Running all TF and P-1 at night
 

I see that it's possible to run stage 2 P-1 factoring only at night by putting the string OnlyRunStage2WithMaxMemory=1 at the end of Prime.ini.

I don't think it's possible to run TF and stage 2 P-1 only at night. I really wish it were possible, though. On my computer, both TF and P-1 factoring go about 30% faster when I am on my computer but without many applications open. At night, when I set my computer to turn off the monitor when it's idle for long enough, I get another 50% increase in speed. The gains for LL testing are 10-20%

Would it be possible to add this feature to future versions of Prime95?

TF does not take much RAM. Doing P-1 stage 2 at night is so that large amounts of RAM are available.

[QUOTE=Uncwilly]TF does not take much RAM. Doing P-1 stage 2 at night is so that large amounts of RAM are available.[/QUOTE]
I know that.

Can you tell us the config of your computer? single or dual, P4 or not, amount of RAM, etc. etc. It is unlikely that the speed up for TF is 30% and for LL is 10-20%. There must be some other explanation.

It would be nice if you could run these test again and post exact timings.

[QUOTE=garo]Can you tell us the config of your computer? single or dual, P4 or not, amount of RAM, etc. etc. It is unlikely that the speed up for TF is 30% and for LL is 10-20%. There must be some other explanation.

It would be nice if you could run these test again and post exact timings.[/QUOTE]Single P4 2.4G processor, 512+256MB ram of some sort.

I didn't do it that long since I just wanted to get some times. Times for TF: 9500 milliseconds (per 500 iterations) while using the computer, 8450 with just internet explorer open and not idle, and 8250 while idle. I won't be doing another P-1 test for another month.

Do {(final speed)-(initial speed)}/(initial speed). Note speed is the reciprocal of those numbers. that gave me a 12.4% improvement while not idle and 15.2% improvement idle. LL tests usually give me at most 5.85% and 10.5% increases in improvement, respectively.

This isn't as extreme as I though. It's possible that I made a calculation error earlier. I also ran this test somewhat artificially, so the numbers might be a bit smaller than last time since Prime95 tends to slow down steadily until my computer goes idle again. The effect is still there, but only to the tune of 7% and 5%. Considering the amount of time a P-1 and TF run lasts relatively (only about 10% of the time) it might not be worth coding this into Prime95 and possibly introducing a bug. :rant:

Found part of it. I had only one real-time program running. When I run two, the times goes up to 10208. So the improvement from this to when it's idle is 24.6%

Since TF and LL tests use different parts of the processor, it may be that the programs you are running conflict with TF more. But that depends totally on the programs you are running. it is very possible if you run other programs the effect will be the opposite.

[QUOTE=dominicanpapi82]I don't think it's possible to run TF and stage 2 P-1 only at night. I really wish it were possible, though.[/QUOTE]Do you run the TF as a separate work type? Then you might try something like this in prime.ini (didn't try it myself though):
[QUOTE] Time=1-5/8:30-17:30
WorkPreference=(something)
Time=1-5/0:00-8:30,1-5/17:30-24:00,6-7/0:00-24:00
WorkPreference=(something else)
[/QUOTE]I can't seem to find any documentation on work preference though, but double-check seems to be 4.