On an aside, if RAC is meaningless can someone tell me why it is included in the "author's stats" in this forum? Just a curious question here.

Because as these threads have demonstrated, peope like to see numbers representing things. They don't have to be accurate or actually representative, but as long as they're there, people like it.
A more usefull number would be average Work Units per day or hour, but based on a weekly (or better yet) monthly sample. But the problem with that is any fluctuations would be significantly damped (and people seem to get all excited when the see a huge spike, and all worked up when there's a huge drop) & major changes would take a while to show through (although that is the case even now when it comes to accurate information).

---

Grant
Darwin NT

Agree 100%! If they are not going to fix this new abortion called a credit system, at least reset the stats!

It has been fixed. It's just that those that have had the advantage of optimised clients don't like the fix because their advantage has been reduced (at least for now).

---

Grant
Darwin NT

Also, I earlier thought that since the credits now are based on mflops it wouldn't matter if you got a low or high AR WU as you got credit for the actual work from the CPU, how ever this has turned out not be the case for some reason...

The impact of low and high AR is even greater with enhanced. You could determine the default credit you get if you pick a wu that has an ar of 0.422 because those are most of the wus.

---

Join BOINC United now!

And as has been said in a previous post, I could accept that if there was a reset and we all started over at zero. That would be a truly level playing field. Anything short of that is unfair to those of us who spent a lot of time and effort building up what we had.

Agree 100%! If they are not going to fix this new abortion called a credit system, at least reset the stats!

Try this on for size. Total up all the members here who used optimzed apps for version 4.xx. Guess what......you are in the minority here. No matter how much good you did for the project, crunching massive amounts of data, purchasing more and more equipment, you are still in the minority of users here and cannot dictate to the majority.

And remember I am part of the optimized 4.xx croud also!

---

Boinc....Boinc....Boinc....Boinc....

Also, I earlier thought that since the credits now are based on mflops it wouldn't matter if you got a low or high AR WU as you got credit for the actual work from the CPU, how ever this has turned out not be the case for some reason...

?
Looking at some of the results for one of you machines shows this.
Time GC
10,349.08 31.44
66.97 0.10
16,404.33 65.22
9,260.03 33.01

More work done, more credit. Less work done, less credit.

---

Grant
Darwin NT

Also, I earlier thought that since the credits now are based on mflops it wouldn't matter if you got a low or high AR WU as you got credit for the actual work from the CPU, how ever this has turned out not be the case for some reason...

The actual number of floating point operations changes as the angle range changes, that's why some work units go faster than "standard" work, and others take very much longer.

---

Attn: Grant, Winterknight, Ned, and others with similar posts

No offense, but you guys are sounding like a broken record now - spewing the same retorts over and over, and it's not working. So please, just stop.

At this point, many of us want answers to the issues you are not even willing to address.

Many of us have been crunching SETI for years, and we are very close to pulling the plug right now. What we want and need right now is to hear from a decision maker, like ERIC!

You are not going to convince us that this new credit system is not flawed!

---

http://www.setiusa.net

OTW


QS

OK, here comes my last compliant w.r.t deleted threads:

I don't think this thread should have been deleted.

It sure was controversial, but there were no flames or abusive statements involved.

Regards Hans

P.S:

I'll keep my big mouth shut now.

---

My daily average for each week has always been within 10-15% of my RAC so let's look at it. Oh my goodness, my daily average for the week is dropping at about the same rate as my RAC.

You must be looking at different graphs to what i am.

Okay, let's look at my daily production numbers. Hmmm, daily production has dropped much faster than my RAC or my daily average. Either way you look at it my claim is the same.

*deep sigh*
Daily production isn't a good indicator as it is very dependant on server outages & will be affected considerably by the variable processing times.
Weekly or monthly will give a better idea of what the average daily or hourly crunching is like.
And once again, as mentioned by Eric, those that have been using Optimised applications will see a greater reduction in credit per hour than thouse using the standard application, as the standard application is what his efforts were based on.

Okay, now I am really confused. When I was quoting my RAC I was being told it was meaningless and that I should be looking at daily or weekly numbers. Now all of a sudden daily numbers are also no good. Well I checked my weekly average as well and it has already dumped by 10% in just 3 days. As my higher production days (which do not mean as much)go off the chart then the weekly numbers will drop dramatically as well. So now which numbers should I look at? No matter which I look at the drop is happening. It may take longer to show on the weekly or monthly numbers but it will happen just the same.

---

I am liking the earlier suggesion of starting over again (since the method changed drastically) more and more.

The drastic change has only been for a small percentage of people. For most people there has been little, if any, change.

A small percentage of people who produce a very large portion of the science.

---

And as has been said in a previous post, I could accept that if there was a reset and we all started over at zero. That would be a truly level playing field. Anything short of that is unfair to those of us who spent a lot of time and effort building up what we had.

Agree 100%! If they are not going to fix this new abortion called a credit system, at least reset the stats!

Try this on for size. Total up all the members here who used optimzed apps for version 4.xx. Guess what......you are in the minority here. No matter how much good you did for the project, crunching massive amounts of data, purchasing more and more equipment, you are still in the minority of users here and cannot dictate to the majority.

And remember I am part of the optimized 4.xx croud also!

We are not trying to dictate. We are asking for a fair shake. If they want a truly level playing field with the new system, then reset the stats and let us all start over under the new system.

---

If they want a truly level playing field with the new system, then reset the stats and let us all start over under the new system.

Just drop me a line in time so I can write down my score...

Regards Hans

---

First I thought the credits was fair, but after a little more closely look
I think the credits look "hinky"

For example, isn't is strange that all WUs for one particular PC always almost get the same CS for a WU as someone else who claim about the same amount of credit.

Like my main server almost always both claims and gets a credit around 65 CS per WU. (and this is exactly around 65 for ever WU)
Like this one, it's a very typical WU for that machine.
http://setiathome.berkeley.edu/workunit.php?wuid=78964999

But my other computer the AMD XP almost always gets WUs that have a completely different range of 61 CS per WU (and this is exactly around 61 for ever WU). This is a very typical WU for that computer.
http://setiathome.berkeley.edu/workunit.php?wuid=78500417


And my other computers show the same "effect", I find it a bit weird if nothing else.

That's probably an artifact of the feeder or the scheduler. For instance, your host 1947001 got 9 WUs on 20 May 2006 at 19:02:46 UTC. From the deadlines I think there were only two different angle ranges, so you'd get several WUs in a row with nearly identical times and claims. The same thing happens with each host, the work isn't evenly distributed.

I hope the reduced load on the servers from setiathome_enhanced will allow the project to better randomize the work distribution.

                                             Joe

Quote: Steve Akers:

"No offense, but you guys are sounding like a broken record now - spewing the same retorts over and over, and it's not working. So please, just stop... At this point, many of us want answers to the issues you are not even willing to address."

Although I myself am behind the new Enhanced credit system 100%, I'd also like to think that I can see both sides to the issue and I agree that there has been a lot of gobbledy-gook spewed out in these threads. I hope that you guys do get the answers you are seeking from the folks who are actually running the show. It is certainly not an unfair request.

My own analysis shows that I am claiming exactly the same credits here on SETI Enhanced as I am claiming over at Einstein, so in that regard I feel that the new system is equitable. I also believe that it is the people who are running optimized crunchers who have historically gotten the short end of the stick because of low credit claims from the benchmarking method. I feel that Enhanced has addressed that issue in making sure that everyone gets the same reward, regardless of whether they use the standard science application, or an optimized application which gets the work done faster. If all projects moved to the FLOP-counting method at this time there would be much less disparity between them.

If I understand correctly what I have read in these threads, it appears that most people do feel that the new method of rewarding our work is fair and equitable, but it is only the amount of credit being awarded per FLOP that is in dispute. I believe Eric stated in his post that he is continuing to look into this.

Regardless, I hope that for all concerned the matter gets resolved to everyone's satisfaction and the fun of crunching for SETI@home may continue as in the past.

Just my opinion, of course...

Dig

You are not going to convince us that this new credit system is not flawed!

How much pending credit do you have?

---

I wasn't in most of yesterday and then forgot to send a reply.

I assume you will withdraw your accusation.

regards
Andy

Hi Andy:

There was no insult intended. I found it amusing that in that specific example you claimed 5.6 time more granted credit (CS/hr) than the poor fellow who took 79,523 seconds to complete his work unit. The Table below shows the 4 sets of results you referred to in your example:

79,523.20 sec., CC = 194.71, GC = 56.46 CC = 8.81 CS/hr, GC = 2.56 CS/hr
46,570.62 sec., CC = 49.37, GC = 56.46 CC = 3.82 CS/hr, GC = 4.36 CS/hr
14,287.27 sec., CC = 56.46, GC = 56.46 CC = 14.23 CS/hr, GC = 14.23 CS/hr
22,595.64 sec., CC = 56.46, GC = 56.46 CC = 9.00 CS/hr, GC = 9.00 CS/hr

The results that I posted with the different combinations of TruXsoft’s BOINC client and various versions of the Seti app were for everyone’s information so that they may see the differences. The data was copied from the Seti results page into MS Excel and results presented as is below. Why certain results were obtained is irrelevant and saves on a lot of useless jaw flapping.

Genuine Intel (R) Pentium(R) 4 650 CPU 3.40GHz running Microsoft Windows XP Professional Edition, Service Pack 2, (05.01.2600.00)

5.3.12.tx36 BOINC client and standard Seti V4.18 (average of 23 work units)
CPU time = 2,072.61 sec., Claimed credit = 16.81 CS/hr., Granted credit = 23.30 CS/hr.

5.3.12.tx36 BOINC client and Crunch3r’s Seti SSE3 V4.11 (average of 97 work units)
CPU time = 3,110.58 sec., Claimed credit = 34.90 CS/hr., Granted credit = 29.81 CS/hr.

5.3.12.tx36 BOINC client and Crunch3r’s Seti SSE3 V5.11 (average of 34 work units)
CPU time = 12,654.50 sec., Claimed credit = 40.92 CS/hr., Granted credit = 15.71 CS/hr.

5.3.12.tx36 BOINC client and Crunch3r’s Seti SSE3 V5.12 (average of 31 work units)
CPU time = 13,751.11 sec., Claimed credit = 14.38 CS/hr., Granted credit = 14.13 CS/hr.

Authentic AMD AMD Athlon(tm) 64 Processor 3000+ running Linux 2.6.16-1.2108_FC4

5.3.12.tx36 BOINC client and Crunch3r’s Seti SSE3 V5.12 (average of 11 work units)
CPU time = 9,033.28 sec., Claimed credit = 13.56 CS/hr., Granted credit = 13.93 CS/hr.

Authentic AMD Athlon (tm) 64 Processor 3000+ running Microsoft Windows 2000 Professional Edition, Service Pack 4, (05.00.2195.00)

5.3.12.tx36 BOINC client and Crunch3r’s Seti SSE3 V5.11 (average of 26 work units)
CPU time = 11,120.90 sec., Claimed credit = 53.18 CS/hr., Granted credit = 20.08 CS/hr.


Today, I signed up with Einstein@home and I am running it on a 50:50 basis with Seti. Upon signing up, Einstein’s version 4.37 for the improved all-sky pulsar search was loaded onto my computer. They also have an optimized SSE3 version U41.04 made available by Akosf which I promptly installed.

From exploring the site, their optimized version is returning the following results:

Authentic AMD Athlon (tm) 64 Processor 3000+ running Microsoft Windows XP Professional Edition, Service Pack 2, (05.01.2600.00)
5.3.12.tx36 BOINC client and Akosf’s SSE S41.07 optimized client (average of 9 work units)
CPU time = 3,087.11 sec., Claimed credit = 44.49 CS/hr., Granted credit = 42.78 CS/hr.

Authentic AMD Dual Core AMD Opteron (tm) Processor 165 running Microsoft Windows XP Professional Edition, Service Pack 2, (05.01.2600.00)
5.3.12.tx36 BOINC client and Akosf’s SSE3 U41.04 optimized client (average of 40 work units)
CPU time = 2,395.42 sec., Claimed credit = 56.16 CS/hr., Granted credit = 54.83 CS/hr.

From the discussions at Einstein the preferred BOINC client is TruXoft’s 5.3.12.tx36 client.

The conclusion to all of this is that if the optimized Seti Clan wants to keep up to the optimized Einstein Clan then you had better lock and load TruXoft’s 5.3.12.tx36 BOINC client and Crunch3r’s 5.11 Seti app. otherwise you will be left eating their dust.

We should all be very grateful to Crunch3r for having done such an excellent job of developing an optimized Seti client that so closely matches Einstein’s optimized client.

I had to download my version of the Einstein optimized SSE3 version U41.04 at the following mirror site. The 1 from the Einstein site was corrupt. http://einstein.terrorfront.info/en/index.htm

Regards
Franz

Daily production isn't a good indicator as it is very dependant on server outages & will be affected considerably by the variable processing times.
Weekly or monthly will give a better idea of what the average daily or hourly crunching is like.

Okay, now I am really confused. When I was quoting my RAC I was being told it was meaningless and that I should be looking at daily or weekly numbers. Now all of a sudden daily numbers are also no good.

Didn't make myself as clear there as i would have liked.
If you look at your daily numbers on those graphs, some days you have huge amounts, others small amounts of credit. Dependant on server outages & so on- even if there is no change in your output.
A daily number of credits, based on a weeks (or a months) crunching would give a more accurate idea of what the daily production really is.

Well I checked my weekly average as well and it has already dumped by 10% in just 3 days. As my higher production days (which do not mean as much)go off the chart then the weekly numbers will drop dramatically as well. So now which numbers should I look at? No matter which I look at the drop is happening. It may take longer to show on the weekly or monthly numbers but it will happen just the same.

To quote myself, reusing the section you also quoted. "And once again, as mentioned by Eric, those that have been using Optimised applications will see a greater reduction in credit per hour than thouse using the standard application, as the standard application is what his efforts were based on."


BTW was April the month you switched to optimised clients, or did you bring more machines on line? There was a huge jump in credit for that month compared to the previous ones. This month so far has given more credits in the month than those early ones, but certainly not as much as in April.

---

Grant
Darwin NT

And as has been said in a previous post, I could accept that if there was a reset and we all started over at zero. That would be a truly level playing field. Anything short of that is unfair to those of us who spent a lot of time and effort building up what we had.

Agree 100%! If they are not going to fix this new abortion called a credit system, at least reset the stats!

Try this on for size. Total up all the members here who used optimzed apps for version 4.xx. Guess what......you are in the minority here. No matter how much good you did for the project, crunching massive amounts of data, purchasing more and more equipment, you are still in the minority of users here and cannot dictate to the majority.

And remember I am part of the optimized 4.xx croud also!

We are not trying to dictate. We are asking for a fair shake. If they want a truly level playing field with the new system, then reset the stats and let us all start over under the new system.

or just simply differentiate the scores between SETI and SETI Enhanced :)

---

You are not going to convince us that this new credit system is not flawed!

How much pending credit do you have?

251 credits at the moment - not much.

---

http://www.setiusa.net