mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Data

Reply
 
Thread Tools
Old 2020-08-08, 15:02   #210
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3·5·431 Posts
Default

Quote:
Originally Posted by James Heinrich View Post
It's an unusual work distribution to be sure, but it could be legitimate. It works out to a bit over 2000 GHz-days of work per day, certainly achievable with a single RTX 2080.
About 98-99% of which is wasted, making it the equivalent of running a Quadro 2000 a few hours a day. Those 85-bit depths are suitable for ~800M exponents, which would take 1/4 as long because f=2kp+1.
Quote:
If it were my call I would suggest that TF credit be capped at the target TF level for that exponent (78-79 for exponents in this range), higher TF results could be accepted but no credit given since (as far as the GIMPS project is concerned) it's a waste of resources.

I'll pass this along to George to see if he wants to investigate further, perhaps talk to "Dirk" about it.
Maybe allow one bit higher than GPU72 target, which may be about right for RTX20xx.
Generally, users ought run TF to GPU72 bounds, P-1 to PrimeNet bounds the one and only time P-1 is run, and PRP with proof, or move on to other exponents.

And note there may be a downward adjustment in TF bit level and P-1 bounds targets coming, thanks to the introduction of and transition to PRP-proof & CERT, reducing verification effort from equal to a primality test, to ~1% of one.
kriesel is offline   Reply With Quote
Old 2020-08-08, 16:35   #211
masser
 
masser's Avatar
 
Jul 2003
Behind BB

7·269 Posts
Default

Quote:
Originally Posted by kriesel View Post
About 98-99% of which is wasted
Wasted for the prime search, yes. Not necessarily wasted if you're interested in finding factors, testing software/hardware, etc.
masser is offline   Reply With Quote
Old 2020-08-08, 17:42   #212
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3×5×431 Posts
Default

Quote:
Originally Posted by masser View Post
Wasted for the prime search, yes. Not necessarily wasted if you're interested in finding factors, testing software/hardware, etc.
If interested in finding factors, P-1 would have been a better choice after 78 bits TF. TF is also not a great hardware/software test, since the usual result is "no factor" which can hide a lot of undetected errors. Memory testing software would be a better choice for finding hardware issues. Seven bits too deep for optimal TF is about enough time to PRP-proof the exponent on even the most TF-heavy gpus.

Per mersenne.ca lookups, RTX2080
TF: 2623 GhzD/day
LL: 65
ratio: 40.4:1
Excess TF at https://www.mersenne.ca/exponent/205132043, 37602. GhzD/day (14.3 days excess TF)
P-1 needed 61.87 GhzDays (0.95 days needed)
LL estimate or PRP-proof: 1750.4 GhzDays (26.9 days; recent Gpuowl takes about 0.6x as long, ~16.2 days), actually ~3% less if allowing for some probability of a P-1 factor found.

Dirk's kit, Dirk's call though.

Last fiddled with by kriesel on 2020-08-08 at 17:57
kriesel is offline   Reply With Quote
Old 2021-12-14, 22:46   #213
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3·5·431 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
Or don't indulge the child.
Predict M52 is equal opportunity debunking within limits.
Our collective record of predicting composites in Predict Mxx remains perfect so far. The computations, not so much. On the one you linked, I had a gpuowl crash during PRP proof generation, and got a completed proof generation on restart.
The temp files are gone, so I can't readily do much about that proof cert fail issue.
kriesel is offline   Reply With Quote
Old 2021-12-14, 23:17   #214
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

2×3×1,741 Posts
Default

Quote:
Originally Posted by kriesel View Post
The temp files are gone, so I can't readily do much about that proof cert fail issue.
So?

I've actually found this an interesting study in just how much can be communicated with one-word answers.

How short can the word be? How important is the surrounding punctuation?

Sorry... I'm Canadian... Eh?
chalsall is offline   Reply With Quote
Old 2021-12-14, 23:25   #215
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3×5×431 Posts
Default

Quote:
Originally Posted by chalsall View Post
So?
0
kriesel is offline   Reply With Quote
Old 2021-12-15, 02:09   #216
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

2×5×73 Posts
Default

Quote:
Originally Posted by kriesel View Post
so I can't readily do much about that proof cert fail issue.
UncWilly again believed it was my fault because another user got a suspected PRP result on an exponent I guessed.

I failed once a little longer than a month ago, Dobri did it too, the application should reserve the temp files until certified.

Quote:
Originally Posted by chalsall View Post
So?

I've actually found this an interesting study in just how much can be communicated with one-word answers.

How short can the word be? How important is the surrounding punctuation?
The 1 word "Trumpet" seems to have the most influence among the modern society.

Last fiddled with by tuckerkao on 2021-12-15 at 02:14
tuckerkao is offline   Reply With Quote
Old 2022-01-26, 16:05   #217
xx005fs
 
"Eric"
Jan 2018
USA

110110002 Posts
Default Here we go again

User "for you" just submitted 3 TF assignments that have no factors with one of them between 2^127 and 2^1024 bits and the other 2 between 2^75 to 2^128 and 2^80 to 2^128 respectively.
xx005fs is offline   Reply With Quote
Old 2022-01-26, 20:39   #218
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

145018 Posts
Default

Yep. It's been discussed on multiple threads. I expect server or firewall countermeasures are being implemented, as well as database repair and possibly other actions. Perhaps new userid holders would think twice about such misbehavior if signing up required a lump of cash security placed in escrow in the MRI bank account, forfeit for such activity, and perhaps some personal identification provided & verified.
Server problems thread beginning at https://mersenneforum.org/showpost.p...postcount=2419
Massive new mover on Top Producers list (glitch!) - patched thread https://mersenneforum.org/showthread.php?t=27391&page=2
Quote:
Originally Posted by xx005fs View Post
User "for you" just submitted 3 TF assignments.
I doubt he had assignments, because the variety of bit levels range from trivially low to absurdly unachievably high. And https://www.mersenne.org/manual_gpu_assignment/ does not go above 85 bits. He pretty clearly submitted forged result records. See the attachment at https://mersenneforum.org/showthread...829#post598829

Last fiddled with by kriesel on 2022-01-26 at 21:08
kriesel is offline   Reply With Quote
Old 2022-02-04, 18:58   #219
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3·5·431 Posts
Default

I wonder how one does TF 21 to 280 in a single report, as by Roland Clarkson for M217656493, M217652731 and others. And why not from 270 to 280, since it had already been done to 270.

Last fiddled with by kriesel on 2022-02-04 at 19:12
kriesel is offline   Reply With Quote
Old 2022-02-04, 20:47   #220
S485122
 
S485122's Avatar
 
"Jacob"
Sep 2006
Brussels, Belgium

5·192 Posts
Default

Quote:
Originally Posted by kriesel View Post
...
And why not from 270 to 280, since it had already been done to 270.
Probably because factoring from 2^28 to 2^70 is less than 0,1% of factoring from 2^28 to 2^80 so why bother looking how much trial factoring has already be done for the exponent. Once you start wanting to do a few extra bits, starting from 0 doesn't make a lot of difference.
The user has a good record with LL and PRP tests and a reasonable rate of success with TF and a very good one rate with P-1.

Last fiddled with by S485122 on 2022-02-04 at 20:49 Reason: trying to be clear
S485122 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Results ET_ Operazione Doppi Mersennes 627 2021-01-31 18:06
Where have all the TF results gone?... lycorn PrimeNet 22 2017-10-02 02:40
PGS Results danaj Prime Gap Searches 0 2017-08-14 18:35
CPU Results last 24 hrs Unregistered Information & Answers 3 2010-07-26 00:49
0x results... Mike PrimeNet 11 2004-05-23 12:55

All times are UTC. The time now is 05:13.


Thu May 19 05:13:33 UTC 2022 up 35 days, 3:14, 0 users, load averages: 0.90, 1.43, 1.48

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, 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