Strategic Double Clicking - Page 142

storm5510 · 2017-07-23, 01:49

Quote:

Originally Posted by nofaith628

Current status with my dual 14 core Xeon machine is (running two workers with prime95):
~1.5ms/iter on 55M exponents (Worker 1)
~1.8ms/iter on 59M exponents (Worker 2)

The machine runs draws around 330W of power according to my UPS display.

You say "dual" 14 core. I would take this to mean two cpu's on one board? 1.5 ms/iter. is amazing. The best I've seen on this 7700 is 2.7 ms/iter.

When I built this two months ago, I was naive of the differences between a 7700 and a 7700K, so I went the 7700. Live and learn. Sometime, I will order a 7700K and part with the 7700. Prices on the "K" dropped off when the "X" series was released. My mobo has OC capability, but I would use it lightly. I added a liquid cooler about a month ago and it made a lot of difference.

I did some power consumption measurements on this setup a few weeks ago. At no load, this rig draws 95W. Running Prime95, it is 172W. Running mfaktc on this long-in-the-tooth GTX-480, it goes to 285W. If I ran both, then it jumps to 380W. The other GPU programs use somewhat less. I didn't measure those. This PSU is 700W. but I really don't care for pushing it. If I run both, that is 2.86 kWh a day, or 85 kWh per month.

nofaith628 · 2017-07-23, 14:47

You are indeed correct, the current setup with this workstation is two E5-2695 V3's on a single motherboard.

At first, 55M exponents would output from 2.5ms/iter and fluctuate up to 7.0ms/iter, realised this was the fault of the motherboard's Intel Xeon Turbo Boost technology. I must say I am not too well versed in the in-depth computer gimmicks, but I observed the clocks of each CPU fluctuate wildly while the LL tests ran. After some tweaking and turning the turbo boost settings off, along with experimenting with the best rate of return by testing different values for threads per worker, I ended up with 2 workers, with each worker assigned ThreadsPerTest=26, NewCPUSpeed=2195, and here I am churning through the double checks efficiently.

And next two months I shall be grabbing my next batch of strategic double checks.

storm5510 · 2017-07-23, 17:20

As a comparison, https://www.mersenne.org/report_expo...9522479&full=1 will take 68 hours to verify at 4.1 ms/iter.

I still cannot get these links to appear correctly!

thyw · 2017-07-23, 20:32

Quote:

Originally Posted by storm5510

As a comparison, https://www.mersenne.org/report_expo...9522479&full=1 will take 68 hours to verify at 4.1 ms/iter.
I still cannot get these links to appear correctly!

Edit: if it's about not showing full links, and showing ... in the centre, it appears to be a forum config option.
https://www.vbulletin.org/forum/showthread.php?t=229729 https://www.vbulletin.com/forum/foru...-not-truncated
"If you don't mind them not being linked, just turn off the "Automatically parse URLs" option when posting, and don't put [url] tags around em."
--
Do you want https://www.mersenne.org/M59522479 this form? It's the same redirect.
Or the LL results? https://www.mersenne.org/report_ll/?...dispdate=1&B1=
Or active assigments? https://www.mersenne.org/assignments...xp_hi=59522479

Or are you talking about the appearance? End of http://mersenneforum.org/showthread.php?t=5758&page=32 and http://mersenneforum.org/showthread.php?t=19716 threads.
^tl;dr (There was a formating bug about the blue menu elements in Firefox 54/55, but it resolved now)

Also these were pretty off topic posts in the end.

science_man_88 · 2017-07-23, 20:58

Quote:

Originally Posted by storm5510

As a comparison, https://www.mersenne.org/report_expo...9522479&full=1 will take 68 hours to verify at 4.1 ms/iter.

I still cannot get these links to appear correctly!

try the green $2^p-1$ symbol:

59522479

storm5510 · 2017-07-24, 00:05

Quote:

Originally Posted by thyw

Also these were pretty off topic posts in the end.

You are correct, it was.

Madpoo · 2017-07-26, 16:10

Updated list of possibly bad stuff, sorted by exponent:

Code:

exponent	Bad	Good	Unk	Sus	Solo	Mis	worktodo
52988959	2	1	1	0	1	0	DoubleCheck=52988959,73,1
53954513	1	0	3	0	3	0	DoubleCheck=53954513,73,1
53990221	1	0	3	0	3	0	DoubleCheck=53990221,73,1
56069149	1	0	3	0	3	0	DoubleCheck=56069149,73,1
57436123	1	0	5	0	5	0	DoubleCheck=57436123,73,1
57441157	1	0	5	0	5	0	DoubleCheck=57441157,73,1
57547283	2	0	2	0	2	0	DoubleCheck=57547283,73,1
57966593	1	0	5	0	5	0	DoubleCheck=57966593,73,1
57966613	1	0	5	0	5	0	DoubleCheck=57966613,73,1
57966761	1	0	5	0	5	0	DoubleCheck=57966761,73,1
61340831	1	0	3	0	3	0	DoubleCheck=61340831,74,1
61340863	1	0	3	0	3	0	DoubleCheck=61340863,74,1
61436833	1	0	3	0	3	0	DoubleCheck=61436833,74,1
63726899	1	0	2	0	2	0	DoubleCheck=63726899,74,1
63727619	1	0	2	0	2	0	DoubleCheck=63727619,74,1
64629557	1	0	2	0	2	0	DoubleCheck=64629557,74,1
64929833	1	0	2	0	2	0	DoubleCheck=64929833,74,1
72314813	1	0	2	0	2	0	DoubleCheck=72314813,75,1
72326587	1	0	1	0	1	0	DoubleCheck=72326587,75,1
72531889	1	0	1	0	1	0	DoubleCheck=72531889,75,1
73009073	1	0	2	0	2	0	DoubleCheck=73009073,75,1
74124077	1	0	2	0	2	0	DoubleCheck=74124077,75,1
74543201	1	0	2	0	2	0	DoubleCheck=74543201,75,1
74810917	1	0	4	0	4	0	DoubleCheck=74810917,75,1
74873957	2	0	2	0	2	0	DoubleCheck=74873957,75,1
74890243	2	0	2	0	2	0	DoubleCheck=74890243,75,1
75009079	1	0	4	0	4	0	DoubleCheck=75009079,75,1
75130961	1	0	4	0	4	0	DoubleCheck=75130961,75,1
75682133	1	0	2	0	2	0	DoubleCheck=75682133,75,1
75691787	1	0	4	0	4	0	DoubleCheck=75691787,75,1
76860967	1	0	1	1	2	0	DoubleCheck=76860967,75,1
76880393	1	0	1	1	2	0	DoubleCheck=76880393,75,1
76953853	1	0	0	3	3	0	DoubleCheck=76953853,75,1
76953869	1	0	0	3	3	0	DoubleCheck=76953869,75,1
76953893	1	0	0	3	3	0	DoubleCheck=76953893,75,1
77042293	1	0	1	0	1	0	DoubleCheck=77042293,75,1
77379023	1	0	5	0	5	0	DoubleCheck=77379023,75,1
77488337	1	0	5	0	5	0	DoubleCheck=77488337,75,1
78049679	1	0	2	0	2	0	DoubleCheck=78049679,75,1
78185621	3	2	1	0	1	0	DoubleCheck=78185621,75,1
78297283	1	0	3	0	3	0	DoubleCheck=78297283,75,1
78297397	1	0	3	0	3	0	DoubleCheck=78297397,75,1
78297437	1	0	3	0	3	0	DoubleCheck=78297437,75,1
80400319	3	0	0	2	2	0	DoubleCheck=80400319,76,1
80492707	3	0	0	2	2	0	DoubleCheck=80492707,75,1
81474373	3	0	1	0	1	0	DoubleCheck=81474373,75,1
81512009	2	0	1	0	1	0	DoubleCheck=81512009,75,1
81581413	1	0	1	0	1	0	DoubleCheck=81581413,75,1
81605137	1	0	0	1	1	0	DoubleCheck=81605137,75,1

cuBerBruce · 2017-07-27, 04:12

Quote:

Originally Posted by Madpoo

Updated list of possibly bad stuff, sorted by exponent:

I took the first two:

Code:

52988959	2	1	1	0	1	0	DoubleCheck=52988959,73,1
53954513	1	0	3	0	3	0	DoubleCheck=53954513,73,1

storm5510 · 2017-07-27, 07:25

My DC result for http://www.mersenne.ca/exponent/59522479 was a mismatch.

I do not know if this is indicating that I have a hardware problem or if this means the original LL and DC were flawed. I have done three matching DC's on this new hardware and they all matched.

M59522479 remains "Unverified" so someone else will have to run it again.

ATH · 2017-07-27, 10:17

Quote:

Originally Posted by storm5510

M59522479 remains "Unverified" so someone else will have to run it again.

I started it on the Titan Black.

storm5510 · 2017-07-27, 16:19

Quote:

Originally Posted by ATH

I started it on the Titan Black.

Thank you!

2017-07-23, 17:20	#1554
storm5510 Random Account Aug 2009 U.S.A. 3404₈ Posts	As a comparison, https://www.mersenne.org/report_expo...9522479&full=1 will take 68 hours to verify at 4.1 ms/iter. I still cannot get these links to appear correctly! Last fiddled with by storm5510 on 2017-07-23 at 17:24

2017-07-27, 07:25	#1560
storm5510 Random Account Aug 2009 U.S.A. 3404₈ Posts	Mismatch My DC result for http://www.mersenne.ca/exponent/59522479 was a mismatch. I do not know if this is indicating that I have a hardware problem or if this means the original LL and DC were flawed. I have done three matching DC's on this new hardware and they all matched. M59522479 remains "Unverified" so someone else will have to run it again.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Double-Double Arithmetic	Mysticial	Software	50	2017-10-30 19:16
Clicking an exponent leads to 404 page	marigonzes	Information & Answers	2	2017-02-14 16:56
x.265 half the size, double the computation; so if you double again? 1/4th?	jasong	jasong	7	2015-08-17 10:56
What about double-checking TF/P-1?	137ben	PrimeNet	6	2012-03-13 04:01
Double the area, Double the volume.	Uncwilly	Puzzles	8	2006-07-03 16:02

2017-07-23, 14:47	#1553
nofaith628 Feb 2017 2·17 Posts	You are indeed correct, the current setup with this workstation is two E5-2695 V3's on a single motherboard. At first, 55M exponents would output from 2.5ms/iter and fluctuate up to 7.0ms/iter, realised this was the fault of the motherboard's Intel Xeon Turbo Boost technology. I must say I am not too well versed in the in-depth computer gimmicks, but I observed the clocks of each CPU fluctuate wildly while the LL tests ran. After some tweaking and turning the turbo boost settings off, along with experimenting with the best rate of return by testing different values for threads per worker, I ended up with 2 workers, with each worker assigned ThreadsPerTest=26, NewCPUSpeed=2195, and here I am churning through the double checks efficiently. And next two months I shall be grabbing my next batch of strategic double checks.