M19 is not prime...
 

Well ok obviously it is. But for some reason one of my copies of Prime95 says it isn't about 10% of the time.

I was running a torture test with w23.4 on a P4 with HT enabled. While the torture test was running I decided to play around with another instance of Prime. I tried LL testing several known Mersenne primes for fun. M216091, M132049, M110503 all came out fine. Then I entered M19 and got this:

UID: ....../C6CA6CD00, M19 is not prime. Res64: FFFFFFFFFFFFFFFF. WZ1: 00260028,16,00000000

I wasn't exactly expecting that. I tried testing it again and it came out as prime several times. Then I tested M19937 followed by M19 again:

UID: ....../C6CA6CD00, M19937 is prime! WZ1: 51779BC2
UID: ....../C6CA6CD00, M19 is not prime. Res64: FFFFFFFFFFFFFFFF. WZ1: 00260028,16,00000000

Obviously this isn't really a problem since Prime isn't meant to test numbers this small. I don't think it could be a hardware issue on my system as the TT is still chugging away. Anyone else ever seen the before? Is this just due to SSE2 rounding, or there something else that could cause the residual to occasionally be all ones?

I get M19 is not prime too. So it's not just your machine.

I believe George has mentioned before that at very low exponents, the code doesn't always work correctly.

http://arstechnica.infopop.net/OpenTopic/page?a=tpc&s=50009562&f=122097561&m=3300925355&r=1820975755#1820975755

And very large exponents too.

[quote="TTn"]And very large exponents too.[/quote]

Based on what?

I think it has been explained several times to you that the machine where the supposed M40 was running on had hardware problems!

I think it's about time that you should stop making thes unfounded statements.

[quote="TTn"]And very large exponents too.[/quote]

Do you mean > 79.3M?
If that is the case, it is not a question of "not working correctly". It´s just that the current version of the program is not meant to test exponents above that limit. With today´s technology, it would be useless anyway... 8). We have already enough exponents to keep us all busy for the next 10 or 15 years :)

[quote]And very large exponents too.[/quote]

Just couldn't resist could you? Keep beating that long dead horse.

[quote="smh"][quote="TTn"]And very large exponents too.[/quote]

Based on what?

I think it has been explained several times to you that the machine where the supposed M40 was running on had hardware problems!

I think it's about time that you should stop making thes unfounded statements.[/quote]

It's way past the time. But the person in question likes being a jerk, so what can you expect. Sad really...

Actually its in the source.
You guys are the jerks.
Should we have a poll to kick you guys out!
Or wait a poll to stop GIMPS because you guys were jerks..
Ya ya thats the ticket!

I did not mention M40, but you retards automatically thought that is what I was saying.

We love you TTn inspite of the name calling. :mrgreen:
[quote="TTn"]You guys are the jerks. [/quote]