• blah
  • blah

Sorted for ease (Ars is good)....
More good life affirming work from Roelof and another dusting of results hits the benchmark tables. Almost 200 results in all for the salivating hordes of statisticians that seem to frequent this odd little outpost of SETI quirkiness. Since I'm not one to miss out on a bit of TLC referral-age I will let Roelof quote himself! "I have updated my SETI Spy web page's CpF table and graph to reflect SETI@home 3.03 performance." Since most of this info is derived from the TLC numbers you might find it interesting to look at a more graphical and up to date visual comparison of processor efficiency. For my own part, some time ago I set about the not trivial task of creating sorted tables that link to each other. At present there are three headings that are sorted so have a look and see what you think. The links and pages are colour coded (olive for time, yellow for CPU type and orange for Cpf) to aid navigation, 'it's pink so it must be Tuesday' sort of idea. I intend to keep all the pages to a maximum of 100 entries to give some relief to those of you who don't like scrolling a page that's several meters long. Though if anyone wants a single (monster) page I will consider adding it as a seperate 'load at your peril' link! Also in the pipeline is a glossary/key to the tables as there seem to be so many acronyms that it's become an elitist forest of tech' terms. News is slow so at present just keeping the tables tweaked for your viewing pleasure seems reasonable enough. Night all.
Max out

More resultage...
Onto a second page as threatened. Another 45 or so more benchmarks added. Nothing more for now just trying to get these pages and links up and correct. Please bear with my sidebar updating which will proceed to the older pages tomorrow. Also Roelof and I have decided that submissions that are to all intents and purposes identical to ones previously submitted will not be included and the owner politely informed. The only court of appeal is me but Roelofs decision is final as far as I'm concerned. This will help to preclude situations where someone manages to shave a few seconds off a bench or changes some pointless setting to no useful effect. From the way it's going so far I can see that v3.03 is going to be very reproductive benchwise so best not to litter the results with irrelevance. Thanks.
Max out

Obtuse poetry reference department open for benchmarks...
Another 27 benchmarks added to the pile and several Xeons in amongst them but still no one challenging for the number one spot. The Xeon times were requested a while back by Roelof in a fairly involved and pleasantly technical Larry Loen thread. In typical competitive nature the discussion has moved on to sifting slower units (VLARs) from your WU lined cache, no doubt the purists would question this tactic - check it out and see what you think. Remember Xeons? These v1 & v2 agent provocateurs that have become SETI's Ozymandii, "Look on my works, ye Mighty, and despair!" Recall their stupidly fast times for clock, now outpaced by other CPUs and outdated by the far less cache intensive v3 - stone legs in the desert matey! They still produce some good times but with the client retirement a done deal many Xeons have been turned over to other distributed pastures. Will we see their like again?
I think the next results update will mutate to multiple pages as 300kb might be peanuts to many of you but there are still plenty who live, work and enjoy the web on 33.3/28.8 modem connections. Considerate or what - don't answer, just take a peek at those lovely numbers and see how you are fairing in benchworld.
Just in case: since I'm feeling generous tonight an Ozymandias link or two or three.
Max out.

After the storm, a pleasant calm...
Well obviously (from the yellow banner above) there are some more results up courtesy of Roelof. In fact there are 20 new entries so happy hunting. The new times span from 5 to 48 hours and our first Whistler beta, but don't get too excited! On the add-on scene there are some minor updates to SETI Spy (now at 3.0.4) and SETI  Driver (1.6.3.1). Also Seti Queue (3.03beta2A) has had a pretty major overhaul recently. So v3.00 has gone the way of all flesh (courtesy of a DNS change) and on time as well, considering previous mooted transition dates. Results are no longer being credited nor WU's served out to anything but v3.03 (the notable exceptions are clients in transition awaiting porting - which has kept a number of you relieved and happy)! After the hectic lunacy of the past two weeks in and around TLC and SETI it really is lovely to have no major news.
As an update to the last piece of wordage (Jan 28) it would seem that Berkeley do have a sort of grip on corrupt results and are managing to keep them out of the database. Eric Korpela in answer to a question about the hacked client and cheating replied...

"The cheating percentage is currently tiny. Out of the last 400000 results, I don't see any evidence that anyone was using the patched client. On the other hand, I see about 25 people that have overclocked to the point of generating garbage, or are having some other type of malfunction. (One of the goals for one of our new hires is to notify people that send us garbage that they may have a problem)... [snip] ...Someday (keep your fingers crossed) we really hope you will be able to look at your own results and see that everything is OK." [Jan 31 2001, Re: The patch (semi-official response) - alt.sci.seti].

I take this to mean that they are able to pick out (some?) erroneous returns and that in the future there might even be an indication of the quality (read error production of your system) in your results/stats. A lot hinges on what level of divergence between results is acceptable before it is flagged as in 'error'. I know that Roelofs adherence to repeatability is very strict concerning your submitted result.sahs. Until you check your system (with the TLC WU say) or Berkeley implements this customer feedback (very long shot) it is still possible to produce junk and be none the wiser - though it won't harm the database it will be a waste of your time.
Max out.