Seti@Home optimized science apps and information
Optimized Seti@Home apps => Discussion Forum => Topic started by: Gecko_R7 on 26 Aug 2007, 02:34:38 pm
-
Question for Joe or Crunch3r:
Been seeing many WUs granting 63.98 credits these past several weeks.
Just compared times on my T7200 -doze box w/ Crunch3r's 2.4 MB ssse3 vs. OSX box T7200 w/ Alex's revised 7.2 for C2D.
The iMac crunched several WUs w/ AR=.010024 in 7751.81 for 63.98 credits
The Dell, WUs w/ AR=.006913 in 8319.08 for the same 63.98
Both rigs have crunched dozens of 63.98 credit WUs, yet the AR of the same credit WUs is distinctly different for each platform.
Different AR = same WU credit?
I know there is a simple explanation for this.
A refresher lesson from the resident professors? :)
-
Question for Joe or Crunch3r:
Been seeing many WUs granting 63.98 credits these past several weeks.
Just compared times on my T7200 -doze box w/ Crunch3r's 2.4 MB ssse3 vs. OSX box T7200 w/ Alex's revised 7.2 for C2D.
The iMac crunched several WUs w/ AR=.010024 in 7751.81 for 63.98 credits
The Dell, WUs w/ AR=.006913 in 8319.08 for the same 63.98
Both rigs have crunched dozens of 63.98 credit WUs, yet the AR of the same credit WUs is distinctly different for each platform.
Different AR = same WU credit?
I know there is a simple explanation for this.
A refresher lesson from the resident professors? :)
You should have a look at the flop counters an compare them too cuz that's how credits are calculated.
-
Question for Joe or Crunch3r:
Been seeing many WUs granting 63.98 credits these past several weeks.
Just compared times on my T7200 -doze box w/ Crunch3r's 2.4 MB ssse3 vs. OSX box T7200 w/ Alex's revised 7.2 for C2D.
The iMac crunched several WUs w/ AR=.010024 in 7751.81 for 63.98 credits
The Dell, WUs w/ AR=.006913 in 8319.08 for the same 63.98
Both rigs have crunched dozens of 63.98 credit WUs, yet the AR of the same credit WUs is distinctly different for each platform.
Different AR = same WU credit?
I know there is a simple explanation for this.
A refresher lesson from the resident professors? :)
Any MB AR from 0.000 to 0.050 gets exactly the same processing and yields 63.98 credits. These are the VLAR units where there's so little motion that all the data remains within the beam width for the full 107.34 second duration of the WU. The only other place the flopcount curve has the right value for a 63.98 claim is in the middle of the cliff at AR 0.400, so I think it's impossible.
For Line feed, the dread 58.7 credit units were the same thing except extended to the 0.083 beam width of that feed.
Joe
-
New tasks are being split at Beta (contrary to what the Server Status page still says), and the _0 copy of both the ones I've downloaded so far had been snapped up in advance by representatives of the 'dark fleet' of anonymous P4s - so there should be plenty of comparison material, if a little slowly.
I wonder how they're going to handle the transition to Main? Nuking the entire task list, and starting from a clean slate, is fine at Beta (though I would have been a bit p****d if it had coincided with with my timed pre-release test runs of the opti apps): but I can't see it going down a storm at Main. Cleanest would be to shut down all splitting for at least one deadline cycle, and suppress re-issues: but we'll have to see whether they have the patience for that.
Let us know when you've got any Mk2s for us to test.