mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Operazione Doppi Mersennes (https://www.mersenneforum.org/forumdisplay.php?f=99)
-   -   Results (https://www.mersenneforum.org/showthread.php?t=17187)

ET_ 2012-09-13 18:31

Results
 
Please, insert here your results.txt files, so that we'll have a unique place to look to when in doubt.

Luigi

rcv 2012-09-13 19:17

Posting here for historic purposes. George's program was not yet ready for the public, but the kernel for MM127 was working fine. George invited me to work on this range. The early kernel kept the GPU so busy, that when run on my primary video card, the video lag / system responsiveness was measured in minutes. I ran this on a secondary card to which no video was connected. Started on about August 22, 2012. Finished on September 10, 2012. (With a software upgrade in the interim.) The work used a dedicated EVGA NVIDIA GTX 570.

The actual results.txt line reported is:

[CODE]no factor for M131071 from 2^68 to 2^69 [mfaktc 0.20 barrett92_mul32_gs][/CODE]In English, the above translates to

[CODE]no factor for MM127 from 2^178 to 2^179 [mfaktc 0.20 barrett92_mul32_gs][/CODE]

While the results line may appear bogus, George didn't yet have the complete input/output interface. "M131071" was code for "MM127". The bit-ranges were adjusted by 110. [That's why the above "translation" is provided.]

science_man_88 2012-09-13 19:21

[QUOTE=rcv;311475]Posting here for historic purposes. George's program was not yet ready for the public, but the kernel for MM127 was working fine. George invited me to work on this range. The early kernel kept the GPU so busy, that when run on my primary video card, the video lag / system responsiveness was measured in minutes. I ran this on a secondary card to which no video was connected. Started on about August 22, 2012. Finished on September 10, 2012. (With a software upgrade in the interim.) The work used a dedicated EVGA NVIDIA GTX 570.

The actual results.txt line reported is:

[CODE]no factor for M131071 from 2^68 to 2^69 [mfaktc 0.20 barrett92_mul32_gs][/CODE]In English, the above translates to

[CODE]no factor for MM127 from 2^178 to 2^179 [mfaktc 0.20 barrett92_mul32_gs][/CODE][/QUOTE]

[code](16:20)>(2^127-1)
%13 = 170141183460469231731687303715884105727[/code] you tested MM17 or it appears so.

rcv 2012-09-13 19:30

[QUOTE=science_man_88;311477]you tested MM17[/QUOTE]
Feel free to retest the range. The software is faster now. It should take 16.7 days on a factory-clocked EVGA GTX 570.

science_man_88 2012-09-13 19:33

[QUOTE=rcv;311479]Feel free to retest the range. The software is faster now. It should take 16.7 days on a factory-clocked EVGA GTX 570.[/QUOTE]

I don't have a GPU, I was just saying I don't see how M131071=MM17 to one bit range is MM127 to another bit range. okay never mind I see the edit.

ET_ 2012-09-13 19:39

[QUOTE=rcv;311475]Posting here for historic purposes. George's program was not yet ready for the public, but the kernel for MM127 was working fine. George invited me to work on this range. The early kernel kept the GPU so busy, that when run on my primary video card, the video lag / system responsiveness was measured in minutes. I ran this on a secondary card to which no video was connected. Started on about August 22, 2012. Finished on September 10, 2012. (With a software upgrade in the interim.) The work used a dedicated EVGA NVIDIA GTX 570.

The actual results.txt line reported is:

[CODE]no factor for M131071 from 2^68 to 2^69 [mfaktc 0.20 barrett92_mul32_gs][/CODE]In English, the above translates to

[CODE]no factor for MM127 from 2^178 to 2^179 [mfaktc 0.20 barrett92_mul32_gs][/CODE]

While the results line may appear bogus, George didn't yet have the complete input/output interface. "M131071" was code for "MM127". The bit-ranges were adjusted by 110. [That's why the above "translation" is provided.][/QUOTE]

So it seems that the early pre-alpha version run on barrett_92 kernel, after a fast hacking on the 95bit limit of the factor. Or maybe the [COLOR="Red"]name[/COLOR] of the kernel was the same while its [COLOR="Red"]content[/COLOR] was changed...

Prime95 2012-09-13 20:08

[QUOTE=ET_;311482]So it seems that the early pre-alpha version run on barrett_92 kernel, after a fast hacking on the 95bit limit of the factor. Or maybe the [COLOR="Red"]name[/COLOR] of the kernel was the same while its [COLOR="Red"]content[/COLOR] was changed...[/QUOTE]

The early, pre-pre-alpha was a complete hack. 131071 in worktodo did mean test MM127 but add 110 to the bit levels. Output of the proper kernel range was not coded. Output of the adjusted bit levels was not coded.

frmky 2012-09-14 05:18

[CODE]no factor for k*2^44+1 from 2^92 to 2^93 (k range: 400T to 700T) [mmff 0.21 mfaktc_barrett96_F32_63gs]
no factor for k*2^44+1 from 2^93 to 2^94 (k range: 0Q to 1Q) [mmff 0.21 mfaktc_barrett96_F32_63gs][/CODE]

flashjh 2012-09-14 13:48

Don't know if you want this here, but here it is:

for assignment MMFactor=127,179,2.5e15
[CODE]
no factor for MM127 from 2^179 to 2^180 (k range: 0 to 2500000000000000) [mmff 0.20mmff mfaktc_barrett183_M127gs]
[/CODE]

frmky 2012-09-14 18:39

[CODE]no factor for k*2^28+1 from 2^76 to 2^77 (k range: 550T to 1000T) [mmff 0.21 mfaktc_barrett89_F0_31gs]
no factor for k*2^28+1 from 2^77 to 2^78 (k range: 550T to 1000T) [mmff 0.21 mfaktc_barrett89_F0_31gs]
[/CODE]

Karl M Johnson 2012-09-15 16:37

[CODE]no factor for MM89 from 2^138 to 2^139 (k range: 500T to 600T) [mmff 0.21 mfaktc_barrett140_M89gs]
no factor for MM89 from 2^139 to 2^140 (k range: 0T to 600T) [mmff 0.21 mfaktc_barrett140_M89gs][/CODE]


All times are UTC. The time now is 09:23.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.