GPU to 72 status...

[Dubslow]

While I'm at it, how did this happen?
http://gpu72.com/reports/workers/ll/

 #	Worker	Own	68	69	70	71	72	73	>73	 	Facts	Saved	Done
1	Bdot	48									49				3,557
2	prime95	70									45				3,268
3	kcjaget	16									30				2,179
4	1997rj7	12									27				1,958
5	Pete	7									25				1,819
6	Chuck	7									18				1,304
7	LaurV	20									14				1,014
8	Jerry Hallett	20									13				859
9	rduerr	33									11		1	144	820
10	markr	2									10				725
11	kdgehman	4									10				725
12	delta_t	6									9				655

Edit: Since I have the post, has anybody else noticed how creepily even the total LL/DC throughput has stayed, from a work/GHz-Days perspective? I know many of us are doing just one or the other, so it's remarkable how close they are. (At the time of this edit, the totals were 22,503/22,480 respectively.)

[James Heinrich]

Originally Posted by Dubslow

While I'm at it, how did this happen?

Since it wasn't immediately obvious to me due to scrolling of the code box, I assume you mean "how did rduerr find a factor with LL"?

[Dubslow]

Originally Posted by James Heinrich

Since it wasn't immediately obvious to me due to scrolling of the code box, I assume you mean "how did rduerr find a factor with LL"?

Indeed.

[Radikalinsky]

Since it wasn't immediately obvious to me due to scrolling of the code box, I assume you mean "how did rduerr find a factor with LL"?

rduerr did a TF up to 73 before the LL test. By finding a factor, he saved about 72 GHz days for the LL test.

[KyleAskine]

Originally Posted by chalsall

OK. A quick hack before I run out for a "Sundowner"...

If you go to the LLTF form and choose the new option "I have a slow card!!!" it will ignore the range and pledge fields, and issue you work between 60M and 61M to TF from 70 to 71.

I would again like to point out on behalf of BDot, Jerry, myself, and all of the other AMD owners that AMD's are around 10-15% faster going to 70 and below than to 71 and 72 due to how the optimal kernel is constructed.

I won't click the slow card button, and I will grab some 72's if the project thinks that is what is better, but I just hope you have this in your mind as a factor to think about. The performance drop is very significant for us above 70.

[bcp19]

Originally Posted by KyleAskine

I would again like to point out on behalf of BDot, Jerry, myself, and all of the other AMD owners that AMD's are around 10-15% faster going to 70 and below than to 71 and 72 due to how the optimal kernel is constructed.

I won't click the slow card button, and I will grab some 72's if the project thinks that is what is better, but I just hope you have this in your mind as a factor to think about. The performance drop is very significant for us above 70.

Hmm, that's a factor I did not take into consideration when I worked out the charts. It wouldn't affect the overall outcome much though, at least until a program to run LL's on those GPUs is made.

[Batalov]

Periodically the reservation system backfires.
When it happened first time, I thought wth but I couldn't be bothered.
Now, this time I happened to have a snapshot of my reservation page, so I defintely know that it was my asignment:

52456903 P-1 72 - - 2012-04-08 19:30:36  2.46

...yet, the final result was "not needed", -- it was done by someone else (J.H.).


P.S. The other J.H.

[James Heinrich]

Originally Posted by Batalov

...yet, the final result was "not needed", -- it was done by someone else (J.H.).

It wasn't me!

[flashjh]

Originally Posted by Batalov

Periodically the reservation system backfires.
When it happened first time, I thought wth but I couldn't be bothered.
Now, this time I happened to have a snapshot of my reservation page, so I defintely know that it was my asignment:

52456903 P-1 72 - - 2012-04-08 19:30:36  2.46

...yet, the final result was "not needed", -- it was done by someone else (J.H.).

It says it was assigned to me today, but I haven't grabbed any P-1 assignments today?

52456903 P-1 72 --2012-04-09 02:20:39 2012-04-09 09:33:23

Edit: Forgot... I did grab 20 P-1s last night, so today Zulu time, for a new system. I don't know what happened?

[rcv]

@chalsall: If I interpret the charts correctly, GPUto72 holds 6588 assignments between 53M and 54M. Of those, 0059 are assigned. The other 99+% are simply being hoarded. While the hoarding isn't as bad in all ranges, the hoarding remains egregious. As of this moment, PrimeNet offers nothing for TF below 65M

On my own behalf, I am requesting that you release some of your hoarded assignments in all available ranges, so that I may acquire some assignments directly from the PrimeNet server. [I respectfully suggest you release 50% of your hoarded assignments. If you run out first, you can ask PrimeNet for 50% of what remains. if PrimeNet runs out first, you should release 50% of what you still hold.]

Thank you.

[chalsall]

Originally Posted by rcv

@chalsall: If I interpret the charts correctly, GPUto72 holds 6588 assignments between 53M and 54M. Of those, 0059 are assigned.

No; you are not interpreting the charts correctly. Please look, and come again.

Originally Posted by rcv

On my own behalf, I am requesting that you release some of your hoarded assignments in all available ranges, so that I may acquire some assignments directly from the PrimeNet server. [I respectfully suggest you release 50% of your hoarded assignments. If you run out first, you can ask PrimeNet for 50% of what remains. if PrimeNet runs out first, you should release 50% of what you still hold.

I have already released 11,000 candidates back to PrimeNet to issue for Trial Factoring. Why don't you take some of them?

As an aside rcv, just how much TF work can you do compared to what we can do?