Overclock.net banner

1 - 20 of 24 Posts

·
Premium Member
Joined
·
5,606 Posts
Pretty soon [email protected] will design a new version of the program that will hopefully make it multi threaded...but for now, like Burn said, 2 instances is the only way
 

·
I remember when......
Joined
·
3,812 Posts
We were discussing this in another thread........Try using the graphical program not the console one (or, run 2 of the console ones).

At least on my system, the graphical folding loads both my cores.

I had some screenshots, but I can't get them small enough to attach....
 

·
Premium Member
Joined
·
5,809 Posts
Quote:


Originally Posted by Thumper

I had some screenshots, but I can't get them small enough to attach....

download irfanview....
 

·
I remember when......
Joined
·
3,812 Posts
Here are my screenshots folks

I used Speedfan, task Manager and PC Probe 2 to show CPU usage, so this is three programs saying the same thing.

Idle

1 Prime95

2 prime95s

Folding

Sorry about the quality, I set the resolution low on the captures when I saved them.

But you can clearly see that at idle my system shows almost 0% usage on both CPU cores. With 1 Prime95 program running, 1 core is clearly fully loaded with the other doing some work, but only approx 15%. Then with 2 Prime95 programs running, both cores show 100% load.

Finally, with only one graphical Folding program running, again, both cores are running at 100%.

The only explanation I have is that the graphical folding program is using both my cores.
 

·
Premium Member
Joined
·
9,824 Posts
There's an easy way to test. Place the graphical client and the console in the same directory. Run each version for one hour and check the number of frames completed. Yes, it's totally safe to swap between versions in the same directory; they work off of the exact same work files. Since both versions will be working on the same WU on the same system, that will definitively tell you if the graphical version is squeezing more folding power out of your system.
 

·
Premium Member
Joined
·
9,824 Posts
It's nothing personal. The results you're observing are just in complete disagreement with what Stanford states about their own software. If you have come across a way to force the software to be multithreaded, a bit of evidence to send to the Stanford guys would be handy.
 

·
I remember when......
Joined
·
3,812 Posts
Quote:


Originally Posted by Taeric

It's nothing personal. The results you're observing are just in complete disagreement with what Stanford states about their own software. If you have come across a way to force the software to be multithreaded, a bit of evidence to send to the Stanford guys would be handy.



LOL

I know, I was just poking fun, I'm not insulted or anything. I'm actually upset that I didn't think of your suggestion to try myself.


It should be done with my current WU by the time I get home tonight (last I looked it was supposed to finish in about 30 minutes, and I won't be home for about 3 hours) so I should have a fresh WU with pleanty of room for testing.

I'll take a screenshot of the starting point and ending point for each to show the difference.
 

·
I remember when......
Joined
·
3,812 Posts
UPDATE:

Sorry, couldn't test last night. The parts for my sister's computer arrived and I wanted to get that build done. best way to relax after a terrible day in a cubicle is to get your hands inside a computer


And today I have to haul stuff around for my family (have truck, will move
)

Tomorrow night I should have time to sit down and try it.......

BTW, does the console version even tell you how many frames you have completed?? I didn't see it.....
 

·
Premium Member
Joined
·
9,824 Posts
The log file for both version are the same. You can check the percent complete in the log; the number in the parenthesis is what you're looking for. Since all you're looking for is a comparison, you can maybe let each version run for 5-10% of a WU and see the difference between them. As a minor caution with that - the computation speed within some WUs varies over the course of the WU (e.g. the first 20% is faster than the rest). You may want to check this on 2-3 different WUs. The nice thing about this is you can always go back and mine info from your current log file, so you only have to check out the other version of the program.

As an example...

[11:12:21] Completed 11500000 out of 50000000 steps (23)
[11:54:45] Writing local files
[11:54:45] Completed 12000000 out of 50000000 steps (24)
[12:37:13] Writing local files
[12:37:13] Completed 12500000 out of 50000000 steps (25)
 

·
I remember when......
Joined
·
3,812 Posts
Quote:


Originally Posted by Taeric

Since all you're looking for is a comparison, you can maybe let each version run for 5-10% of a WU and see the difference between them. As a minor caution with that - the computation speed within some WUs varies over the course of the WU (e.g. the first 20% is faster than the rest). You may want to check this on 2-3 different WUs.


Well, here's test #1.......

not sure why the times are funky in the log........I ran both for exactly 2 hours based on the Windows clock, which matches my other clocks......I have screenshots and I'll cut and past the log file.......

First, 2 hours using the graphical version.......

Graphical Start

Graphical End

...and the log....


[14:40:39] - Ask before connecting: No
[14:40:39] - Use IE connection settings: Yes
[14:40:39] - User name: [email protected] (Team 37726)
[14:40:39] - User ID: 17187AD630DC6997
[14:40:39] - Machine ID: 1
[14:40:39]
[14:40:39] Loaded queue successfully.
[14:40:39] Initialization complete
[14:40:39] + Benchmarking ...
[14:40:43]
[14:40:43] + Processing work unit
[14:40:43] Core required: FahCore_78.exe
[14:40:43] Core found.
[14:40:43] Working on Unit 02 [March 20 14:40:43]
[14:40:43] + Working ...
[14:40:43]
[14:40:43] *------------------------------*
[14:40:43] [email protected] Gromacs Core
[14:40:43] Version 1.86 (August 28, 2005)
[14:40:43]
[14:40:43] Preparing to commence simulation
[14:40:43] - Looking at optimizations...
[14:40:43] - Files status OK
[14:40:44] - Expanded 851307 -> 4394946 (decompressed 516.2 percent)
[14:40:44]
[14:40:44] Project: 761 (Run 53, Clone 84, Gen 4)
[14:40:44]
[14:40:44] Assembly optimizations on if available.
[14:40:44] Entering M.D.
[14:41:04] (Starting from checkpoint)
[14:41:04] Protein: p761_vln3.3pbc
[14:41:04]
[14:41:04] Writing local files
[14:41:04] Completed 60000 out of 500000 steps (12)
[14:41:05] Extra SSE boost OK.
[14:53:34] Writing local files
[14:53:34] Completed 65000 out of 500000 steps (13)
[15:06:04] Writing local files
[15:06:04] Completed 70000 out of 500000 steps (14)
[15:18:29] Writing local files
[15:18:29] Completed 75000 out of 500000 steps (15)
[15:30:48] Writing local files
[15:30:48] Completed 80000 out of 500000 steps (16)
[15:43:03] Writing local files
[15:43:03] Completed 85000 out of 500000 steps (17)
[15:55:14] Writing local files
[15:55:14] Completed 90000 out of 500000 steps (18)
[16:07:32] Writing local files
[16:07:32] Completed 95000 out of 500000 steps (19)
[16:19:50] Writing local files
[16:19:50] Completed 100000 out of 500000 steps (20)
[16:32:04] Writing local files
[16:32:04] Completed 105000 out of 500000 steps (21)
[16:44:22] Writing local files
[16:44:22] Completed 110000 out of 500000 steps (22)
[16:56:47] Writing local files
[16:56:47] Completed 115000 out of 500000 steps (23)
[17:09:06] Writing local files
[17:09:06] Completed 120000 out of 500000 steps (24)
[17:21:22] Writing local files
[17:21:22] Completed 125000 out of 500000 steps (25)
[17:33:29] Writing local files
[17:33:29] Completed 130000 out of 500000 steps (26)
[17:45:30] Writing local files
[17:45:30] Completed 135000 out of 500000 steps (27)
[17:57:30] Writing local files
[17:57:30] Completed 140000 out of 500000 steps (28)
[18:09:31] Writing local files
[18:09:31] Completed 145000 out of 500000 steps (29)
[18:21:35] Writing local files
[18:21:35] Completed 150000 out of 500000 steps (30)
[18:33:39] Writing local files
[18:33:39] Completed 155000 out of 500000 steps (31)
[18:45:42] Writing local files
[18:45:42] Completed 160000 out of 500000 steps (32)
[18:57:45] Writing local files
[18:57:45] Completed 165000 out of 500000 steps (33)
[19:09:49] Writing local files
[19:09:49] Completed 170000 out of 500000 steps (34)
[19:21:52] Writing local files
[19:21:52] Completed 175000 out of 500000 steps (35)
[19:33:56] Writing local files
[19:33:56] Completed 180000 out of 500000 steps (36)
[19:46:01] Writing local files
[19:46:01] Completed 185000 out of 500000 steps (37)
[19:58:05] Writing local files
[19:58:05] Completed 190000 out of 500000 steps (38)
[20:10:09] Writing local files
[20:10:09] Completed 195000 out of 500000 steps (39)

[email protected] Client Shutdown.


and now 2 hours using the console version.....

Console Start

Console End

and this is the log from using the console version


[20:30:50] - Ask before connecting: No
[20:30:50] - Use IE connection settings: Yes
[20:30:50] - User name: [email protected] (Team 37726)
[20:30:50] - User ID: 17187AD630DC6997
[20:30:50] - Machine ID: 1
[20:30:50]
[20:30:50] Loaded queue successfully.
[20:30:50] + Benchmarking ...
[20:30:52]
[20:30:52] + Processing work unit
[20:30:52] Core required: FahCore_78.exe
[20:30:52] Core found.
[20:30:52] Working on Unit 02 [March 20 20:30:52]
[20:30:52] + Working ...
[20:30:52]
[20:30:52] *------------------------------*
[20:30:52] [email protected] Gromacs Core
[20:30:52] Version 1.86 (August 28, 2005)
[20:30:52]
[20:30:52] Preparing to commence simulation
[20:30:52] - Looking at optimizations...
[20:30:52] - Files status OK
[20:30:53] - Expanded 851307 -> 4394946 (decompressed 516.2 percent)
[20:30:53]
[20:30:53] Project: 761 (Run 53, Clone 84, Gen 4)
[20:30:53]
[20:30:53] Assembly optimizations on if available.
[20:30:53] Entering M.D.
[20:31:14] (Starting from checkpoint)
[20:31:14] Protein: p761_vln3.3pbc
[20:31:14]
[20:31:14] Writing local files
[20:31:14] Completed 197482 out of 500000 steps (39)
[20:31:14] Extra SSE boost OK.
[20:37:18] Writing local files
[20:37:18] Completed 200000 out of 500000 steps (40)
[20:49:14] Writing local files
[20:49:15] Completed 205000 out of 500000 steps (41)
[21:01:11] Writing local files
[21:01:11] Completed 210000 out of 500000 steps (42)
[21:13:08] Writing local files
[21:13:08] Completed 215000 out of 500000 steps (43)
[21:25:05] Writing local files
[21:25:05] Completed 220000 out of 500000 steps (44)
[21:37:02] Writing local files
[21:37:02] Completed 225000 out of 500000 steps (45)
[21:48:59] Writing local files
[21:48:59] Completed 230000 out of 500000 steps (46)
[22:00:56] Writing local files
[22:00:56] Completed 235000 out of 500000 steps (47)
[22:12:53] Writing local files
[22:12:53] Completed 240000 out of 500000 steps (48)
[22:24:50] Writing local files
[22:24:50] Completed 245000 out of 500000 steps (49)

[email protected] Client Shutdown.


I can't explain the time difference in the log. As you can see from my screenshots, the console test ended at 2:30, yet the log states it ended at 10:24........

I'll run some more tests, but it'll be at least 2 weeks as I am going to be spending the next 2 weeks moving.
 

·
Premium Member
Joined
·
9,824 Posts
The log file is based on GMT if that helps.
 

·
I remember when......
Joined
·
3,812 Posts
Quote:


Originally Posted by Taeric

The log file is based on GMT if that helps.

It does.

Well, then according to that data, you are right, they fold at the exact same rate, which means the graphical can't be using both cores.

But I would love an explanation from Stanford as to why it spools both my cores up, especially as it obviously causes heat buildup.

I happen to like the graphical look, but if it's burning up my CPU for no reason, I guess it's wasteful.
 
1 - 20 of 24 Posts
Top