Double Check - A discussion thread - Page 2

Sloth · 2009-01-24, 03:35

I am starting to put together a spreadsheet to see how longs tests are taking. I only have a couple useful results at this point. Between systems that were finishing other work while also trying to do LLR testing and a power outage that shut down the server so it did not get results as they were finished, I only have good numbers for a couple systems.

All are using cllr-3-7-1c.exe
C2D E4500 2.2GHz running at 2.9GHz (800FSB pushed to 1066)
222113*2^6267213+1 took just over 26 hours. This system looks to be doing a test in 26 to 26.5 hours per core. I only have one other set of results from this system and it was around 28 hours. Not sure what was going on during that time but the 26 hour test all the system really did was the test.
C2D 4500 2.2GHz at 2.2 looks to finish just short of 30 hours. This one has 1 core working 24/7 and the other core doing some boinc projects 12 hours a day (using 1 core). Another 4500 @ 2.2 that is doing about 32 hours per core - running 2 cores on LLR full time.
C2D 6300 1.86GHz @ 2.33GHz - each core came back at 37.5 hours.

I have a Q6700 with no useful times yet and a 3GHz hyperthreaded system that is running 2 tests that I am still waiting for. I want to play with this one and see how much time is saved doing 2 tests at a time vs. 1. A few more days and I can give you times on a couple other processors and not just 4500s. I currently have about 20 cores running.

S.

VJS · 2009-01-25, 22:54

Just curious since like Lars,

I'm trying to build up a i7-920, my wife said get rid of some of the odl stuff first. So thats step one.

shauge · 2009-01-26, 09:50

I have some experience with i7-920 from FP, when overclocked to 3GHz, it does its loops in about 27ms running at 8 instances of LLR. A Core2quad (9550) does its loops in about 20-21 ms, at the same clock freq. So the three memory channels certainly improve things :)

VJS · 2009-01-28, 19:00

Interesting!!!

so that basically means 27ms / 2 so 13 to 14 ms as opposed to 20-21 wow!!!

That somewhat fits my old q660 mumbers which were around 17-18 @ 3.4Ghz if I recall. I also remember getting better perforamce at 400 MHz fsb over 333 Mhz but I don't remember it being that much.

Tobit · 2009-01-28, 19:16

I really like my Core i7 rig. I'm still new to all of this but I am pleased with my Sieving performance using sr2sive 1.8.6.

VJS · 2009-02-01, 14:42

post some numbers !!!

2009-01-24, 03:35	#12
Sloth Mar 2006 1011110₂ Posts	I am starting to put together a spreadsheet to see how longs tests are taking. I only have a couple useful results at this point. Between systems that were finishing other work while also trying to do LLR testing and a power outage that shut down the server so it did not get results as they were finished, I only have good numbers for a couple systems. All are using cllr-3-7-1c.exe C2D E4500 2.2GHz running at 2.9GHz (800FSB pushed to 1066) 2221132^6267213+1 took just over 26 hours. This system looks to be doing a test in 26 to 26.5 hours per core. I only have one other set of results from this system and it was around 28 hours. Not sure what was going on during that time but the 26 hour test all the system really did was the test. C2D 4500 2.2GHz at 2.2 looks to finish just short of 30 hours. This one has 1 core working 24/7 and the other core doing some boinc projects 12 hours a day (using 1 core). Another 4500 @ 2.2 that is doing about 32 hours per core - running 2 cores on LLR full time. C2D 6300 1.86GHz @ 2.33GHz - each core came back at 37.5 hours. I have a Q6700 with no useful times yet and a 3GHz hyperthreaded system that is running 2 tests that I am still waiting for. I want to play with this one and see how much time is saved doing 2 tests at a time vs. 1. A few more days and I can give you times on a couple other processors and not just 4500s. I currently have about 20 cores running. S. Last fiddled with by Sloth on 2009-01-24 at 03:37*

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Double checking discussion thread	philmoore	Five or Bust - The Dual Sierpinski Problem	20	2011-12-24 20:21
Double Check	Unregistered	Information & Answers	3	2011-10-01 04:38
First check and double check llrnet servers.	opyrt	Prime Sierpinski Project	3	2009-01-02 01:50
Double-check check?	M0CZY	Software	15	2008-10-30 14:20
Here's why we double-check...	gd_barnes	No Prime Left Behind	0	2008-02-11 19:23

2009-01-25, 22:54	#13
VJS Dec 2004 13·23 Posts	Just curious since like Lars, I'm trying to build up a i7-920, my wife said get rid of some of the odl stuff first. So thats step one.

2009-01-26, 09:50	#14
shauge Jan 2006 Stavanger, Norway 5 Posts	I have some experience with i7-920 from FP, when overclocked to 3GHz, it does its loops in about 27ms running at 8 instances of LLR. A Core2quad (9550) does its loops in about 20-21 ms, at the same clock freq. So the three memory channels certainly improve things :)

2009-01-28, 19:00	#15
VJS Dec 2004 100101011₂ Posts	Interesting!!! so that basically means 27ms / 2 so 13 to 14 ms as opposed to 20-21 wow!!! That somewhat fits my old q660 mumbers which were around 17-18 @ 3.4Ghz if I recall. I also remember getting better perforamce at 400 MHz fsb over 333 Mhz but I don't remember it being that much.

2009-01-28, 19:16	#16
Tobit Jan 2009 2 Posts	I really like my Core i7 rig. I'm still new to all of this but I am pleased with my Sieving performance using sr2sive 1.8.6.

2009-02-01, 14:42	#17
VJS Dec 2004 453₈ Posts	post some numbers !!!