Strategic Double Clicking

[Madpoo]

New list... the lowest unassigned exponent from each of these machines. I've started to think it's a good practice to just pick one more exponent from the machine at a time to test and see how it goes. Some systems are so obviously bad it's probably fine to mention all of them at once, but I'll generate those separately.

Exponent	Bad	Good	Unk	Sus	Solo	Mis	worktodo
44323283	1	0	5	0	5	0	DoubleCheck=44323283,72,1
44554957	1	0	2	0	2	0	DoubleCheck=44554957,72,1
44999533	14	6	1	0	1	0	DoubleCheck=44999533,72,1
45024641	2	0	1	0	1	0	DoubleCheck=45024641,72,1
45446179	1	0	1	0	1	0	DoubleCheck=45446179,72,1
45474857	1	0	1	0	1	0	DoubleCheck=45474857,72,1
45513541	1	0	4	0	4	0	DoubleCheck=45513541,72,1
45782263	1	0	1	0	1	0	DoubleCheck=45782263,72,1
46701197	5	2	1	0	1	0	DoubleCheck=46701197,72,1
49760719	1	0	5	0	5	0	DoubleCheck=49760719,73,1
50017559	118	41	39	0	34	5	DoubleCheck=50017559,73,1
55737653	1	0	3	0	3	0	DoubleCheck=55737653,74,1
56292121	7	1	2	0	2	0	DoubleCheck=56292121,73,1
59080517	1	0	7	0	7	0	DoubleCheck=59080517,73,1

[Madpoo]

Originally Posted by kladner

I am glad to see you have this one, Aaron.
https://www.mersenne.org/report_expo...2775709&full=1

Yeah, I've generally been doing good at picking up new triple-checks fairly quick. Time will tell if I can manage to catch up and then stay on top of them. Right now I've got all but a handful (maybe 10) assigned to me now, and the work I already have assigned should keep me busy for the next couple weeks.

Some days it's kind of crazy how many new mismatches come in, and other days there are hardly any.

Of course, the lists I put up here of suspected bad first-time runs feeds into that. If I do the analysis right, that means more mismatches...LOL

[Prime95]

I took these:

44323283	1	0	5	0	5	0	DoubleCheck=44323283,72,1
44554957	1	0	2	0	2	0	DoubleCheck=44554957,72,1
44999533	14	6	1	0	1	0	DoubleCheck=44999533,72,1
45024641	2	0	1	0	1	0	DoubleCheck=45024641,72,1
45446179	1	0	1	0	1	0	DoubleCheck=45446179,72,1
45474857	1	0	1	0	1	0	DoubleCheck=45474857,72,1
45513541	1	0	4	0	4	0	DoubleCheck=45513541,72,1
45782263	1	0	1	0	1	0	DoubleCheck=45782263,72,1

[kladner]

Originally Posted by Madpoo

Yeah, I've generally been doing good at picking up new triple-checks fairly quick. Time will tell if I can manage to catch up and then stay on top of them.

I am fairly confident of my result as it was preceded by and followed by quite a few matches. My hardware setup has been unchanged for some time, too.

[Mark Rose]

I've taken the remaining double checks from this thread. I should have them done in a month.

Exponents needing a triple check:

DoubleCheck=46701197,72,1
DoubleCheck=61739537,74,1
DoubleCheck=62206849,74,1
DoubleCheck=62481227,74,1
DoubleCheck=69407521,74,1
DoubleCheck=71627189,75,1
DoubleCheck=74590297,75,1
DoubleCheck=74658247,75,1
DoubleCheck=74771029,75,1
DoubleCheck=74991149,75,1
DoubleCheck=75194293,75,1
DoubleCheck=75216997,75,1
DoubleCheck=78187933,75,1
DoubleCheck=79748423,75,1

[kladner]

Originally Posted by kladner

I am fairly confident of my result as it was preceded by and followed by quite a few matches. My hardware setup has been unchanged for some time, too.

Pride goes before a fall.My result was wrong.
https://www.mersenne.org/report_expo...2775709&full=1

[nofaith628]

I've taken the following triple checks:

DoubleCheck=74771029,75,1
DoubleCheck=74991149,75,1
DoubleCheck=75194293,75,1
DoubleCheck=75216997,75,1
DoubleCheck=78187933,75,1
DoubleCheck=79748423,75,1

[Jayder]

Taking:

Originally Posted by Mark Rose

DoubleCheck=46701197,72,1
DoubleCheck=61739537,74,1
DoubleCheck=62206849,74,1
DoubleCheck=62481227,74,1
DoubleCheck=69407521,74,1
DoubleCheck=74590297,75,1
DoubleCheck=74658247,75,1

Leaving the one I did the double-check on:

DoubleCheck=71627189,75,1

[Dubslow]

Originally Posted by kladner

Pride goes before a fall.My result was wrong.
https://www.mersenne.org/report_expo...2775709&full=1

Process before results. If in retrospect, everything you knew before the result came in *did* allow good confidence (from a probability perspective), then it wasn't wrongful confidence, just random. Have there been subsequent bad tests from the machine?

[kladner]

Originally Posted by Dubslow

Process before results. If in retrospect, everything you knew before the result came in *did* allow good confidence (from a probability perspective), then it wasn't wrongful confidence, just random. Have there been subsequent bad tests from the machine?

It has had a clean run of many** DCs, since. In general, I do trust its operation. The bad one surprised me.

**It does a current range DC in about 25 hours in its current setup: i7-6700K @ 4.3 GHz full time, DDR4-2666 RAM running at 3200. Iterations hover around 2.2 ms. At its best it holds about 2.17 ms.

[Mark Rose]

Might be worth backing off the timings slightly then.