SMP Project 2605 - Post Your Run Times

[Apoptosis]

Just fired up some F@H on the old V8 system and got project 2605 to work on. Anyone else currently running this? If you are please post up a small section of the log, so I can see how long it takes to complete a step. Be sure to include the CPU info and memory amount!

[Apoptosis]

I'm running the Intel V8 test system at default settings.

intel_v8.jpg (161.97 KiB) Viewed 3264 times

System Specifications:
Dual Quad-Core Intel Xeon Processors X5365 (3GHz/1333MHzFSB)
Intel Workstation Board S5000XVN
12GB of Total Memory:
4 x 2GB Kingston 667Mhz ECC FB-DIMM Memory Modules
4 x 1GB Kingston 667Mhz ECC FB-DIMM Memory Modules
1KW PC Power and Cooling PSU
ATI Radeon HD 3450 GPU

It's running Linux with FAH version 6 Beta 2 and looks like it is taking 8 minutes 20 seconds or so to finish each step. Both clients are turning in the same times. Oh and just for fun I plugged in the power meter and the system is bouncing around 470-480W while running F2H on all eight cores. The system is using a passive cooled Radeon 3450 graphics card too...

CLIENT 1:

[17:20:45]
[17:20:45] Project: 2605 (Run 10, Clone 259, Gen 31)
~~~~~~~~~
[17:21:09] Completed 0 out of 500000 steps (0 percent)
[17:29:34] Writing local files
[17:29:34] Completed 5000 out of 500000 steps (1 percent)
[17:37:58] Writing local files
[17:37:58] Completed 10000 out of 500000 steps (2 percent)
[17:46:18] Writing local files
[17:46:18] Completed 15000 out of 500000 steps (3 percent)
[17:54:38] Writing local files
[17:54:38] Completed 20000 out of 500000 steps (4 percent)
[18:02:56] Writing local files
[18:02:56] Completed 25000 out of 500000 steps (5 percent)
[18:11:13] Writing local files
[18:11:13] Completed 30000 out of 500000 steps (6 percent)
[18:19:33] Writing local files
[18:19:33] Completed 35000 out of 500000 steps (7 percent)
[18:27:57] Writing local files
[18:27:57] Completed 40000 out of 500000 steps (8 percent)
[18:36:20] Writing local files
[18:36:20] Completed 45000 out of 500000 steps (9 percent)
[18:44:39] Writing local files
[18:44:39] Completed 50000 out of 500000 steps (10 percent)

CLIENT 2

[17:20:50] Project: 2605 (Run 10, Clone 260, Gen 31)
~~~~~~~~~~~~~~~
[17:21:13] Completed 0 out of 500000 steps (0 percent)
[17:29:29] Writing local files
[17:29:29] Completed 5000 out of 500000 steps (1 percent)
[17:37:48] Writing local files
[17:37:48] Completed 10000 out of 500000 steps (2 percent)
[17:46:12] Writing local files
[17:46:12] Completed 15000 out of 500000 steps (3 percent)
[17:54:32] Writing local files
[17:54:32] Completed 20000 out of 500000 steps (4 percent)
[18:02:50] Writing local files
[18:02:50] Completed 25000 out of 500000 steps (5 percent)
[18:11:07] Writing local files
[18:11:07] Completed 30000 out of 500000 steps (6 percent)
[18:19:28] Writing local files
[18:19:28] Completed 35000 out of 500000 steps (7 percent)
[18:27:51] Writing local files
[18:27:51] Completed 40000 out of 500000 steps (8 percent)
[18:36:15] Writing local files
[18:36:15] Completed 45000 out of 500000 steps (9 percent)
[18:44:37] Writing local files
[18:44:37] Completed 50000 out of 500000 steps (10 percent)

[Bwall]

Impressive system but I really thought it would be faster. I'll get my folding log tonight to see what project my Linux client was working on, here is what I have off the top of my head.

E6700 @3.2GHz 2GB DDR2 5-5-5-15 @ ~1000MHz-- Runs it right at 10min/percent (whatever 2394ppd works out to be)

E8400@4GHz 2GB DDR 2 4-4-4-12@ 890MHz-- 7:57min/percent....so the E8400 overclocked is a little faster than a 3GHz Xeon in a program that supports 4 cores??! I would estimate it runs about the same speed as a 3.6GHz E8400....maybe I can put my lower clocked E8400 to the test tonight.

I hope we see some more results from others!

[dicecca112]

8 Minutes is damn fast compared to the 17-18 a 2.0Ghz Merom gets

[Bwall]

Ok, finally home.

Project 2605
E6700 @ 3.2 GHz - 10:21/percent = 2448 ppd
E8400 @ 4GHz -7:54/percent=3208 ppd

[Apoptosis]

I took out 4GB of memory, so the system is running 8GB and not 12GB now... Those FB-DIMM modules get VERY VERY hot and I couldn't even hold them after they were folding for 36 hours...

Fired up the system again and got another 2605 Work Unit...

[04:26:42] Project: 2605 (Run 6, Clone 235, Gen 32)
~~~~~~~~~~~~~~~~
[04:26:50] Completed 0 out of 500000 steps (0 percent)
[04:35:15] Writing local files
[04:35:15] Completed 5000 out of 500000 steps (1 percent)
[04:43:35] Writing local files
[04:43:35] Completed 10000 out of 500000 steps (2 percent)
[04:51:54] Writing local files
[04:51:54] Completed 15000 out of 500000 steps (3 percent)
[05:00:11] Writing local files
[05:00:11] Completed 20000 out of 500000 steps (4 percent)
[05:08:28] Writing local files
[05:08:28] Completed 25000 out of 500000 steps (5 percent)
[05:16:48] Writing local files
[05:16:48] Completed 30000 out of 500000 steps (6 percent)
[05:25:07] Writing local files
[05:25:07] Completed 35000 out of 500000 steps (7 percent)
[05:33:25] Writing local files
[05:33:25] Completed 40000 out of 500000 steps (8 percent)
[05:41:45] Writing local files
[05:41:45] Completed 45000 out of 500000 steps (9 percent)
[05:50:06] Writing local files
[05:50:06] Completed 50000 out of 500000 steps (10 percent)

Looks like 8min 17 seconds per percent, so going from 12GB of memory to 8GB of memory had no impact on folding performance. It really reduced the heat and I am guessing the total power consumption at load. Before it was 480W and the next time I restart it I'll check again.

[Apoptosis]

Power Consumption with the same work units is now 430W

By removing four 1GB Samsung 667MHz FB-DIMM modules the system went from 480W at load to 430W... 50W for four 1GB modules... So each modules consumes 12.5W?? No wonder those memory modules get so hot...

[KnightRid]

Power Consumption with the same work units is now 430W

By removing four 1GB Samsung 667MHz FB-DIMM modules the system went from 480W at load to 430W... 50W for four 1GB modules... So each modules consumes 12.5W?? No wonder those memory modules get so hot...

Holy cow! I wonder if more memory on the chip relates to more power and heat needed, as the reverse would then be lower power and heat. So it bears the question...

Is it better to have 2 1gb chips in a 2gb memory computer, or 4 512mb chips..things that make you go hmmmmmm

Mike

Is an interesting concept though - will 2 512mb chips use less power and create less heat than a 1gb chip.. curious minds want to know