Possible extention to the "GPU to 72 Tool" project?

[davieddy]

Originally Posted by Chuck

Last night it gave me 2,533.628 GHz days credit for exponent 48276133 and indicates I found a factor. Gimps says

Manual testing 48276133 NF 2011-11-13 00:06no factor for M48276133 from 2^70 to 2^71 4.9533

Manual testing 48276133 NF 2011-11-13 00:06no factor for M48276133 from 2^69 to 2^70 2.4767


Chuck

I was wondering why the "throughput" was escalating so much.

David

[Mr. P-1]

Originally Posted by Chuck

Last night it gave me 2,533.628 GHz days credit for exponent 48276133 and indicates I found a factor. Gimps says

Manual testing 48276133 NF 2011-11-13 00:06no factor for M48276133 from 2^70 to 2^71 4.9533

Manual testing 48276133 NF 2011-11-13 00:06no factor for M48276133 from 2^69 to 2^70 2.4767


Chuck

M48276133 has the 79-bit factor 581409625692942066907489, almost certainly found by P-1

k=2^4 * 3 * 7^3 * 3413 * 4817 * 22247

That's an easy stage 1 find.

[davieddy]

Originally Posted by Mr. P-1

M48276133 has the 79-bit factor 581409625692942066907489, almost certainly found by P-1

k=2^4 * 3 * 7^3 * 3413 * 4817 * 22247

That's an easy stage 1 find.

I took Chuck's post to convey the wonderment he felt
about receiving 2.5K GHz days, rather than the expected ~2.5

David

[chalsall]

Originally Posted by Chuck

Last night it gave me 2,533.628 GHz days credit for exponent 48276133 and indicates I found a factor.

Whoops!!! Thanks -- my first bug (actually, two)...

Someone else found the factor using P-1 after you'd TFed from 69 to 71, and the script gave both you and them the (miscalculated) credit for the find.

Fixed now....

[KyleAskine]

Are you planning on getting your worker progress page updated to show DC work? Or will that be going somewhere else?

[KyleAskine]

Originally Posted by KyleAskine

Are you planning on getting your worker progress page updated to show DC work? Or will that be going somewhere else?

Whoops, I see that you are showing it. Ignore me!

[lycorn]

I´ve just started using GPU to 72. Pretty smooth, nice and easy. Great work, chalsall! Another nice feature added to help on the GIMPS overall effort.
Now two points I would like to clarify:
I got 4 exponents to P-1 (won´t be doing any TF, at least in the near future, for lack of a suitable GPU); all of them have N/A instead of the hex key in the corresponding worktodo.txt line, which is normal (am I right?...).
When the first exponent finished, the second one started straightaway but the result for the first one wasn´t immediately sent to Primenet Server, as I have Prime95 set to not automatically send results to the server. Some hours later, I sent the result to the server (via Prime95, not the manual pages). The result was accepted, credit correctly given, but the 3rd and 4th exponents in the queue, which had not any work done on them yet, got kicked out of worktodo.txt. No big issue, I rebuilt the lines with the information on my Assignments Report in GPU to 72 tool; now the thing is, next time I send a result to the server the same thing will happen. If instead of 2 exps one has 20 or more, it starts to be a real pain...
I see 2 workarounds for this situation: the worker receives the hex key from GPU to 72 together with the assignment OR the worker reports the completion of the work unit using the Manual Pages of Primenet. The first would be the most practical, of course.
What you think?

Second point:

Originally Posted by chalsall

OK. Some LL Workers might want to take exponents which only HAD P-1 done because they don't have enough memory to do P-1 well. Simple to make the both options (plus "either way") available to requesters.

Is the option to allow people to choose low LL (or DC) tests from GPU to 72 in the works, or has it been abandoned?

[KyleAskine]

Originally Posted by lycorn

I´ve just started using GPU to 72. Pretty smooth, nice and easy. Great work, chalsall! Another nice feature added to help on the GIMPS overall effort.
Now two points I would like to clarify:
I got 4 exponents to P-1 (won´t be doing any TF, at least in the near future, for lack of a suitable GPU); all of them have N/A instead of the hex key in the corresponding worktodo.txt line, which is normal (am I right?...).
When the first exponent finished, the second one started straightaway but the result for the first one wasn´t immediately sent to Primenet Server, as I have Prime95 set to not automatically send results to the server. Some hours later, I sent the result to the server (via Prime95, not the manual pages). The result was accepted, credit correctly given, but the 3rd and 4th exponents in the queue, which had not any work done on them yet, got kicked out of worktodo.txt. No big issue, I rebuilt the lines with the information on my Assignments Report in GPU to 72 tool; now the thing is, next time I send a result to the server the same thing will happen. If instead of 2 exps one has 20 or more, it starts to be a real pain...
I see 2 workarounds for this situation: the worker receives the hex key from GPU to 72 together with the assignment OR the worker reports the completion of the work unit using the Manual Pages of Primenet. The first would be the most practical, of course.
What you think?

Second point:


Is the option to allow people to choose low LL (or DC) tests from GPU to 72 in the works, or has it been abandoned?

1st point: After I manually alter my worktodo.txt I need to shutdown and restart prime95, or it does as yours does, and destroys the new stuff in worktodo.txt when it talks to the server. Once I started doing that, it worked fine.

2nd point: I would think he is working on it, but he can probably respond!

[lycorn]

Originally Posted by KyleAskine

1st point: After I manually alter my worktodo.txt I need to shutdown and restart prime95, or it does as yours does, and destroys the new stuff in worktodo.txt when it talks to the server. Once I started doing that, it worked fine.

Yep, you got it right, I had forgotten about that. Thx a bunch!

[chalsall]

Originally Posted by lycorn

I´ve just started using GPU to 72. Pretty smooth, nice and easy. Great work, chalsall! Another nice feature added to help on the GIMPS overall effort.

Thanks very much.

Originally Posted by lycorn

I got 4 exponents to P-1 (won´t be doing any TF, at least in the near future, for lack of a suitable GPU); all of them have N/A instead of the hex key in the corresponding worktodo.txt line, which is normal (am I right?...).

Yes, that is normal.

Originally Posted by lycorn

When the first exponent finished, the second one started straightaway but the result for the first one wasn´t immediately sent to Primenet Server, as I have Prime95 set to not automatically send results to the server. Some hours later, I sent the result to the server (via Prime95, not the manual pages). The result was accepted, credit correctly given, but the 3rd and 4th exponents in the queue, which had not any work done on them yet, got kicked out of worktodo.txt.

Hmmm... Damn. I had thought that the N/A would result in Prime95/mprime not trying to report back to PrimeNet until the work was completed. Obviously incorrect...

Originally Posted by lycorn

I see 2 workarounds for this situation: the worker receives the hex key from GPU to 72 together with the assignment OR the worker reports the completion of the work unit using the Manual Pages of Primenet. The first would be the most practical, of course. What you think?

The reason I'm hesitant to give the real AID is that the ownership of the candidate would then be transferred to the Worker when it checked in with PrimeNet, and possibly unreserved when the P-1 work was completed.

Perhaps manual submission of the results would be better?

Originally Posted by lycorn

Is the option to allow people to choose low LL (or DC) tests from GPU to 72 in the works, or has it been abandoned?

It's in the works. But it's going to be a few more days.

I have "real" work which has to be done too....

[KyleAskine]

Originally Posted by chalsall

Hmmm... Damn. I had thought that the N/A would result in Prime95/mprime not trying to report back to PrimeNet until the work was completed. Obviously incorrect...

It isn't a problem. The problem was with P95, not your assignment.