r***** saved file

[bayanne]

Normally the saved files for Prime95 start with p***** and q*****.

However I am running a large exponent at present and it is just saving one r***** file.

Is this something new, or is it because the exponent is large. I am running 24.14

[cheesehead]

In order not to lose anything if a glitch interrupts the savefile operation, Prime95 doesn't delete either of the two old files until after it has successfully written the newest one.

It goes:

Write r*****
If no error, then delete p*****
If no error, then rename q***** to p*****
If no error, then rename r***** to q*****
Resume iterations

So, if you're watching the directory, the r***** file will exist only a few seconds before it is renamed to q*****.

[Prime95]

So if the r****** file is not being deleted there must be some problem deleting the p****** file. Try deleting the p****** file by hand.

[bayanne]

err .... what p***** file

Well, actually there is a p**** file but it is 1 kb in size, and dated from 22:45 05/11/2006, where as the r***** file is from 18.31 07/11/2006

[bayanne]

I did delete that tiny p***** file, which could only be done when Prime95 was not running.

On a restart it renamed the r***** file to a p***** file.

Not sure why the p***** file was still hanging around, but sorted now.

Worth adding to FAQ ......

[cheesehead]

Originally Posted by cheesehead

Write r*****
If no error, then delete p*****
If no error, then rename q***** to p*****
If no error, then rename r***** to q*****
Resume iterations

So, if you're watching the directory, the r***** file will exist only a few seconds before it is renamed to q*****.

Oops, I interchanged the roles of p***** and q*****.

Make that:

Write r*****
If no error, then delete q*****
If no error, then rename p***** to q*****
If no error, then rename r***** to p*****
Resume iterations

So, if you're watching the directory, the r***** file will, normally, exist only a few seconds before it is renamed to p*****.

Originally Posted by bayanne

Well, actually there is a p**** file but it is 1 kb in size, and dated from 22:45 05/11/2006, where as the r***** file is from 18.31 07/11/2006

Prime95 trial factoring also names its tiny save files as p*****. Perhaps a TF run was interrupted before it finished, else it would've been deleted upon completion.

On single-CPU, non-HT systems Prime95 wouldn't be running both TF and LL simultaneously. On dual-CPU, Prime95 attaches ".001" to the end of all save files generated on CPU 1, while using no suffix on those from CPU 0. On HT, I dunno what George does to avoid conflicts (just in case someone would try TF and LL on the same exponent simultaneously on the same CPU).