Core i7 folding benchmarks??

ludeboy12 · 11-03-08, 04:24 PM

Anyone have any links that show comparisons between C2Q and Core i7 when it comes to folding?

drshivas · 11-03-08, 04:38 PM

http://techreport.com/articles.x/15818/12

Is_907 · 11-03-08, 05:31 PM

Looks like not having HT will be more useful for Folding... am I reading that graph correctly?
I'm also curious to see some numbers on the i7's for Rosetta... in case someone has found them already...

tgxiii · 11-03-08, 05:40 PM

I'm not really understanding the article. On all the graphs for individual work units, the no-HT i7-965 Extreme leads in ppd and the one with HT is at 6th or 7th place. However, the last graph (for Total Project PPD) shows it ahead of all except the dual QX9775 and the non-HT is at 5th.

harlam357 · 11-03-08, 06:47 PM

Wish they would have run some SMP numbers.

ChasR · 11-03-08, 07:20 PM

The techreport benchmark is all but useless without a1 and a2 core SMP WUs. You can figure it to be faster, but by how much on each type of SMP WU is a guess. Where it should trounce the current Qs is in running multiple SMP instances with the a1 core.

ludeboy12 · 11-03-08, 08:33 PM

Yea would be nice to see some clock for clock SMP comparisons.....

LandShark · 11-04-08, 08:24 PM

Quote:

Originally Posted by ChasR

The techreport benchmark is all but useless without a1 and a2 core SMP WUs. You can figure it to be faster, but by how much on each type of SMP WU is a guess. Where it should trounce the current Qs is in running multiple SMP instances with the a1 core.

exactly my thought!

w/ HT, total of 8 cores, running 2x WinSMP client to fully use all cores. or if one has enough ram in the system, how 'bout 4x VMWare+LinuxSMP? or better yet, fill all those PCIe 16x w/ GPUs, then run SMP (Win or Linux, your choice) on the rest of the cores!!

may be Stanford will lift the 4 cores hard coded soon...??

Shelnutt2 · 11-04-08, 08:39 PM

Quote:

Originally Posted by LandShark

may be Stanford will lift the 4 cores hard coded soon...??

I didn't realize they still had that with the v6 clients? I know a while ago they said if you asked and had just reason they would get you a client that had 8 or 16 threads if you proved you had the hardware. This was mostly talking about a cluster setup though.

ChasR · 11-04-08, 10:45 PM

I think 2 clients x 4 cores is going to be faster than 1 x 8. If I am not mistaken, the Linux 6.23 client isn't hardcoded to 4 threads. If you have 8 cores, -smp 8 will run one instance across 8 cores. 4 is still the minimum.

11-03-08, 04:24 PM	Thread Starter #1
ludeboy12 Join Date: Mar 2005 Location: Wheeling, WV	Core i7 folding benchmarks?? Anyone have any links that show comparisons between C2Q and Core i7 when it comes to folding? __________________ C2Q Rig.... C2Q Q6600 @ 3.6ghz \| Abit IP35 Pro \| 2 x 1gb G.Skill DDR2 PC6400 \| eVGA 8800gts 320 x 2 \| Silverstone TJ-09 i7 Rig.... i7 920 @ 3.8ghz \| eVGA x58 \| 3 x 1gb G.Skill DDR3 1600 \| eVGA 8800gts g92 x 2 \| Caseless FAH Quad / GPU Farm....To be continued.... My Folding Stats...
	QUOTE Thanks

11-03-08, 04:38 PM	#2
drshivas Sith Lord Join Date: Jun 2003 Location: Maryland, USA	http://techreport.com/articles.x/15818/12
	QUOTE Thanks

11-03-08, 05:31 PM	#3
Is_907 Join Date: May 2002 Location: Keller, TX	Looks like not having HT will be more useful for Folding... am I reading that graph correctly? I'm also curious to see some numbers on the i7's for Rosetta... in case someone has found them already... __________________ Castiel (Desktop Rig) - i7-3770; 16GB DDR3; 256GB Crucial M4 + 600GB RAID0 array; Asus GTX460 1GB.
	QUOTE Thanks

11-03-08, 05:40 PM	#4
tgxiii Member Join Date: Feb 2004 Location: Folding @ Eagle Rock, CA	I'm not really understanding the article. On all the graphs for individual work units, the no-HT i7-965 Extreme leads in ppd and the one with HT is at 6th or 7th place. However, the last graph (for Total Project PPD) shows it ahead of all except the dual QX9775 and the non-HT is at 5th.
	QUOTE Thanks

11-03-08, 06:47 PM	#5
harlam357 Senior Fold-a-holic Join Date: Sep 2004 Location: @Home... Folding@Home	Wish they would have run some SMP numbers. __________________ HFM.NET - Client Monitoring Application for Folding@Home \| Team 32 Super Sticky \| Project Quadteron \| Heat (66-0-0)
	QUOTE Thanks

Core i7 folding benchmarks??

Internet Services

Website Development

Tech Hardware

11-03-08, 07:20 PM	#6
ChasR Senior Member Join Date: Apr 2004 Location: Atlanta	The techreport benchmark is all but useless without a1 and a2 core SMP WUs. You can figure it to be faster, but by how much on each type of SMP WU is a guess. Where it should trounce the current Qs is in running multiple SMP instances with the a1 core. __________________ All ppd from friends and family. Many thanks to HayesK for his support and congrats on being the #1 Folder on Team 32. Special thanks to ChrisC and ColinC
	QUOTE Thanks

11-03-08, 08:33 PM	Thread Starter #7
ludeboy12 Join Date: Mar 2005 Location: Wheeling, WV	Yea would be nice to see some clock for clock SMP comparisons..... __________________ C2Q Rig.... C2Q Q6600 @ 3.6ghz \| Abit IP35 Pro \| 2 x 1gb G.Skill DDR2 PC6400 \| eVGA 8800gts 320 x 2 \| Silverstone TJ-09 i7 Rig.... i7 920 @ 3.8ghz \| eVGA x58 \| 3 x 1gb G.Skill DDR3 1600 \| eVGA 8800gts g92 x 2 \| Caseless FAH Quad / GPU Farm....To be continued.... My Folding Stats...
	QUOTE Thanks

11-04-08, 10:45 PM	#10
ChasR Senior Member Join Date: Apr 2004 Location: Atlanta	I think 2 clients x 4 cores is going to be faster than 1 x 8. If I am not mistaken, the Linux 6.23 client isn't hardcoded to 4 threads. If you have 8 cores, -smp 8 will run one instance across 8 cores. 4 is still the minimum. __________________ All ppd from friends and family. Many thanks to HayesK for his support and congrats on being the #1 Folder on Team 32. Special thanks to ChrisC and ColinC
	QUOTE Thanks

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search