mfaktc: a CUDA program for Mersenne prefactoring

[sonjohan]

Originally Posted by kjaget

Once we get automated Primenet interaction it won't be a big deal.

I so like the manual upload of results. I like to see the extreme jump in stats.

Also, with automated Primenet interaction, it's impossible to force TF to 72, which I'm currently doing for all exponents I received.

[kladner]

Originally Posted by Dubslow

<SNIP>
@kladner: 'start' is not its own executable, which is (AFAIK) what shortcuts can recognize. cmd.exe is (obviously) its own executable.

Did you ever work anything out on this front?

And a related question, why do you avoid batch files? Something similar to what James Heinrich said -start without cmd in a batch- is what I'm using now. It does all the affinity/priority stuff as before, but now passes Ctl-C to mfaktc.

Of course it's your choice, to bat or not to bat. I'm just curious as to why you don't like them.

[oswald]

Originally Posted by kladner

Of course it's your choice, to bat or not to bat. I'm just curious as to why you don't like them.

The Riddler will only riddle The Batman.

Yes, if you are going to type it more than a couple of times, script or batch it. I must admit I'm an Algorithm Junky. The beauty of it when you make it work is a joy into itself. The challenge of debug. Oops, sorry. I sound like Monty Python.

Anyway, I'd like to know the outcome as well.

[Dubslow]

I was trying to see if it was possible in the shortcut, with its greater simplicity (also, while not really relevant here, for the 'average' user, 'batch file' will put them off completely while 'shortcut' is less terrifying). Also, I only have one instance, unlike the vast majority of people here. If I get unlazy I might re-implement it.

[bcp19]

Originally Posted by Uncwilly

(51 * 50 * 49 * 48) / 4! to 1
Your pocket 8 digit calculator can do that one. That is a super trivial calculation.

Originally Posted by LaurV

This post is good for xyzzy's top. Sure that program was not doing something else too? There are only 249900 of them, and even a 8-bit Sinclair could spit them out in a blink, or use a matrix (needle) printer to print them in less then 12 hours...

1) Not a math major
2) you're forgetting I am not a major programmer
3) program was running in basic
4) misspoke... program was looking for the MINIMUM number of tickets it would take to get ALL possible 4 number combos; IE: 1,2,3,4,5,6 has 1,2,3,4 1,2,3,5 1,2,3,6 1,2,4,5 1,2,4,6 1,2,5,6 1,3,4,5 1,3,4,6 1,2,5,6 2,3,4,5 2,3,4,6 2,3,5,6 2,4,5,6 3,4,5,6, so that is 1 card.

6 nested for next loops 1 to 46, 2 to 47, 3 to 48, 4 to 49, 5 to 50 and 6 to 51 = 9,474,296,896 iterations of the calculation process. Don't ask me for the program, this was around 20 years ago and I am going from memory with what I have here. If I am remembering correctly, it ran through this more than once, as on the first run through it would only write to the 'array' if there were no duplicate instances; IE: 1,2,3,4-6,x,x+1 to 51 would be bypassed after 1,2,3,4,5,6 was processed, seems like it looked for all 14, then for 10/14, then 6/14 and then just spit out the remaining unchecked 'array' items as it was easier to use human calculations than continue deeper into the computer calculations.

[ckdo]

Originally Posted by ckdo

Both of you are, of course, completely wrong in case bcp19 didn't talk about a "4 out of 51" lottery in the first place.

Originally Posted by bcp19

1,2,3,4,5,6 [...] is 1 card.

QED.

[Dubslow]

Originally Posted by oswald

OK, seems I misunderstood.

I made a shortcut by right clicking on mfaktc-win-64.exe and picking sendto desktop (shortcut).

Then I right clicked on the shortcut and picked properties.

Target box had:

"C:\Program Files\mfaktc\mfaktc-win-64.exe"

The options were the added:

"/low /affinity C mfaktc-win-64.exe -v 1"

The Target box now has:

"C:\Program Files\mfaktc\mfaktc-win-64.exe" "/low /affinity C mfaktc-win-64.exe -v 1"

Note that quotes pairs are needed for the program .exe. Quote pairs for the program command line options seem to be optional.

Double click on the shortcut and the programs starts.

If I'm bugging you or really misunderstand, just say so, I will stop.

Whatever I was doing then, it wasn't right. I just tried this again, with opposite results: ^C is passed to mfaktc. Unfortunately, the options aren't recognized and the affinity isn't set properly (not set at all).

So that still leaves me clueless.

[LaurV]

Originally Posted by bcp19

1) 1,2,3,4,5,6 .... is 1 card.

Originally Posted by ckdo

QED.

They are much less in this case, and their number is easier to compute, and they are easy to "spread" (how do you call this process in English? is it "spread"? like writing down all permutations of some set, or all coefficients of some polynomial multiplication, or some sum/series formula?) with a simple algorithm if you want them printed. It will take even less time then I said initially.

Hint: you have maximum 249k of them with the trivial cover: each 1234 is covered by a 1234xy with a random x,y. From here you start cutting. As C(6,4)=15, the theoretical minimum is 16660, assuming you can cover it in a disjunctive manner. Can you?

[bcp19]

Originally Posted by LaurV

They are much less in this case, and their number is easier to compute, and they are easy to "spread" (how do you call this process in English? is it "spread"? like writing down all permutations of some set, or all coefficients of some polynomial multiplication, or some sum/series formula?) with a simple algorithm if you want them printed. It will take even less time then I said initially.

Hint: you have maximum 249k of them with the trivial cover: each 1234 is covered by a 1234xy with a random x,y. From here you start cutting. As C(6,4)=15, the theoretical minimum is 16660, assuming you can cover it in a disjunctive manner. Can you?

As mentioned before, I am not a math major, nor a major programmer. When I wrote the program I did not take into consideration the 1,2,3,4,x,y as you state, so the program ran through all 9.4 billion iterations.

EDIT: One of the other things to consider, the 8088 was a 5MHz processor and probably had at most 64MB memory, which means the program had to do it's array through disc access, which would considerable slow the process, in addition to the slowdown already caused by using basic.

[kladner]

Originally Posted by Dubslow

I was trying to see if it was possible in the shortcut, with its greater simplicity (also, while not really relevant here, for the 'average' user, 'batch file' will put them off completely while 'shortcut' is less terrifying). Also, I only have one instance, unlike the vast majority of people here. If I get unlazy I might re-implement it.

OK. I get your point. I guess I've dealt with batches for so long it comes naturally, at least for simple things.

[kladner]

Originally Posted by bcp19

EDIT: One of the other things to consider, the 8088 was a 5MHz processor and probably had at most 64MB memory,

That would be KB, right?