New P-1 CPU credit formula

[jrk]

If this code is anything to go by (posted by George some weeks ago):

// P-1 - no factor
// timing / 86400 * (1.5 * B1 + 0.05 * B2)
// P-1 - factor found in stage 2
// timing / 86400 * (1.5 * B1 + 0.05 * B2)
// P-1 - factor found in stage 1, B2 = 0
// timing / 86400 * (1.5 * B1)
function credit_cpu_PM1_factoring( $exponent, $fftlen, $B1, $B2 )
{
    $timing = credit_get_FFT_timing( $exponent, $fftlen );
    return ( $timing * ( 1.5 * $B1 + 0.05 * $B2) / 86400.0 );
}

Then if stage 2 isn't performed, the value of B2 given here should be 0 anyway.

BUT now that the function is calculating B2-B1 then it must check for the condition B2==0, or else you'll actually end up being under-credited for not doing B2...

George?

[MatWur-S530113]

@ jrk: if you look in the 'factoring limits' page ('result queries') you'll see that B2 always is set to B1 if no stage 2 is done, the same happens if you look up such a result in your own results page of your acc.

I agree with Jacob and garo, the boundaries of already done P-1 work should be subtractet in the formula. But if Old_B1>B1 then (B1 - Old_B1) should be set to 0 or you get CPU-Credit subtractet. Same if Old_B2>B2.

best regards,

Matthias

[Prime95]

Originally Posted by jrk

George?

Actual PHP code is:

	if ( $B1 >= $B2 )
		return ( $timing * ( 1.45 * $B1 ) / 86400.0 );
	else
		return ( $timing * ( 1.45 * $B1 + 0.079 * ($B2 - $B1) ) / 86400.0 );

[jrk]

Ok thanks

[jrk]

Originally Posted by MatWur-S530113

I agree with Jacob and garo, the boundaries of already done P-1 work should be subtractet in the formula. But if Old_B1>B1 then (B1 - Old_B1) should be set to 0 or you get CPU-Credit subtractet. Same if Old_B2>B2.

I think it will be trickier than that.

If Old_B1 = Old_B2 = 100e3, no stage 2

Then someone does B1 = B2 = 500e3, no stage 2. This should be credited for the difference in B1 values. That's easy (I think?).

But what if, instead:

Old_B1 = 100e3, Old_B2 = 2e6, this time stage 2 was done before

And someone does B1 = B2 = 500e3, no stage 2. Should this receive the same credit for the B1 difference as before? Arguably it is not as useful.

And if they do B1 = 500e3, B2=2e6, no difference in B2 but the new stage 2 work should receive some credit since the B1 was extended, and that is more useful than just extending B1 without redoing the stage 2.

Things to ponder.

[jrk]

I typo'd the B2 numbers in the post above. Fixed now.

[S485122]

Originally Posted by jrk

I think it will be trickier than that.

If Old_B1 = Old_B2 = 100e3, no stage 2

...

Things to ponder.

If the resulting credit is negative, award no credit. Otherwise the formula will get to complex. Any way suppose P-1 done to 3 000 000 ; 3 000 000, then another attempt with 1 000 000 ; 30 000 000 would earn negative credit ... but is that second P-1 factoring attempt usefull ? If one redoes P-1 it must be worth it.

Originally Posted by garo

I think your reasoning is incorrect. If the correct B1 coefficient should be 1.45, then in your equivalent formula, if no stage 2 is done the B2 term simply adds to the B1 term to result in the correct amount of credit.

Incorrect ? I just said the same thing as you do. I perhaps forgot to mention that I assumed that when no stage 2 is done, B2 is considered equal to B1.

Anyway the new PHP code is clearer :-)

Jacob

[jrk]

Originally Posted by S485122

If the resulting credit is negative, award no credit.

I made no mention of negative credit. However you decide to calculate it, a negative credit would clearly be a bug in logic.

Any way suppose P-1 done to 3 000 000 ; 3 000 000, then another attempt with 1 000 000 ; 30 000 000 would earn negative credit

Only if the logic is bad.

... but is that second P-1 factoring attempt usefull ? If one redoes P-1 it must be worth it.

Well, if the point is to deduct P-1 that's already been completed... then whatever you are going to do about that will result in not crediting work that isn't useful.

[MatWur-S530113]

Hello Jayson,

Originally Posted by jrk

I think it will be trickier than that.

If Old_B1 = Old_B2 = 100e3, no stage 2

Then someone does B1 = B2 = 500e3, no stage 2. This should be credited for the difference in B1 values. That's easy (I think?).

But what if, instead:

Old_B1 = 100e3, Old_B2 = 2e6, this time stage 2 was done before

And someone does B1 = B2 = 500e3, no stage 2. Should this receive the same credit for the B1 difference as before? Arguably it is not as useful.

Yes, I think one should get the same credit as before. You think about the 'loose' of information for stage 2 in the range [500e3,2e6]. But IMHO this 'loose' is very low compared to the information of the new stage 1. As Jacob said, a redoing of a P-1 must be worth it. And Jacob and I think more about redoing a P-1 by the same person, with a save-file from the old P-1. You are thinking more about a redoing of another person, who has to do the complete work again. I agree that this is a little bit more complicate than I thought first.

Originally Posted by jrk

And if they do B1 = 500e3, B2=2e6, no difference in B2 but the new stage 2 work should receive some credit since the B1 was extended, and that is more useful than just extending B1 without redoing the stage 2.

Things to ponder.

Stage 2 seems to be the problem, yes. In opposite to stage 1, stage 2 can not be expanded. It must be done again every time completely for its own. But the only thing I wanted to say was, that the CPU-Credit shouldn't become negative.

Atm I think about this problem:
Should a person who is redoing a P-1 completely new get (only) the same CPU-Credit for this work as a person who is redoing this P-1 with an already stored save-file?

best regards,

Matthias

[Vidmar]

Originally Posted by petrw1

The previous points showed up (in the total only - not detail lines) ... however ... a result that completed only a few hours after this announcement received less than half the credit similar assignments got:

Woody 50756567 NF-PM1 2009-01-27 05:35 49.0 B1=610000, B2=19520000 1.8746 
Woody 50722993 NF-PM1 2009-01-24 23:02 46.8 B1=610000, B2=19520000 4.0079 
Woody 50756449 NF-PM1 2009-01-22 16:20 44.5 B1=610000, B2=19520000 4.0079 
Woody 50756369 NF-PM1 2009-01-20 08:51 42.1 B1=610000, B2=19520000 4.0079

Where are you able to find this data? Since the switch to PrimeNetv5 I've been unable to locate my returned results.

[Vidmar]

Originally Posted by Vidmar

Where are you able to find this data? Since the switch to PrimeNetv5 I've been unable to locate my returned results.

Never mind. Found it. The + expand option needs to be re-thought. I wonder how many others don't realize it's there because they can't see it?