We're not talking the lost time, but the reason described and long discussed of the time that goes towards IO and why WCG changed the default work distribution for CEP2: https://secure.worldcommunitygrid.org/forums/...ad,30161_offset,50#300323

No, I'm talking about the 30-45 minutes most everyone experiences with CEP2 on Linux. The USB IO is another fish to fry... covered in the next post, nearly ready to post... waiting on completion of more C4CW tasks and an HFCC.

Sticking to the topic would be a major breath of fresh air... this one is about poor crunching (efficiency).

Intermediate data is building... I'm hopeful to bring all sciences back to the 97-99%+ and CEP2 to 95% instead of 90-92%. The WTD extending helped already quite allot.

VM and RAM disk are from what I've read no big issue.... you may want to research that and add that to the pool of knowledge... substantially, something needs to back up what's in there on a periodic basis or booting becomes quite costly in progress loss... even for 24/7 crunchers.

Though all avenues on the swap file angle has not paid any dividend for crunching, it has made my quad system more user-responsive not having any, but that is a Linux thing. For Windows a VM is almost a must (W7 and maybe Vista and earlier) since MS uses strategies such as writing portions of RAM retained data to VM, so it can literally unmap it from RAM instantenously and point it to the VM mirror when RAM is actually needed for something else.

My CEP2 usage:
I7 with 6GB = 95%-97%
Q6600 with 2GB = 90-92%

Sek, what's the amount ram on your rig? With your numbers and mine it seems more ram means higher efficiency.

Some more "sysstat" performance data. No significant disk I/O bottlenecks detected.


Last CEP2 job completed on this device:

CPU time: 12 Hours
Elapsed time: 12.4 Hours

<result>
<name>E200486_431_A.25.C21H13NOS2.127.3.set1d06_0</name>
<final_cpu_time>43200.006800</final_cpu_time>
<final_elapsed_time>44937.790886</final_elapsed_time>
<exit_status>0</exit_status>
<state>5</state>
<platform>i686-pc-linux-gnu</platform>
<version_num>619</version_num>
<stderr_out>

Job starts:

[07:26:42] Starting job 0,CPU time has been restored to 0.000000.
[07:29:07] Starting job 1,CPU time has been restored to 116.610272.
[07:36:38] Starting job 2,CPU time has been restored to 537.081350.
[10:47:09] Starting job 3,CPU time has been restored to 11198.435577.
[10:55:36] Starting job 4,CPU time has been restored to 11667.995193.
[11:03:05] Starting job 5,CPU time has been restored to 12091.264846.
[11:10:58] Starting job 6,CPU time has been restored to 12535.115370.
[11:18:36] Starting job 7,CPU time has been restored to 12967.189684.
[11:29:58] Starting job 8,CPU time has been restored to 13621.074278.
[11:37:24] Starting job 9,CPU time has been restored to 14046.812555.
[11:45:53] Starting job 10,CPU time has been restored to 14528.093389.
[12:06:55] Starting job 11,CPU time has been restored to 15741.876865.
[12:17:13] Starting job 12,CPU time has been restored to 16320.548893.
[13:06:41] Starting job 13,CPU time has been restored to 19164.929481.
[15:30:07] Starting job 14,CPU time has been restored to 27527.013251.
[17:39:39] Starting job 15,CPU time has been restored to 35087.966811.
Killing job because cpu time has been exceeded. Subjob start time = -266494131, Subjob current time = 1088496126
[19:55:25] Finished Job #15

Performance log entries (same time or very close to same) for job start times listed above:

# sar -u

07:17:01 AM     CPU     %user     %nice   %system   %iowait    %steal     %idle
07:17:01 AM     all      0.03     98.22      1.76      0.00      0.00      0.00
07:27:01 AM     all      0.06     96.53      1.58      1.30      0.00      0.53
07:37:01 AM     all      0.03     91.82      7.06      0.22      0.00      0.87
10:47:01 AM     all      0.03     93.81      6.15      0.01      0.00      0.00
10:57:01 AM     all      0.02     92.82      6.23      0.13      0.00      0.79
11:07:01 AM     all      0.02     94.64      4.78      0.20      0.00      0.37
11:17:01 AM     all      0.03     93.87      5.42      0.17      0.00      0.51
11:27:01 AM     all      0.03     95.86      3.66      0.02      0.00      0.43
11:37:01 AM     all      0.03     95.79      3.69      0.05      0.00      0.44
11:47:01 AM     all      0.02     94.22      4.61      0.21      0.00      0.93
12:07:01 PM     all      0.02     95.95      3.50      0.16      0.00      0.36
12:17:01 PM     all      0.02     94.12      5.70      0.15      0.00      0.00
12:27:01 PM     all      0.03     98.20      1.17      0.34      0.00      0.26
12:37:01 PM     all      0.02     95.36      4.50      0.12      0.00      0.00
01:07:01 PM     all      0.02     97.90      1.62      0.01      0.00      0.45
03:27:01 PM     all      0.03     98.41      1.55      0.01      0.00      0.00
03:37:01 PM     all      0.02     95.99      3.50      0.00      0.00      0.49
05:37:01 PM     all      0.02     98.44      1.54      0.00      0.00      0.00
05:47:01 PM     all      0.02     93.57      5.60      0.36      0.00      0.45
07:47:02 PM     all      0.03     94.01      5.96      0.00      0.00      0.00
07:57:02 PM     all      0.07     91.93      5.71      1.73      0.00      0.57

The below averages are for all the performance recorded entries since midnight:

08:07:01 PM     CPU     %user     %nice   %system   %iowait    %steal     %idle
Average:        all      0.03     96.05      3.77      0.07      0.00      0.08

Swap pages in/out:
# sar -W

08:07:02 PM  pswpin/s pswpout/s
Average:         0.00      0.00

Disk I/O performance:
# sar -d

08:07:02 PM       DEV       tps  rd_sec/s  wr_sec/s  avgrq-sz  avgqu-sz     await     svctm     %util
Average:       dev8-0      1.37      0.07    721.30    526.90      0.43    315.47     11.40      1.56

Memory performance:
# sar -r

08:07:02 PM kbmemfree kbmemused  %memused kbbuffers  kbcached  kbcommit   %commit
Average:       638499   1409089     68.82    259136    885776    351947      8.48

Interesting, %memused (68.82%) for all system/user processes is less than the 75% max memory setting in the BOINC Manager while in use. The CEP2 system requirements say "1,024 MB Memory Available". I wonder if I reduce system RAM from 2 GB to 1.5 GB what effect that would have on SWAP usage and any increased disk I/O activity bottleneck.