:-) Yes, I knew too already from dual boot testing of the still substantial gap, which the "fun with numbers" man has not (and what we knew of the Windows Beta test mean ;). Initially the average dropped to 1.54 per the history (the fast resturners) but then climbed to 1.60/1.61. Working with that, the project [now genuinely phase 1 as we know from the scientists discussing a second phase], could complete before year end, including mid year slowing, lest even more work is pulled into the library. Makes a GPU implementation for HCC1 kind of subject to rethinking (!)

--//--

The thought had crossed, but considerable resources still required would probably prohibit that route. Anyhow, the rethinking could include taking the remainder as launch into a GPU version as ''validation'' and ''proof of system readiness'', for scientist comments suggested that more added work is not improbable [the target library was already expanded twice] and for when the biggy GPU science would arrive at WCG, at some unknown time in the futuro.

--//--

That would make Linux about 7% faster at crunching HCC than XP, on your system.
It's a 3 generation old CPU though, so more modern CPU's could well be different.
Are both operating systems x86, or is it a Cox's Orange Pippin to Bramley comparison.

My stock Q6600 at 2.4Ghz on W7-64 does them somewhat quicker, but I've switched back to Linux as that adds unbeatable efficiency... multiple percents reduction in differential between Wall and CPU time. At any rate, on the volumes we are now processing the daily averages can be put against a ruler and not deviate by more than a fraction, less than 1% it seems, which is how consistent the crystals process on the daily body of work, though individual machines may see the 23% variation that Ingleside was citing.

Today we're heading for a new validated WUs record on HCC1... between 320 and 330k... project completion time now moved to October, 2011 (Don't panic , Mr. Mainwaring... HFCC's return end of June or sooner will slow it again, by a fraction ;>)

--//--

edit: Thé Link for those not knowing what thát is about ;P

Will surely add in the coming 2 weeks, but would pre-crunchers not stock up completed tasks in their caches for some extra uplift? (of course not :)

Not sure what's going on, as the total HCC1 CPU time now pretty much matches the pre conversion totals whilst HFCC is < 10 years for the day. Maybe the late Easter closely followed by May 1 has something to do with this or other "(un)natural" events on the globe to put on the breaks somewhat.

As for testing the servers, think there was a note to say that WCG was ready to handle 1 million tasks a day... the techs just need swinging the work generator for HCC1 a little harder.... can't have the pentathloners have a "no work available", now can't we :)

(Saw that "my" team will be in the games too, so duty is calling to help augmenting ''our'' numbers a bit.)

--//--