[QUOTE=Dubslow;282378]How much? Certainly not 1600/1333*100%, right?[/QUOTE]

I wish I took accurate timings, but at the time I wanted hurry up and get onto mfaktc.

-- Craig

[strikethrough?] Nice post count ;) [/s]

Awwww..... :(

[strike]Test[/strike]

Note that you need to allocate more than 2GiB to get the Brent-Suyama extension to kick in. We think ~4GiB gives you E=6 and ~6-7 GiB gives you E=12.

I tried that.

[strike]Apparently not.[/strike]

[QUOTE=Xyzzy;282427][strike]Test[/strike]

Note that you need to allocate more than 2GiB to get the Brent-Suyama extension to kick in. We think ~4GiB gives you E=6 and ~6-7 GiB gives you E=12.[/QUOTE]

With all due respect, I think the numbers must be a bit lower.

Prime95-64 allocation 2048MB day, 3096MB night.
MaxHighMemWorkers=2 (with 3 workers doing P-1)

From results.txt:

[CODE]UID: kladner/pod64, M49389281 completed P-1, B1=475000, B2=9381250, E=6, We4: FD4212C2
[Sat Dec 10 23:42:42 2011]
UID: kladner/pod64, M49929079 completed P-1, B1=485000, B2=9700000, E=6, We4: 0B55E23E
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx snip
[Sun Dec 11 09:45:25 2011]
UID: kladner/pod64, M49366459 completed P-1, B1=475000, B2=9381250, E=6, We4: FDB411B4
[Mon Dec 12 03:16:35 2011]
UID: kladner/pod64, M49933799 completed P-1, B1=485000, B2=9700000, We4: 0B49E217
[Mon Dec 12 09:38:57 2011]
UID: kladner/pod64, M51878003 completed P-1, B1=505000, B2=10226250, E=6, We4: 3EBC689D
[Mon Dec 12 13:48:12 2011]
UID: kladner/pod64, M49938787 completed P-1, B1=485000, B2=9700000, E=6, We4: 0B62E25F
[Tue Dec 13 04:25:57 2011]
P-1 found a factor in stage #2, B1=485000, B2=9700000.
UID: kladner/pod64, M49936763 has a factor: 172394453366165817972339097
[Tue Dec 13 14:39:24 2011]
UID: kladner/pod64, M51885271 completed P-1, B1=505000, B2=10226250, E=6, We4: 3ECD6832
[Tue Dec 13 21:22:56 2011]
UID: kladner/pod64, M49394069 completed P-1, B1=475000, B2=9500000, E=6, We4: FD7642BA
[Wed Dec 14 07:42:00 2011]
UID: kladner/pod64, M49936853 completed P-1, B1=485000, B2=9700000, E=6, We4: 0B7EE240
[Wed Dec 14 18:08:55 2011]
UID: kladner/pod64, M49394131 completed P-1, B1=475000, B2=9500000, E=6, We4: FD654284
[Thu Dec 15 03:21:18 2011]
UID: kladner/pod64, M51885277 completed P-1, B1=505000, B2=10226250, E=6, We4: 3EE66847
[Thu Dec 15 10:29:39 2011]
UID: kladner/pod64, M49397371 completed P-1, B1=475000, B2=9500000, E=6, We4: FE97425E
[Fri Dec 16 00:12:49 2011]
UID: kladner/pod64, M51951883 completed P-1, B1=505000, B2=10226250, E=6, We4: 40A667BB
[Fri Dec 16 05:01:57 2011]
UID: kladner/pod64, M49936853 completed P-1, B1=485000, B2=9700000, E=6, We4: 0B7EE240
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx snip
[Fri Dec 16 17:34:34 2011]
UID: kladner/pod64, M51951373 completed P-1, B1=505000, B2=10226250, E=6, We4: 40AB67B6
[/CODE]

We are allocating 2GiB per test and we do not get the E=6.

:mike:

What do you mean "per test"? I only allocate it "per day time" and "per night time" :smile: which I believe is for ALL tests, totally, and not per each test. In case of two workers wanting stage 2 in the same time, well, let P95 deal with scheduling. And for 4096 we get E=12.

[QUOTE=LaurV;282533]In case of two workers wanting stage 2 in the same time, well, let P95 deal with scheduling. And for 4096 we get E=12.[/QUOTE]

Hmm. I've never gone that high. E=6 is the limit of what I've seen. I need to check out the results.txt on the Linux box. It has 2400MB allocated between two workers, no MaxHighMemWorkers set.

EDIT:
[Tue Dec 6 02:01:42 2011]
UID: kladner/POD-95, M51960971 completed P-1, B1=505000, B2=10100000, E=6, We8: 40F30865
[Tue Dec 6 10:35:36 2011]
UID: kladner/POD-95, M52181383 completed P-1, B1=505000, B2=10100000, E=6, We8: 455533E1

[Fri Nov 18 08:25:59 2011]
UID: kladner/POD-95, M60408709 completed P-1, B1=710000, B2=19170000, E=6, We8: 1BCD7822, AID: 4F00F4FFxxxxxxxxxxxxxxxxxxxxxxxxxxxx
UID: kladner/POD-95, M60410869 completed P-1, B1=710000, B2=19170000, E=6, We8: 1BC17838, AID: 796D0521xxxxxxxxxxxxxxxxxxxxxxxxxxxx


EDIT2: Opteron 180, 3GB RAM, Ubuntu 11.10 64bit

I don't know what drives this, but I do have these results.

[QUOTE]What do you mean "per test"?[/QUOTE]We have never been able to get the memory scheduling worked out so we run 4 instances of Mprime for better control.

Yes, memory scheduling isn't particularly great (though changing memory settings at any given time is very difficult), and individual thread management is impossible in MPrime.

[QUOTE=Dubslow;282549]Yes, memory scheduling isn't particularly great (though changing memory settings at any given time is very difficult), and individual thread management is impossible in MPrime.[/QUOTE]

I have 2GB allocated for P-1 tests, and I get E=6.

Luigi