• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Intell i7 and linux were just meant to fold!

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Methal

Methal

Member
​
Joined
Aug 5, 2008
Location
DC
So i've been messing around with my bios settings. I figured out that at 2.7ghz i'll get 1% on -smp about...

Code: 
[04:52:55] Completed 220010 out of 250000 steps  (88%)
[04:57:50] Completed 222500 out of 250000 steps  (89%)

...that offten.

that is with one smp. with two it goes up about 1.5-2 minutes between percentages.

that is at stock.

One cool thing is:

Code: 
root@erik-desktop:~# cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 1596.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 0
cpu cores	: 4
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5306.69
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 2793.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 3
cpu cores	: 4
apicid		: 6
initial apicid	: 6
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.07
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 2
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 2793.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 1
cpu cores	: 4
apicid		: 2
initial apicid	: 2
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 3
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 2793.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 2
cpu cores	: 4
apicid		: 4
initial apicid	: 4
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 4
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 1596.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 3
cpu cores	: 4
apicid		: 7
initial apicid	: 7
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 5
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 1596.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 2
cpu cores	: 4
apicid		: 5
initial apicid	: 5
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 6
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 1596.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 1
cpu cores	: 4
apicid		: 3
initial apicid	: 3
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 7
vendor_id	: GenuineIntel
cpu family	: 6
model		: 26
model name	: Intel(R) Core(TM) i7 CPU         920  @ 2.67GHz
stepping	: 4
cpu MHz		: 2793.000
cache size	: 8192 KB
physical id	: 0
siblings	: 8
core id		: 0
cpu cores	: 4
apicid		: 1
initial apicid	: 1
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 lahf_lm ida tpr_shadow vnmi flexpriority ept vpid
bogomips	: 5307.08
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

(yes thats a crap ton of info) I've got 4 cores running at 2.7ghz. And 4 running at 1.5ghz. (saves on power and heat) that is with 1 smp. with two it goes up of course.


Now when I clock my proc up to about 3.3ghz I get 1% ever 4 minutes, with two I get 1% every 6.

up to 4.1ghz I am getting 1% every 3 minutes and 1% every 5 minutes with two smps running

This thing is a folding God =D
 
An i7 running -smp 8 on one instance should come within a couple of hundred ppd of one running two instances. From a science standpoint, it doesn't make sense to delay the return of a WU by almost 100% to gain 2% ppd.
 
At stock, running one instance, you're equaling a Q6600 running @ 3.0 GHz.

4:55/ frame is 5623 ppd.

BTW on a 1920 point WU:
3:00/frame = 9216 ppd
4:00/frame = 6912 ppd
5:00/Frame = 5530 ppd
6:00/frame = 4608 ppd

At 4.1 GHz your reported ppd would be 13,824 ppd, which would put you 2nd on the most powerful folders list at FCF with far and away the best i7 numbers.

It would help to know which WU (p2677 is faster than p2669 for instance), and wouldn't hurt to have exact times (so I don't have to calculate ppd and when you get into the 3:00/frame range, every second is 51 ppd), or even easier output from HFM benchmarking:

p2677_IBX in water

Name: Val2 VM2 (q9450 @ 3.2GHz)
Path: \\Val2-vm2\chasr vm2\
Number of Frames Observed: 300

Min. Time / Frame : 00:08:51 - 3124.1 PPD
Avg. Time / Frame : 00:09:16 - 2983.6 PPD
Cur. Time / Frame : 00:09:26 - 2930.9 PPD
R3F. Time / Frame : 00:09:29 - 2915.4 PPD
All Time / Frame : 00:09:21 - 2957 PPD
Eff. Time / Frame : 00:09:26 - 2930.9 PPD
 
ChasR said:
At stock, running one instance, you're equaling a Q6600 running @ 3.0 GHz.

4:55/ frame is 5623 ppd.

BTW on a 1920 point WU:
3:00/frame = 9216 ppd
4:00/frame = 6912 ppd
5:00/Frame = 5530 ppd
6:00/frame = 4608 ppd

At 4.1 GHz your reported ppd would be 13,824 ppd, which would put you 2nd on the most powerful folders list at FCF with far and away the best i7 numbers.

It would help to know which WU (p2677 is faster than p2669 for instance), and wouldn't hurt to have exact times (so I don't have to calculate ppd and when you get into the 3:00/frame range, every second is 51 ppd), or even easier output from HFM benchmarking:

p2677_IBX in water

Name: Val2 VM2 (q9450 @ 3.2GHz)
Path: \\Val2-vm2\chasr vm2\
Number of Frames Observed: 300

Min. Time / Frame : 00:08:51 - 3124.1 PPD
Avg. Time / Frame : 00:09:16 - 2983.6 PPD
Cur. Time / Frame : 00:09:26 - 2930.9 PPD
R3F. Time / Frame : 00:09:29 - 2915.4 PPD
All Time / Frame : 00:09:21 - 2957 PPD
Eff. Time / Frame : 00:09:26 - 2930.9 PPD
Click to expand...

after messing around with it at 4.1 ghz. I scaled back down to 3.7. sometime last night my computer restarted. Checking my log there was some error.

I just set up the -smp 8. Going to see how that works =D

Code: 
[17:03:10] Project: 2677 (Run 31, Clone 82, Gen 22)

is that what you wanted to know?

Code: 
Making 3D domain decomposition 2 x 2 x 2
starting mdrun '22860 system in water'
5750000 steps,  11500.0 ps (continuing from step 5500010,  11000.0 ps).
[17:03:18] Resuming from checkpoint
[17:03:18] Verified work/wudata_01.log
[17:03:18] Verified work/wudata_01.trr
[17:03:18] Verified work/wudata_01.xtc
[17:03:18] Verified work/wudata_01.edr
[17:03:19] Completed 10 out of 250000 steps  (0%)
[17:07:01] Completed 2500 out of 250000 steps  (1%)

I think this is at 3.7ghz. If I do a cat /proc/cpuinfo for some reason it aborts my folding.

Going to do more research and see if I can get this thing to run stable at 4.1

Now if I were to run -advmethodes would it increase my ppd? or increase my core from idle to low? or big insteasd of small or normal?
 
Last edited:
Methal said:
I think this is at 3.7ghz. If I do a cat /proc/cpuinfo for some reason it aborts my folding.

Going to do more research and see if I can get this thing to run stable at 4.1

Now if I were to run -advmethodes would it increase my ppd? or increase my core from idle to low? or big insteasd of small or normal?
Click to expand...

*scratches head* Hmmm, that does sound strange, do you get an 'aborted' or 'client shutdown' output to the terminal, or does the output in terminal just all disappear?

One question: is this native Linux, or in a VM? Reason I asked is that I had a Linux VM guest whose terminal behaved oddly, couldn't "copy & paste" to terminal with right mouse button, had to use the menu bar(edit>paste).
 
Right now it appears you'll get a mix of WUs with or without -advmethods. With all mine set to -adv, I've got 16 a2 WU and 5 a1 WUs. The small/normal/big have no effect on SMP, all the WUs are big.
 
ihrsetrdr said:
*scratches head* Hmmm, that does sound strange, do you get an 'aborted' or 'client shutdown' output to the terminal, or does the output in terminal just all disappear?

One question: is this native Linux, or in a VM? Reason I asked is that I had a Linux VM guest whose terminal behaved oddly, couldn't "copy & paste" to terminal with right mouse button, had to use the menu bar(edit>paste).
Click to expand...

This is native linux. The exact error is "client aborted by user" which is says 3 times. then shuts the core down.

with my cpu running at 3.7ghz this is what I am getting:

Code: 
[18:09:45] Completed 15010 out of 250000 steps  (6%)
[18:13:24] Completed 17500 out of 250000 steps  (7%)
[18:17:05] Completed 20000 out of 250000 steps  (8%)
[18:20:45] Completed 22500 out of 250000 steps  (9%)
[18:24:24] Completed 25000 out of 250000 steps  (10%)
[18:28:03] Completed 27500 out of 250000 steps  (11%)
 
Code: 
erik@erik-desktop:~$ uname -a
Linux erik-desktop 2.6.28-13-generic #45-Ubuntu SMP Tue Jun 30 22:12:12 UTC 2009 x86_64 GNU/Linux
 
I could swear that I've heard of this happening, at some point in the past. I can't imagine why such would occur, have never seen input in one terminal affect the activity in a different terminal. *am curious*.
 
I don't understand all th workings of cat /proc/cpuinfo. But in my little brain it would make sense if when I "pinged" the cpu it may have caused a hickup in what fah was doing
 
Reading up on FCF, it looks like the consensus for max ppd is to run 2 instances with -smp 8. At 3.8 GHz, road runner is reporting almost 10,000 ppd. vs 7850 ppd from someone else on a single instance at 3.8 GHz
 
ChasR said:
Reading up on FCF, it looks like the consensus for max ppd is to run 2 instances with -smp 8. At 3.8 GHz, road runner is reporting almost 10,000 ppd. vs 7850 ppd from someone else on a single instance at 3.8 GHz
Click to expand...

Yes that is with smp 8. It brings my cpu up to 65ish%

I don't think im going to stick another client on this machine. I've got an almost unlimited supply of computers that I can (legitimately) put it on. we the one click version. no gpus allowed, or system tray kind. which is why i've got almost 200 machines folding, but not producing much ppd. if I get 20k a day im happy...for now =D

Whats fcf?
Click to expand...

dito.
 
Methal said:
I don't understand all th workings of cat /proc/cpuinfo. But in my little brain it would make sense if when I "pinged" the cpu it may have caused a hickup in what fah was doing
Click to expand...

When you "cat" the /proc/cpuinfo file you're just reading a plain text file containing information collected about the processor. Bogomips is a tool that measures a CPU's speed in MIPS (millions of instructions per second). This is more of a calculation operation, and does not involve anything proactive such as a 'ping', as used in networking.
 
wow those numbers make me jelous: currently im getting about 1% done every 9 minutes on my desktop, and 1% every 14min on my server giving me only about 4000ppd i think, i havnt been running very long.

This is from my server:
Code: 
[16:45:29] Completed 97500 out of 250000 steps  (39%)
[16:59:48] Completed 100000 out of 250000 steps  (40%)
[17:14:06] Completed 102500 out of 250000 steps  (41%)
[17:28:23] Completed 105000 out of 250000 steps  (42%)
[17:42:40] Completed 107500 out of 250000 steps  (43%)
[17:57:01] Completed 110000 out of 250000 steps  (44%)
[18:11:21] Completed 112500 out of 250000 steps  (45%)
[18:25:35] Completed 115000 out of 250000 steps  (46%)
[18:39:52] Completed 117500 out of 250000 steps  (47%)
[18:54:16] Completed 120000 out of 250000 steps  (48%)

i mite post the output of my desktop later.

if only i could afford an i7 :(
 
Last edited:
You must log in or register to reply here.

Similar threads

Charr
Core 3 is faster?
Replies
3
Views
847
splat
splat
Hack30
floating point exception
Replies
12
Views
1K
Hack30
Hack30
Back