Strategic Double Clicking - Page 136

Madpoo · 2017-06-06, 17:27

Quote:

Originally Posted by Prime95

I think it would be a good idea to strategically double-check at least half of this users exponents. Based on the data that comes back, we might then double-check the other half.

I've got a query already setup that will pull the smallest unassigned exponent for a particular year/month of any matching criteria (in this case a specific CPU). It comes in handy for spot checking...

Here's that list, in chronological order.

Code:

Exponent	Bad	Good	Unk	worktodo
43191191	0	14	7	DoubleCheck=43191191,72,1
45537641	0	2	3	DoubleCheck=45537641,72,1
41953489	0	4	7	DoubleCheck=41953489,72,1
46388297	0	1	1	DoubleCheck=46388297,72,1
44405707	0	1	4	DoubleCheck=44405707,72,1
43518661	0	7	2	DoubleCheck=43518661,72,1
42316943	0	2	6	DoubleCheck=42316943,72,1
41721523	0	4	3	DoubleCheck=41721523,72,1
47556689	0	0	4	DoubleCheck=47556689,72,1
44258833	0	2	1	DoubleCheck=44258833,72,1
44169109	0	2	6	DoubleCheck=44169109,72,1
45602677	0	0	4	DoubleCheck=45602677,72,1
45414253	0	1	7	DoubleCheck=45414253,72,1
45414401	0	4	2	DoubleCheck=45414401,72,1
44095913	0	5	3	DoubleCheck=44095913,72,1
43849327	0	3	6	DoubleCheck=43849327,72,1
40289297	0	8	3	DoubleCheck=40289297,72,1
45591499	0	8	1	DoubleCheck=45591499,72,1
46157549	0	2	3	DoubleCheck=46157549,72,1
50928337	0	1	14	DoubleCheck=50928337,73,1
43220833	0	1	8	DoubleCheck=43220833,72,1
50752483	0	1	7	DoubleCheck=50752483,73,1
48738539	0	0	10	DoubleCheck=48738539,72,1
43454707	0	1	14	DoubleCheck=43454707,72,1
43482251	0	4	10	DoubleCheck=43482251,72,1
45921341	0	1	11	DoubleCheck=45921341,72,1
50505359	0	2	12	DoubleCheck=50505359,73,1
50488463	0	2	17	DoubleCheck=50488463,73,1
50737123	0	0	8	DoubleCheck=50737123,73,1
50513509	0	0	24	DoubleCheck=50513509,73,1
47880529	0	1	2	DoubleCheck=47880529,72,1
53644573	0	1	22	DoubleCheck=53644573,73,1
51220769	1	3	17	DoubleCheck=51220769,73,1
50419043	0	2	31	DoubleCheck=50419043,73,1
50051017	1	2	17	DoubleCheck=50051017,73,1
47304703	0	3	27	DoubleCheck=47304703,72,1
51781361	1	1	14	DoubleCheck=51781361,73,1
51466607	1	4	17	DoubleCheck=51466607,73,1
55061443	0	0	23	DoubleCheck=55061443,73,1
55109911	0	0	9	DoubleCheck=55109911,73,1
55329581	0	0	17	DoubleCheck=55329581,73,1
54430973	0	0	18	DoubleCheck=54430973,73,1
51223423	0	0	24	DoubleCheck=51223423,73,1
60475829	0	0	6	DoubleCheck=60475829,74,1
51914539	0	0	22	DoubleCheck=51914539,73,1
51916897	0	2	14	DoubleCheck=51916897,73,1
58945283	0	1	17	DoubleCheck=58945283,73,1
56899343	0	0	17	DoubleCheck=56899343,73,1
57421031	0	0	28	DoubleCheck=57421031,73,1
60002291	1	2	19	DoubleCheck=60002291,74,1
62647661	0	1	13	DoubleCheck=62647661,74,1
58907749	1	5	11	DoubleCheck=58907749,73,1
61500731	7	6	11	DoubleCheck=61500731,74,1
55897211	5	9	19	DoubleCheck=55897211,73,1
61222081	6	4	7	DoubleCheck=61222081,74,1
60939001	0	2	21	DoubleCheck=60939001,74,1
61037629	2	2	16	DoubleCheck=61037629,74,1
63181423	1	2	13	DoubleCheck=63181423,74,1
63530827	0	0	15	DoubleCheck=63530827,74,1
63683693	0	0	14	DoubleCheck=63683693,74,1
63683929	0	0	10	DoubleCheck=63683929,74,1
63683621	0	0	14	DoubleCheck=63683621,74,1
65062643	0	0	17	DoubleCheck=65062643,74,1
64681307	0	0	14	DoubleCheck=64681307,74,1
66229081	0	0	19	DoubleCheck=66229081,75,1
66698507	2	2	8	DoubleCheck=66698507,75,1
72629587	2	2	4	DoubleCheck=72629587,75,1
72629749	1	2	1	DoubleCheck=72629749,75,1
67854097	0	0	4	DoubleCheck=67854097,75,1
75059249	0	0	4	DoubleCheck=75059249,75,1
67854053	0	0	12	DoubleCheck=67854053,75,1
68360059	0	0	4	DoubleCheck=68360059,75,1
68699989	0	0	8	DoubleCheck=68699989,75,1
72893761	0	1	16	DoubleCheck=72893761,75,1
73829731	0	0	6	DoubleCheck=73829731,75,1
73656907	0	0	7	DoubleCheck=73656907,75,1
77486791	0	8	3	DoubleCheck=77486791,75,1
77486441	0	3	4	DoubleCheck=77486441,75,1
78212009	0	0	4	DoubleCheck=78212009,75,1
78431653	0	1	3	DoubleCheck=78431653,75,1

Madpoo · 2017-06-06, 21:25

Quote:

Originally Posted by Prime95

I think it would be a good idea to strategically double-check at least half of this users exponents. Based on the data that comes back, we might then double-check the other half.

By the way, I just thought of this and it would be a little bit of a "cheat" but we could "persuade" the general masses to do strategic double-checking by marking potentially wrong results as "suspect" and make them available for first-time checks.

When a computer itself encounters certain errors and turns in a suspect result, that's exactly what happens anyway... in this case we'd be doing what *should* have happened and mark it as suspect so it's checked again much sooner.

I think I saw that a "suspect" result has a 50/50 chance of being correct, if I remember what I found when analyzing the error codes... I think the lists we're pulling of "probably bad" is actually better than that at spotting bad results.

Anyway, just a thought, although it might mess up some reporting where we've already crossed several milestones for first-time checks, but suddenly marking a bunch as suspect and available for a "new first time check" would gum things up for a bit.

GP2 · 2017-06-06, 21:40

Quote:

Originally Posted by Madpoo

By the way, I just thought of this and it would be a little bit of a "cheat" but we could "persuade" the general masses to do strategic double-checking by marking potentially wrong results as "suspect" and make them available for first-time checks.

But the "general masses" will just churn and expire those strategic exponents. We want them to actually get done, and in a timely fashion. Hence this thread.

ewmayer · 2017-06-07, 02:14

Quote:

Originally Posted by GP2

The triple checks on exponents 53648893, 53648981, 53648423, 53647547 will finish within a few hours, in that order. The first one is already at 51M and the slowest is almost at 49M. All interim residues are still matching.

I see the first 3 matched my Mlucas-on-overheating-Ryzen-system results - what is ETA on the fourth?

GP2 · 2017-06-07, 04:04

Quote:

Originally Posted by ewmayer

I see the first 3 matched my Mlucas-on-overheating-Ryzen-system results - what is ETA on the fourth?

As of right this very instant, it's 99.72% complete... so in about 40 minutes from now.

Madpoo · 2017-06-07, 04:37

Quote:

Originally Posted by GP2

But the "general masses" will just churn and expire those strategic exponents. We want them to actually get done, and in a timely fashion. Hence this thread.

Most of them would be sub category zero (if that's a thing...)

ewmayer · 2017-06-07, 21:15

Quote:

Originally Posted by GP2

As of right this very instant, it's 99.72% complete... so in about 40 minutes from now.

And, I see we match on that one as well. Thanks for jumping on these - big relief to know I haven't been wasting all those Ryzen cycles David Stanfill has generously gifted me with. (Not even in the subpar-performance sense, as on Ryzen I get ~90% the throughput of mprime, avx2-builds for both.)

Madpoo · 2017-06-13, 20:26

Here's a short list... this machine has been trending towards the "bad" side of the equation, now with overall stats of 20 bad, 21 good.

Below is a list of the smallest available exponent for any month where the bad>=good (and it might mean no bad/no good). It'll help fill in some unknowns about when the system started to rot... seems like around late 2010 but there are gaps there where no DC's have been done yet.

Code:

Exponent	Bad	Good	Unk	worktodo
41654687	0	0	2	DoubleCheck=41654687,72,1
44526941	0	0	1	DoubleCheck=44526941,72,1
46594763	0	0	7	DoubleCheck=46594763,72,1
46708141	2	0	2	DoubleCheck=46708141,72,1
47521781	0	0	4	DoubleCheck=47521781,72,1
50613733	1	1	3	DoubleCheck=50613733,73,1
50896471	0	0	4	DoubleCheck=50896471,73,1
52944533	1	1	2	DoubleCheck=52944533,73,1
53674261	0	0	4	DoubleCheck=53674261,73,1
55885723	0	0	4	DoubleCheck=55885723,73,1

Madpoo · 2017-06-14, 15:27

I don't know if it's just me, but I found one of the user names in this exponent report to be hilarious in its honesty:
M46205657

That's the 2nd bad result from I've seen from it (it doesn't have any more outstanding).

storm5510 · 2017-06-19, 01:50

Quote:

Originally Posted by Madpoo

I don't know if it's just me, but I found one of the user names in this exponent report to be hilarious in its honesty:
M46205657

That's the 2nd bad result from I've seen from it (it doesn't have any more outstanding).

I see where you ran another DC on June 14. There must be something 'flaky' going on with his hardware. Of course, this is just a guess on my part.

I suppose a lot of what he/she ran will need to be checked...

Madpoo · 2017-06-19, 15:25

Quote:

Originally Posted by storm5510

I see where you ran another DC on June 14. There must be something 'flaky' going on with his hardware. Of course, this is just a guess on my part.

I suppose a lot of what he/she ran will need to be checked...

Yeah, I did a spot check of another one but it matched, so maybe I just hit a few bad ones as a result of my triple-checking.

I may check it out again or just do another roundup of "maybe bad, but let's test one more and see" kind of work.

2017-06-13, 20:26	#1493
Madpoo Serpentine Vermin Jar Jul 2014 6323₈ Posts	Here's a short list... this machine has been trending towards the "bad" side of the equation, now with overall stats of 20 bad, 21 good. Below is a list of the smallest available exponent for any month where the bad>=good (and it might mean no bad/no good). It'll help fill in some unknowns about when the system started to rot... seems like around late 2010 but there are gaps there where no DC's have been done yet. Code: Exponent Bad Good Unk worktodo 41654687 0 0 2 DoubleCheck=41654687,72,1 44526941 0 0 1 DoubleCheck=44526941,72,1 46594763 0 0 7 DoubleCheck=46594763,72,1 46708141 2 0 2 DoubleCheck=46708141,72,1 47521781 0 0 4 DoubleCheck=47521781,72,1 50613733 1 1 3 DoubleCheck=50613733,73,1 50896471 0 0 4 DoubleCheck=50896471,73,1 52944533 1 1 2 DoubleCheck=52944533,73,1 53674261 0 0 4 DoubleCheck=53674261,73,1 55885723 0 0 4 DoubleCheck=55885723,73,1

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Double-Double Arithmetic	Mysticial	Software	50	2017-10-30 19:16
Clicking an exponent leads to 404 page	marigonzes	Information & Answers	2	2017-02-14 16:56
x.265 half the size, double the computation; so if you double again? 1/4th?	jasong	jasong	7	2015-08-17 10:56
What about double-checking TF/P-1?	137ben	PrimeNet	6	2012-03-13 04:01
Double the area, Double the volume.	Uncwilly	Puzzles	8	2006-07-03 16:02

2017-06-14, 15:27	#1494
Madpoo Serpentine Vermin Jar Jul 2014 CD3₁₆ Posts	I don't know if it's just me, but I found one of the user names in this exponent report to be hilarious in its honesty: M46205657 That's the 2nd bad result from I've seen from it (it doesn't have any more outstanding).