Seti credit vs Einstein Credit


log in

Advanced search

Message boards : Number crunching : Seti credit vs Einstein Credit

Previous · 1 · 2
Author Message
Josef W. SegurProject donor
Volunteer developer
Volunteer tester
Send message
Joined: 30 Oct 99
Posts: 4227
Credit: 1,042,363
RAC: 343
United States
Message 1329030 - Posted: 19 Jan 2013, 7:53:06 UTC - in response to Message 1328883.

http://boincstats.com/en/stats/-1/cpcs Shows the comparisons.

The comparison chart at http://boinc.netsoft-online.com/e107_plugins/boinc/get_cpcs.php gives more detail and is better documented. It also says that SETI gives ten times more credit than Einstein, which I'm certain is false.

I think the comparison figures for both projects are distorted by the use of GPUs at both projects: the comparison charts are based on CPU time only. Overall, I think the administrators at both projects strive conscientiously to honour the original credit definition as closely as possible: anyone wishing to choose between them should run their own tests on their own hardware. Or, preferably, choose on the basis of the value of the science on their own personal scale.

It should also be noted that the code to update the credit_per_cpu_sec field in the host records was deleted from BOINC about 2.5 years ago (changeset f65072a). At any project which is using a newer version of the BOINC server code, those values are whatever was previously there. Old hosts have a value, new hosts all have zeroes. In short, unless you know a project is running an ancient version of BOINC, the figures are meaningless.
Joe

Profile Raistmer
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 16 Jun 01
Posts: 3395
Credit: 46,337,294
RAC: 9,952
Russia
Message 1329033 - Posted: 19 Jan 2013, 8:42:33 UTC - in response to Message 1328993.

What I find interesting is in looking at the two charts, the credit difference between Seti and SetiBeta. Same work units. Makes me think that the measuring stick is broken.

You right, it's broken and long ago. And looks like never will be repaired. That's the single and right answer on original question. Because credits are not calibrated. Period.
____________

Profile James SotherdenProject donor
Avatar
Send message
Joined: 16 May 99
Posts: 8679
Credit: 33,044,069
RAC: 56,521
United States
Message 1329035 - Posted: 19 Jan 2013, 8:58:00 UTC

It should not matter who gives the most credit. It should be what project do you really want to crunch. For me Its Seti@Home. The other two I do like, but they are there for back up when I cant crunch S@H.
____________

Old James

ClaggyProject donor
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4067
Credit: 32,875,694
RAC: 6,992
United Kingdom
Message 1329036 - Posted: 19 Jan 2013, 9:13:26 UTC - in response to Message 1328993.
Last modified: 19 Jan 2013, 9:20:48 UTC

What I find interesting is in looking at the two charts, the credit difference between Seti and SetiBeta. Same work units. Makes me think that the measuring stick is broken.

Seti and Seti Beta don't have the same workunits (any more), Seti is using v6 Wu's still, while Seti Beta is using v7 Wu's,
the data going into the Wu's is effectively the same, But the Beta Wu's are effectively larger now because of the added autocorrection signal searches,
you can tell the difference whether or not the header has autocorrection values:

<workunit_header>
<name>08ja11ab.19544.13736.140733193388041.14.162</name>
Snip
<analysis_cfg>
<spike_thresh>24</spike_thresh>
<spikes_per_spectrum>1</spikes_per_spectrum>
<autocorr_thresh>17.7999992</autocorr_thresh>
<autocorr_per_spectrum>1</autocorr_per_spectrum>
<autocorr_fftlen>131072</autocorr_fftlen>
<gauss_null_chi_sq_thresh>2.26359248</gauss_null_chi_sq_thresh>
<gauss_chi_sq_thresh>1.41999996</gauss_chi_sq_thresh>
<gauss_power_thresh>3</gauss_power_thresh>
<gauss_peak_power_thresh>3.20000005</gauss_peak_power_thresh>
<gauss_pot_length>64</gauss_pot_length>
<pulse_thresh>20.2139301</pulse_thresh>
<pulse_display_thresh>0.5</pulse_display_thresh>
<pulse_max>40960</pulse_max>
<pulse_min>16</pulse_min>
<pulse_fft_max>8192</pulse_fft_max>
<pulse_pot_length>256</pulse_pot_length>
<triplet_thresh>9.21395588</triplet_thresh>
<triplet_max>131072</triplet_max>
<triplet_min>16</triplet_min>
<triplet_pot_length>256</triplet_pot_length>
<pot_overlap_factor>0.5</pot_overlap_factor>
<pot_t_offset>1</pot_t_offset>
<pot_min_slew>0.00209999993</pot_min_slew>
<pot_max_slew>0.0104999999</pot_max_slew>
<chirp_resolution>0.1665</chirp_resolution>
<analysis_fft_lengths>262136</analysis_fft_lengths>
<bsmooth_boxcar_length>8192</bsmooth_boxcar_length>
<bsmooth_chunk_size>32768</bsmooth_chunk_size>
<chirps>
<chirp_parameter_t>
<chirp_limit>30</chirp_limit>
<fft_len_flags>262136</fft_len_flags>
</chirp_parameter_t>
<chirp_parameter_t>
<chirp_limit>100</chirp_limit>
<fft_len_flags>65528</fft_len_flags>
</chirp_parameter_t>
</chirps>
<pulse_beams>1</pulse_beams>
<max_signals>30</max_signals>
<max_spikes>8</max_spikes>
<max_autocorr>8</max_autocorr>
<max_gaussians>0</max_gaussians>
<max_pulses>0</max_pulses>
<max_triplets>0</max_triplets>
<keyuniq>-1325094</keyuniq>
<credit_rate>2.8499999</credit_rate>
</analysis_cfg>
<sb_id>0</sb_id>
<iq_modified>0</iq_modified>
</group_info>
<subband_desc>
<number>162</number>
<center>1419086303.7109</center>
<base>1419082031.25</base>
<sample_rate>9765.625</sample_rate>
</subband_desc>
<sb_id>0</sb_id>
</workunit_header>


Claggy

Profile tullioProject donor
Send message
Joined: 9 Apr 04
Posts: 3660
Credit: 369,027
RAC: 222
Italy
Message 1329054 - Posted: 19 Jan 2013, 11:01:29 UTC

I am running SETI@home, Einstein@home and Test4Theory@home on one CPU. They never run in high priority. On the other CPU I run Albert@home (an Einstein Beta), LHC@home, Test4Theory@home and SETI@home Astropulse with a Lunatics app. Also CPDN@home with extended deadlines. They never run in high priority since I have a very small cache (0.25 day). All this on Linux. And yes, the second CPU also hosts a Solaris Virtual Machine with BOINC and SETI@home app by Dotsch.
Tullio
____________

Profile Bernie Vine
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 26 May 99
Posts: 6897
Credit: 25,657,318
RAC: 38,885
United Kingdom
Message 1329058 - Posted: 19 Jan 2013, 12:09:38 UTC - in response to Message 1328873.

Compared to S@H, Einstein has very short deadlines for the WUs, so your computer "is forced" to run Einstein with high priority.

The reason I stopped Einstein, I am in in for the science not being forced to crunch one project over another.

Einstein forces no such thing: I don't think I've ever seen an Einstein task run in high priority.

What forces high priority is a user choice of cache settings which don't play 'nice' with project deadlines - we used to see that a lot when SETI had 7-day deadlines for shorties (before the task durations were doubled by increasing the sensitivity of the search). The rule of thumb when running more than one project is that your chosen cache size should be no more than [shortest deadline of any task on any of the projects] / [number of separate projects running].

So, SETI + Einstein: lowest deadline is 14 days at either project (actually, slightly shorter at SETI) - don't cache more than 7 days, and make sure you choose the right value pair (0,7 for BOINC v6, 7,0 for BOINC v7). That will calm things down a lot, though I don't personally ever run more than a 2 day cache.


I must have had my cache set wrong then because I had never seen a task run at high priority till I joined Einstein, if I remember all the GPU tasks had very short deadlines and forced SETI to take a back seat till they had all cleared, I haven't touched my cache setting since I set them for Boinc 7.

____________


Today is life, the only life we're sure of. Make the most of today.

Richard HaselgroveProject donor
Volunteer tester
Send message
Joined: 4 Jul 99
Posts: 8460
Credit: 48,804,539
RAC: 82,044
United Kingdom
Message 1329083 - Posted: 19 Jan 2013, 13:23:05 UTC - in response to Message 1329058.

Compared to S@H, Einstein has very short deadlines for the WUs, so your computer "is forced" to run Einstein with high priority.

The reason I stopped Einstein, I am in in for the science not being forced to crunch one project over another.

Einstein forces no such thing: I don't think I've ever seen an Einstein task run in high priority.

What forces high priority is a user choice of cache settings which don't play 'nice' with project deadlines - we used to see that a lot when SETI had 7-day deadlines for shorties (before the task durations were doubled by increasing the sensitivity of the search). The rule of thumb when running more than one project is that your chosen cache size should be no more than [shortest deadline of any task on any of the projects] / [number of separate projects running].

So, SETI + Einstein: lowest deadline is 14 days at either project (actually, slightly shorter at SETI) - don't cache more than 7 days, and make sure you choose the right value pair (0,7 for BOINC v6, 7,0 for BOINC v7). That will calm things down a lot, though I don't personally ever run more than a 2 day cache.

I must have had my cache set wrong then because I had never seen a task run at high priority till I joined Einstein, if I remember all the GPU tasks had very short deadlines and forced SETI to take a back seat till they had all cleared, I haven't touched my cache setting since I set them for Boinc 7.

The Einstein project itself has a 14 day deadline for all tasks.

Their Beta project 'Albert' does set shorter deadlines of ~3 days for tasks/applications which are under active test, especially when they are approaching transfer to their main servers. That's done to ensure a rapid turnround, so that the staff can check the returned results quickly without delaying deployment - I think it's a legitimate tactic for a Beta project. [They are better resourced than SETI, staff-wise, so new application development and deployment happens at a faster pace there.]

Profile soft^spirit
Avatar
Send message
Joined: 18 May 99
Posts: 6374
Credit: 28,631,059
RAC: 94
United States
Message 1329163 - Posted: 19 Jan 2013, 18:44:24 UTC

I would say Einstein offers SLIGHTLY more credits than SETI. Not massively more.

If you just want big numbers to impress your friends with (and if so seriously reconsider your social circle) then try collatz. Honestly I feel I shed 50 IQ points while I crunched it. But my RAC was HUGE.
____________

Janice

Profile betregerProject donor
Avatar
Send message
Joined: 29 Jun 99
Posts: 2230
Credit: 4,706,594
RAC: 10,165
United States
Message 1329174 - Posted: 19 Jan 2013, 19:25:57 UTC - in response to Message 1329163.

+1
____________

juan BFBProject donor
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 5220
Credit: 283,848,129
RAC: 451,101
Brazil
Message 1329237 - Posted: 19 Jan 2013, 22:47:49 UTC - in response to Message 1329163.

I would say Einstein offers SLIGHTLY more credits than SETI. Not massively more.

If you just want big numbers to impress your friends with (and if so seriously reconsider your social circle) then try collatz. Honestly I feel I shed 50 IQ points while I crunched it. But my RAC was HUGE.

Ou go to GPUGRID, they "pay" 115k for a single WU! Almos 10x the SETI amount!

____________

Profile Fred J. Verster
Volunteer tester
Send message
Joined: 21 Apr 04
Posts: 3236
Credit: 31,683,525
RAC: 6,779
Netherlands
Message 1329252 - Posted: 20 Jan 2013, 0:01:07 UTC - in response to Message 1329237.
Last modified: 20 Jan 2013, 0:02:34 UTC

I would say Einstein offers SLIGHTLY more credits than SETI. Not massively more.

If you just want big numbers to impress your friends with (and if so seriously reconsider your social circle) then try collatz. Honestly I feel I shed 50 IQ points while I crunched it. But my RAC was HUGE.

Ou go to GPUGRID, they "pay" 115k for a single WU! Almos 10x the SETI amount!


But at GPUGrid, it heavily depends on the GPU you use, a FERMI or KEPPLER gives
far more credit, compaired to the 8000/900 & 200 NVidia series.

And for Collatz C. similar, Single Precision AMD/ATI GPUs do very well, especially 5770; 5870 (D.P.).
____________

juan BFBProject donor
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 5220
Credit: 283,848,129
RAC: 451,101
Brazil
Message 1329282 - Posted: 20 Jan 2013, 2:00:09 UTC - in response to Message 1329252.

I would say Einstein offers SLIGHTLY more credits than SETI. Not massively more.

If you just want big numbers to impress your friends with (and if so seriously reconsider your social circle) then try collatz. Honestly I feel I shed 50 IQ points while I crunched it. But my RAC was HUGE.

Ou go to GPUGRID, they "pay" 115k for a single WU! Almos 10x the SETI amount!


But at GPUGrid, it heavily depends on the GPU you use, a FERMI or KEPPLER gives
far more credit, compaired to the 8000/900 & 200 NVidia series.

And for Collatz C. similar, Single Precision AMD/ATI GPUs do very well, especially 5770; 5870 (D.P.).

Yes but "if" the question is "just" about credit... not my case of course. I just want to show the big diference on the credit systems. Credit is just a way to compare the performance of diferent hosts beetween the same project, compare one project with the other ussing the credit is a compleate waste of time.
____________

Profile Raistmer
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 16 Jun 01
Posts: 3395
Credit: 46,337,294
RAC: 9,952
Russia
Message 1329632 - Posted: 20 Jan 2013, 23:14:23 UTC - in response to Message 1329282.

Credit is just a way to compare the performance of diferent hosts beetween the same project, compare one project with the other ussing the credit is a compleate waste of time.

Correct.

____________

Lionel
Send message
Joined: 25 Mar 00
Posts: 544
Credit: 223,729,094
RAC: 215,178
Australia
Message 1330096 - Posted: 22 Jan 2013, 8:55:04 UTC - in response to Message 1328770.

Why is the credit for work done for Seti one tenth of that done for Einstein? Credit granted for Seti work is very small.


Credit has gone down over time as a defacto means of encouraging people to leave Seti and donate resources elsewhere. This reduces the pressure on Seti's resources and thereby pushes out any urgent need for capital/infrastructure enhancement/replacement. Given the forthcoming increase in computational power of GPUs and the general migration of the base upwards (in terms of total processing power) I suspect that credit will once again be reduced in the next 12 months.


____________

Previous · 1 · 2

Message boards : Number crunching : Seti credit vs Einstein Credit

Copyright © 2014 University of California