Newer milestone thread

[rudy235]

74 million

DONE!

ATH	5960X	73968971	C	2017-09-19 20:21	5.9	208.5514
	0B24A31B1BAD71__

[Madpoo]

Still waiting on M74095187 now. The history of check-ins from the cpu doing it is a little spotty.

Lately it's only been checking in every 3 days or so, and only doing < 1% per day, if it even progressed at all.

Strange, because it started out okay and even managed to do 18% in a day, but that was a month ago.

Basically, in the past 30 days it's only moved from 68.4% to 80.2% ... avg 0.4% daily.

I'm kind of waiting to see if it checks in again in the next 24 hours since it should be soon, if it sticks to that every 3 days thing. I guess we'll see.

[rudy235]

Madpoo: there is a previous result AE5C73EB430E2D__ for 74095187 from July 2017 by mikeblas that is considered "suspect". Do you know why is that?

[VictordeHolland]

Originally Posted by rudy235

is considered "suspect". Do you know why is that?

Suspect results have an error code when submitted. Basicly the program ran into trouble at some point of the calculation and detected that. Prime95 has pretty decent error-detection and can sometimes do the iterations with errors again (or with a different FFT size). So results that have an error code when reported can still produce the right residue, but we're less confident of the results (which is why it is handed out for an extra LL test immediately).

By the way, if a LL result has zero error code it doesn't automaticly mean no errors occurred. About 1-2% of the LL tests without error codes are still proven wrong when they are DoubleCheck/TripleChecked.

[Madpoo]

Originally Posted by VictordeHolland

Suspect results have an error code when submitted. Basicly the program ran into trouble at some point of the calculation and detected that. Prime95 has pretty decent error-detection and can sometimes do the iterations with errors again (or with a different FFT size). So results that have an error code when reported can still produce the right residue, but we're less confident of the results (which is why it is handed out for an extra LL test immediately).

By the way, if a LL result has zero error code it doesn't automaticly mean no errors occurred. About 1-2% of the LL tests without error codes are still proven wrong when they are DoubleCheck/TripleChecked.

Ditto all that. Last time I checked, results marked suspect had about a 50% chance of being wrong, so it's a great idea to treat those as if they're bad and re-assign them as a first time check again.

As you mentioned, even a "clean" run still has that potential to be bad. Some of the machines out there that generated almost all bad results were coming in as "clean" runs, so even the typical error detection routines (roundoff errors, etc) weren't catching it. Probably just lousy RAM in those cases, not CPU related, if I had to guess.

[rudy235]

Originally Posted by Madpoo

Still waiting on M74095187 now. The history of check-ins from the cpu doing it is a little spotty.

Lately it's only been checking in every 3 days or so, and only doing < 1% per day, if it even progressed at all.
.

This a real joke! It progressed from 80.20% to 80.30% in three days.

If cancer progressed this fast, nobody would die of cancer. (Something else would get them first!)

[science_man_88]

Originally Posted by rudy235

If cancer progressed this fast, nobody would die of cancer. (Something else would get them first!)

really so you think under 8.3 years is enough for something else to kill them first ?

[VBCurtis]

Originally Posted by rudy235

This a real joke! It progressed from 80.20% to 80.30% in three days.
If cancer progressed this fast, nobody would die of cancer. (Something else would get them first!)

Perhaps you should check your math, sir: 0.1% in 3 days is 1% per month is 12% per year. If cancer is diagnosed even at 20% of the way to the end, that's around 7 years to reach 100% of the end. If I were told cancer would get me in 7 years, I'm pretty sure I'd feel like I'd die of cancer.

[rudy235]

Originally Posted by VBCurtis

Perhaps you should check your math, sir: 0.1% in 3 days is 1% per month is 12% per year. If cancer is diagnosed even at 20% of the way to the end, that's around 7 years to reach 100% of the end. If I were told cancer would get me in 7 years, I'm pretty sure I'd feel like I'd die of cancer.

Ohh my G*d. They took me literally.

I was simply making a 'figure of speech". I expect that given the level of the forum participants most would be familiar with a "trope". i.e. a figurative or metaphorical use of a word or expression.

I suspect if I mention that a process is "as slow as watching paint dry" someone would come out and argue that you can't actually notice any change in the appearance of the paint while it dries and thusly there is no value in actually watching.

But yeah it was a lot of fun to read the responses. In the meantime, the exponent it remains at 80.30%

[science_man_88]

Originally Posted by rudy235

Ohh my G*d. They took me literally.

I was simply making a 'figure of speech". I expect that given the level of the forum participants most would be familiar with a "trope". i.e. a figurative or metaphorical use of a word or expression.

I suspect if I mention that a process is "as slow as watching paint dry" someone would come out and argue that you can't actually notice any change in the appearance of the paint while it dries and thusly there is no value in actually watching.

But yeah it was a lot of fun to read the responses. In the meantime, the exponent it remains at 80.30%

they also know their fair share of tripe when they read it.

[VBCurtis]

Originally Posted by rudy235

Ohh my G*d. They took me literally.

I don't find cancer to be a joking matter, so taking it literally seemed better than taking offense.