GIMPS project has been going on for 25+ years and we've learned quite a lot along the way. We're happy to share what we've learned.

1) The standard way of comparing LL test programs is to output the low order 64-bits of the last LL iteration. We call that the residue. If the residues of two LL runs match we can be fairly certain both programs and computers succussfully tested the Mersenne number.

2) You shouldn't need help in deciding how your program scales. You can time 100 LL iterations of any Mersenne number and extrapolate how long a full LL test will take. Using this method how long will it take your i5 computer to testing exponents around 1 million, 10 million, 100 million, 1 billion? You can assume modern CPUs are 10 or 20 times faster than an old i5. I think you'll find an exponent near 1 billion will take a long, long time.
Prime95 is online now   Reply With Quote
Old 2023-01-03, 04:39   #15
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

1113410 Posts
Default

Quote:
Originally Posted by Dr Autonomy View Post
If you would like to try the program, feel free; it's right there.
If not, that's great too.

You want screen shots? Here's an idea: run the program and take your own damn screen shots.
Revolutionary, huh?
You are asking us to trust you. Most of us won't until your software has been shown not to be malicious. So, you should dial down the attitude. Please seriously consider sharing the source code. If you don't people will have a hard time trusting you.

You say that your program can test big numbers, yet you use a very, very, very slow way of carrying out important functions in the program. It is not worth the time to use your program then. Your method does not scale well.
You have included a TF function when that is well covered by other programs. Your local .db appears to be an attempt to ignore the great source of data that PrimeNet is.

We are skeptical with cause.

If you want to get suggestions and help improving your program, please have an attitude that shows that your are willing to learn and work with people.
Uncwilly is offline   Reply With Quote
Old 2023-01-03, 06:51   #16
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
San Diego, Calif.

2×3×1,733 Posts
Default

Quote:
Originally Posted by kriesel View Post
Link, or forum username?
There are plenty -
https://www.mersenneforum.org/showthread.php?t=20489
https://www.mersenneforum.org/showthread.php?t=23963
https://www.mersenneforum.org/showthread.php?t=23994
https://www.mersenneforum.org/showthread.php?t=22838
etc. (too many pages to scroll, but you can easily find more)
Batalov is offline   Reply With Quote
Old 2023-01-03, 07:09   #17
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

8,069 Posts
Default

I did a forum search on "Intel-compatible" before posing the question. Post 5 is pretty specific. I didn't find a match. It's not important, just curiosity. If the OP here did assembly code an fft (and we haven't yet seen any source code to confirm that), that's a cut above what we usually get as newcomers to the forum, and above what most of us forumites have done.
kriesel is offline   Reply With Quote
Old 2023-01-03, 09:28   #18
kruoli
 
kruoli's Avatar
 
"Oliver"
Sep 2017
Porta Westfalica, DE

111000001012 Posts
Default

FWIW, VirusTotal comes up clean.
kruoli is offline   Reply With Quote
Old 2023-01-03, 12:01   #19
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

8,069 Posts
Default

Under processes extra info after clicking activity summary full reports, zenbox at kruoli's link, there are screen shots.
An example displaying setting k range 1 to 100,000 for M756839 might mean it is doing TF inefficiently.
For nearby p=756923, fmax = 2 kmax p +1 = 151,384,600,001 ~37.14 bits.
Compare to the speed-optimized target 44 bits for https://www.mersenne.ca/exponent/756923;
Its kmax ~ 244/756923/2 ~ 11,620,855.8, ~116. times higher.
There's no P-1 in evidence in the zenbox screen shots.
https://www.mersenne.org/various/math.php

Hmm: https://learn.microsoft.com/en-us/wi...ndbox-overview How good is that?

Last fiddled with by kriesel on 2023-01-03 at 12:24
kriesel is offline   Reply With Quote
Old 2023-01-03, 17:37   #20
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

8,069 Posts
Default

GP_LLT has been added to the available software listing attachment at http://www.mersenneforum.org/showpos...91&postcount=2, which will likely be further updated as more information becomes available.

17 I wonder if it will accept command line parameters at launch, to avoid the GUI dialogs
18 Which processors here are supported? https://en.wikipedia.org/wiki/List_o...ors#Pentium_4F

Last fiddled with by kriesel on 2023-01-03 at 17:51
kriesel is offline   Reply With Quote
Old 2023-01-03, 21:37   #21
Dr Autonomy
 
Jan 2023
Alberta, Canada

22×3 Posts
Default

Well, thanks, I think..

For the record, Grade school squaring is NOT time complexity O(p2), it's O(Tp). The latter is approx. half the former.

Also, GP_LLT constantly updates the E.T.A. of the L.L.T.'s completion date while it runs the test, so there's never any confusion about how long it will take to finish.

As for command-line parameters, I would welcome any suggestions in that regard from anyone who has run GP_LLT on their system (especially if that system is a Threadripper system of any kind).

I may also add a "total time taken" display field to the Proof.exe program, which would simply total all the microsecond amounts in each iteration's record. Of course, that would depend on whether or not anyone actually runs the program(s)..

The best system I've seen out there thus far is one with 128 cores, therefore capable of 256 hardware-implemented concurrent threads of execution. If running GP_LLT on a BDD prime candidate, that translates to (approximately) 16 software threads per hardware-implemented concurrent thread. That's still pretty shaky, but it's close.

Lastly, many people still believe it is impossible to discover a BDD prime with current technology.

GP_LLT was written to achieve that "impossible" task..
Dr Autonomy is offline   Reply With Quote
Old 2023-01-03, 21:50   #22
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

255768 Posts
Default

Quote:
Originally Posted by Dr Autonomy View Post
Lastly, many people still believe it is impossible to discover a BDD prime with current technology.

GP_LLT was written to achieve that "impossible" task..
So, how long does it take to complete 1 BDD test using your software on your machine?

"Impossible" vs "impractical" are 2 different things.
Uncwilly is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Any way to pipe GMP mpz_t into PFGW? hansl Software 3 2019-04-09 19:44
Is there any program... mart_r Software 2 2009-11-15 20:06
So you think you can program rogue Lounge 5 2009-10-02 15:02
Program Primeinator Information & Answers 5 2009-07-16 21:42
which program? drakkar67 Prime Sierpinski Project 14 2005-11-29 06:25

All times are UTC. The time now is 01:29.


Sun Sep 24 01:29:15 UTC 2023 up 10 days, 23:11, 0 users, load averages: 1.08, 0.88, 0.79

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2023, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔