Hope this helps. I've always run stock clients/apps.

GenuineIntel
Intel(R) Pentium(R) 4 CPU 2.80GHz Microsoft Windows XP
Home Edition, Service Pack 2, (05.01.2600.00)
Hyperthreaded

Running BOINC 5.4.9

SETI Enhanced
CPU time (sec)-------claimed credit
21,944.61------------31.63
6,315.30-------------14.05
36,626.88------------64.85

Einstein
CPU time (sec) ------claimed credit
29,327.67------------39.62
31,084.58------------42.00
29,049.84------------39.25
30,188.92------------40.79
28,848.78------------39.01
29,881.89------------40.41
29,218.27------------39.51
29,807.89------------40.31

This machine also runs SETI Beta, Rosetta, CPDN, and Pirates.

GenuineIntel
Intel(R) Pentium(R) M processor 1400MHz Microsoft Windows XP
Professional Edition, Service Pack 2, (05.01.2600.00)

Running BOINC 5.4.9
SETI Enhanced
CPU time (sec) ------claimed credit
25,551.73------------61.77

I have 3 other results on this machine with -9 errors. Won't include those here.

Einstein
CPU time (sec) ------claimed credit
24,875.01------------53.85
25,051.37------------54.23
25,409.01------------55.01
25,856.38------------55.98
26,812.03------------54.49
25,606.02------------52.04
27,857.62------------56.61
25,333.60------------55.49
25,345.20------------55.52
25,496.37------------55.85

Predictor
CPU time (sec) ------claimed credit
4,614.47-------------8.86
4,221.85-------------8.11
4,981.66-------------9.90
4,343.06-------------8.63

Rosetta
CPU time (sec) ------claimed credit
14,121.92------------19.08
13,347.08------------18.03
10,999.77------------14.86
16,102.42------------21.75
13,995.20------------18.91
16,898.88------------22.83
14,105.61------------19.08
14,134.56------------19.12
12,268.56------------16.59

Ufluids
CPU time (sec) ------claimed credit
25,635.11------------55.50
20,385.80------------41.43
47,924.39------------97.39
46,203.16------------98.94
140,792.49-----------301.12

World Community Grid
Averages:
Avg. Run Time Per Calendar Day (y:d:h:m:s) 0:000:03:28:01
Avg. Run Time Per Result (y:d:h:m:s) 0:000:06:22:45
Avg. Points Per Hour of Run Time 54.12534
Avg. Points Per Calendar Day 187.65217
Avg. Points Per Result 345.28000
Avg. Results Per Calendar Day 0.54348

---

Wendy



Click Here for BOINC FAQ Service

My opinion only : It was'nt broken so why did Berkeley fix it?

If you had been here for any length of time you'd know the amount of noise & carry on the benchmark based credit granting caused.
It makes the present Cruch3r/Tony episode just a minor blip in comparison.

---

Grant
Darwin NT

Anyone able to recommend a good logger to pull this info from each host as WUs are completed?

---

Anyone able to recommend a good logger to pull this info from each host as WUs are completed?

SetiMapView and BOINCLogX come to mind. I havent tried them yet, but apparently you are able to save the results in in a .csv format to export to Excel. (correct?)

Machine                 CPDN    EAH    PAH    SAH    LHC   Rose    SIMAP   SZTAKI   uFluids

2.66GHz Xeon HT         10.13   4.33   6.50   4.07   5.88  4.45    7.44    4.55     5.51

3.4GHz P4 HT             9.75   --     --      --    --    6.90    7.55    6.39     7.07

2.8GHz P4 HT            10.55   5.08   7.07   7.12   5.51  5.72    5.81    5.03     5.89

1.73GHz Pentium M        --     9.32   13.15  9.75   8.77   --      --      --       --

2x2.8GHz Pentium D       --     7.96   10.40  11.80  7.66   --      --      --       --

2.2GHz Celeron M         --     6.42   4.91   4.39   7.20   --      --      --       --

2GHz Celeron             --     6.73   3.89   5.67   6.20   --      --      --       --

3Ghz P4 HT               --     7.15   7.23    --    5.39   --      --      --       --

edit: Thanks cdr100560.

---

BBcode format tags of [ pre][ /pre] is what you're after :)

You missed this part in Erics request

But please only a stock BOINC core client and stock applications. Using optimized core clients or applications would throw off the results.

Having you information is nice, but what is really needed is data from stock clients with stock apps.

Hi Robert:

The data was compiled mainly for my purposes so that I could see what all the fuss is about and draw my own conclusions.

Limiting the analysis only to stock clients and stock apps simply won’t give you the true picture of the credit problem. It will only give you the answer you want to hear.

Regards
Franz

hi web03 and others,

SETI Enhanced
CPU time (sec)-------claimed credit
[...]

Claimed credits do not help very much. The granted credits are the important numbers.

Regards,
Carsten

Hi Franz,

Limiting the analysis only to stock clients and stock apps simply won't give you the true picture of the credit problem. It will only give you the answer you want to hear.

You still deny the fact that the users of optimized applications are a minority. You simply cannot adjust the credit claims to the demands of a minority. The result would be an overclaiming of the mass. This would mean that the whole project cheats the other projects.

Regards,
Carsten

Anyone able to recommend a good logger to pull this info from each host as WUs are completed?

SetiMapView and BOINCLogX come to mind. I havent tried them yet, but apparently you are able to save the results in in a .csv format to export to Excel. (correct?)

Yes, BoincLogX saves most of the information available on the host to .csv files. Its History display shows part of that, the user_total_credit is there as well as the cpu time for each WU, so you can get a credits/time value by comparing changes in total credit to the sum of cpu times for a given period. Doing that for recent work may give erroneous values because of pending credit, but works well for extended periods.

*************
What's really needed is for the stats from the projects to include total CPU time as well as total credit for each host. Then it would be a simple matter to compare credit/time across projects.

                                            Joe

Carsten -

In my case for SETI Enhanced, claimed and granted are identical. Just let me know if you need anything else. I just downloaded BoinclogX to log my projects on my main computer.

hi web03 and others,

SETI Enhanced
CPU time (sec)-------claimed credit
[...]

Claimed credits do not help very much. The granted credits are the important numbers.

Regards,
Carsten

---

Wendy



Click Here for BOINC FAQ Service

Hi web03,

In my case for SETI Enhanced, claimed and granted are identical.

I'm not surprised ... ;-) Maybe I should have cited another line from your original message ... ;-)

Just let me know if you need anything else.

Eric needs the granted credits of the other projects. These numbers indicate what you really get from these projects (you can claim anything, but what you really get for your work is important). Then he can decide if the credit system here at seti needs some adjustment.

Regards,
Carsten

Hope this helps. I've always run stock clients/apps.

GenuineIntel
Intel(R) Pentium(R) 4 CPU 2.80GHz Microsoft Windows XP
Home Edition, Service Pack 2, (05.01.2600.00)
Hyperthreaded

Running BOINC 5.4.9

SETI Enhanced
CPU time (sec)-------claimed credit
21,944.61------------31.63
6,315.30-------------14.05
36,626.88------------64.85

Einstein
CPU time (sec) ------claimed credit----------granted credit
29,327.67------------39.62-------------------pending
31,084.58------------42.00-------------------42.00
29,049.84------------39.25-------------------34.71
30,188.92------------40.79-------------------30.12
28,848.78------------39.01-------------------pending
29,881.89------------40.41-------------------pending
29,218.27------------39.51-------------------pending
29,807.89------------40.31-------------------pending

This machine also runs SETI Beta, Rosetta, CPDN, and Pirates.

GenuineIntel
Intel(R) Pentium(R) M processor 1400MHz Microsoft Windows XP
Professional Edition, Service Pack 2, (05.01.2600.00)

Running BOINC 5.4.9
SETI Enhanced
CPU time (sec) ------claimed credit
25,551.73------------61.77

I have 3 other results on this machine with -9 errors. Won't include those here.

Einstein
CPU time (sec) ------claimed credit----------granted credit
24,875.01------------53.85-------------------37.82
25,051.37------------54.23-------------------37.83
25,409.01------------55.01-------------------48.80
25,856.38------------55.98-------------------31.17
26,812.03------------54.49-------------------54.49
25,606.02------------52.04-------------------pending
27,857.62------------56.61-------------------51.17
25,333.60------------55.49-------------------49.05
25,345.20------------55.52-------------------44.78
25,496.37------------55.85-------------------pending

Predictor
CPU time (sec) ------claimed credit----------granted credit
4,614.47-------------8.86--------------------14.87
4,221.85-------------8.11--------------------17.56
4,981.66-------------9.90--------------------11.68
4,343.06-------------8.63--------------------11.59

Rosetta
CPU time (sec) ------claimed credit----------granted credit
14,121.92------------19.08 ------------------19.08
13,347.08------------18.03 ------------------18.03
10,999.77------------14.86 ------------------14.86
16,102.42------------21.75 ------------------21.75
13,995.20------------18.91 ------------------18.91
16,898.88------------22.83 ------------------22.83
14,105.61------------19.08 ------------------19.08
14,134.56------------19.12 ------------------19.12
12,268.56------------16.59 ------------------16.59

Ufluids
CPU time (sec) ------claimed credit----------granted credit
25,635.11------------55.50-------------------36.52
20,385.80------------41.43-------------------41.43
47,924.39------------97.39-------------------63.81
46,203.16------------98.94-------------------0 - Client Error
140,792.49-----------301.12------------------result purged from website.

World Community Grid
Averages:
Avg. Run Time Per Calendar Day (y:d:h:m:s) 0:000:03:28:01
Avg. Run Time Per Result (y:d:h:m:s) 0:000:06:22:45
Avg. Points Per Hour of Run Time 54.12534
Avg. Points Per Calendar Day 187.65217
Avg. Points Per Result 345.28000
Avg. Results Per Calendar Day 0.54348

Added Granted Credits above.

---

Wendy



Click Here for BOINC FAQ Service

Okay, here's some data:

                                Avg Time & Credit/Hour
  Machine  | Project  | App |  Seconds  | Claim | Grant

P4 2.6GHz  | QMC      | Std | 87,487.16 |  7.38 |  7.38 
XP Home    | SETI     | Std |  7,983.88 |  7.29 |  7.29

P4 3.0GHz  | SZTAKI   | Std |  3,009.58 |  9.16 |  7.44
XP Pro     | SETI     | Std | 21,889.05 |  7.94 |  7.94

P4 3.06GHz | Leiden   | Std |  1,208.60 |  9.29 |  5.63
XP Home    | SETI     | Std | 23,139.72 |  6.60 |  6.60

P4 3.4GHz  | Einstein | Std | 22,476.03 | 10.23 |  8.52
XP Pro     | SETI     | Std | 18,068.37 |  7.77 |  7.77

P4 3.4GHz  | Einstein | Opt |  4,416.78 | 10.39 | 26.69
XP Pro     | SETI     | Opt | 11,173.16 | 14.46 | 14.46

Each of these machines is configured with HT on, and using the 5.3.12.tx37 BOINC client (with calibration disabled) configured to set CPU affinity and run each project on a separate (logical) CPU, with identical resource shares. So, this should be a reasonably good "apples to apples" comparison based on CPU time.

Also, the bottom two machines are identical in every way, with the exception that the second one runs optimized applications and is generally more heavily utilized than the first. The claimed/granted credit for the optimized Einstein client is a reasonably good example of the need for client calibration (which I intend to re-enable for E@H on that particular machine as soon as I post this message).

Hope it helps!

n.b. No, I don't know why the P4/2.6 has such short SETI WU times, unless it's just been lucky enough to get a run of "sweet" work units, or else QMC accesses memory or processes in a way differently enough to avoid the normal hit that results from HyperThreading (or perhaps some of the games - it's my wife's machine - played on it mess with the clock in some manner). However, even though the time reported seems short, the credit per hour seems reasonably in line with the other machines (when comparing with the 3.06, the 2.6 is an 800MHz FSB as opposed to the 533MHz FSB on the 3.06 machine).

n.b.2 Why does the BBCode "code" parameter not seem to be implemented on any of these boards? It would be much easier to format text than the way the "pre" command inserts extra linefeeds.

Data for a P4 3.2GHz HT (limited to 1 cpu for BOINC) Xp Pro SP2, mgr 5.4.9, std apps except Einstein as noted. I don't have any records left of the Einstein wus crunched with the std app but I remember credit being higher with them. I could go back to the std app and crunch a few to check.

Project         # Wus  Ave. Granted/Hr
CPDN-Slab        3     17.56
CPDN-Sulphur   < 1     17.76
CPDN-TCM       < 1     13.73
CP-Seasonal      1     12.71
Einstein         1     10.27  using 4.50, "official beta" optimized app
LHC              3      9.58
Rosetta          3      9.78
Fluids           4      9.31
Predictor        8     10.44
QMC             10      9.57
WCG/FAAH         3      9.99
Seti             2     10.09
Seti Beta        3     11.69  all crunched with 5.15

---

P3 925 MHz
Seti 4.18 - 4.766 cr/hr (extrapolated from Beta forum_thread.php?id=434#4562
Seti Beta windows V5.12 7.82 cr/hr
Seti Enhanced between 3.3 and 4.8 cr/hr, average 4.22 cr/hr

Einstein,
Einstein (v 4.79) units 3.95 cr/hr
Albert (v 4.37) units 3.86 cr/hr

Pent M 1.86
Seti Beta windows ver 5.12
AR=0.53, 29.53 cr/hr
AR=2.8, 36.08 cr/hr

Seti Enhanced
common ar 14.42 cr/hr
High ar 17.92 cr/hr
Low ar all errored out

Einstein
Einstein (v 4.79) units 12.42 cr/hr
Albert (v 4.37) units 15.7 cr/hr

CPDN Sulphur 17.19 cr/hr

Andy

Hi all,


Just a little something to add to the mix:

Here's some results from my old but faithful P3/550MHz:

335616368 80467894 2 Jun 2006 11:25:20 UTC 6 Jun 2006 13:29:08 UTC Over Success Done 117,268.47 52.98 52.61
333544562 79996379 29 May 2006 1:11:57 UTC 31 May 2006 8:57:39 UTC Over Success Done 111,663.89 62.24 62.24
330392411 79258357 22 May 2006 17:29:39 UTC 28 May 2006 16:15:46 UTC Over Success Done 125,287.00 67.23 67.21


So, that's around 120,000 seconds (33 hrs) to gain 60 odd credits. That's 2 credits per hour......time it got retired from SETI and instead used on another project with less intensive workloads.... ;-)


These are using crunch3r's v5.12 opti-client - if I didn't use this, it would be even worse - I've currently got one WU that has a "to completion time" of 42 hours....I thought WU's were going to be issued dependant on the CPU "crunching power" - so low MHz CPU's get the "easier" workload and "beefier" CPU's get the more "intensive" WU's..... oh well....!

---

regards,
Tim
Founder, UK BOINC Team
Join us @ UK BOINC Team: http://www.ukboincteam.org.uk/newforum

I've been crunching some Rosetta@Home work recently and here are 5 results:

CPU Time    Credits   Credits/hour
10,113      85.48     30.43
10,633      89.88     30.43
11,080      93.66     30.43
10,311      87.15     30.43
10,221      86.40     30.43

Here's some previous SETI@Home Enhanced work (using Crunch3r's 5.12 app):

CPU Time    Credits   Credits/hour     AR       Est Std App
 7,755      31.11     14.44            0.8098    8.66
 9,027      34.41     13.72            0.7181    8.23
 9,928      49.25     17.86            0.6029   10.72
12,019      62.82     18.82            0.4259   11.29
 3,589      15.47     15.52            0.0439    9.31

I also estimated what the Credits/hour would be for the standard application.

The system is a Pentium D830 with 2GB of memory, PCI Express, and SATA drives.

Updating the information with Rosetta@home information:

CPU Time   Credits     Credits/Hr

 9,644     81.52       30.43
 9,980     84.36       30.43
10,783     91.15       30.43
10,322     87.24       30.43

---

Eric here are some of my latest results from Predictor.

Pappa did say you looked at my standard seti results before anyhow.


6 Jun 2006 21:39:44 UTC Over Success Done 6,070.90 12.41 11.21

6 Jun 2006 21:39:58 UTC Over Success Done 6,120.00 12.51 12.51

6 Jun 2006 21:39:44 UTC Over Success Done 6,059.45 12.39 18.98

5 Jun 2006 21:19:48 UTC Over Success Done 6,157.00 12.59 12.59

5 Jun 2006 21:20:03 UTC Over Success Done 6,223.00 12.72 13.21

5 Jun 2006 21:19:32 UTC Over Success Done 6,034.00 12.34 12.54

5 Jun 2006 21:19:48 UTC Over Success Done 6,195.00 12.67 13.22

4 Jun 2006 21:17:45 UTC Over Success Done 6,110.00 12.49 13.03

Hope it helps.
Edit/Just to add some close seti_E results done with nonstandard app.

6 Jun 2006 21:41:00 UTC Over Success Done 4,762.00 12.98 12.94

6 Jun 2006 21:39:17 UTC Over Success Done 4,739.06 12.83 12.79

6 Jun 2006 21:39:17 UTC Over Success Done 4,685.00 12.83 12.79

5 Jun 2006 21:20:44 UTC Over Success Done 4,762.00 12.83 12.79

5 Jun 2006 21:19:22 UTC Over Success Done 4,863.00 12.83 12.83

---

Thanks for all these. It'll take me a while to enter them into a spreadsheet.

Keep em coming if it's not too much trouble.

Eric

---