To my fellow crunchers, please accept my appoligies, I was not "flaming" you. My frustrations come from WCG, they increased the run time on all cpu projects, big notice, to slow down the number of cpu wu processed in order to allow for more GPU wu.
While I am all for completing HCC asap, I also want FAAH, GFAM, SN2S, HPF2, CEP2, and anything new done asap. The current system is slowing everything for the benifit of HCC/GPU, this to me is wrong.
Before any of you states the obvious, I do know that cpu wu run time has come down, only becuase I pay attention, WCG does not care enough about the cpu cruncher to post this. I also suspect that once the 2x HCC wu hit the streets that they will go back up and GPU quotas will be increased.
The WCG servers as they are now appear to be limited by the wu/day of approx 2 mil. As more gpu wu are done the cpu wu WILL be restricted. This is NOT RIGHT. There is more than HCC, they deserve equal time. If HCC is all WCG can do, so be it, but at least let the other projects know they are no longer supported. EDIT: Or should I say only minimumly supported?

As was noted several weeks ago, the Techs already have cut a number of sciences back to their shorter lengths [Specifically GFAM, which got the greatest extension and FAAH, latter in fact cut back to the May lengths], as they found their work in months prior to HCC-GPU launch to make the system more robust *did* pay off! For sure, the speed up of HCC has *not* impacted the other sciences [total effective work processed], the techs simply put more dockings/jobs/positions in a single task for the sciences that allowed to cut them differently. E.g. if previously 100 dockings were put in 3 tasks, they temporarily put 100 dockings in 2 tasks... total processing time same, for the same amount of work performed as before.

Given that the work is completed to put 2 images in the HCC tasks, double their time, it is giving room to return other sciences to their shorter run times [with great confidence] and prepare for more additions and variations of how and on what WCG sciences are processed. It's taken time going through the motions, and now near completion, think we can move on.

Thank you.

P.S. Just noted in the Performance Chart of HPF2, that their runtimes now also have returned to pre-extention. Dare not say anything about DSFL because of the great variability of the processing time needed for different targets of that science, but these too are now running on average shorter than they have been in the last 6 months. HFCC and CEP2 were never resized. Leaves SN2S. Not seeing anything acute in run time change.

More of what you qualify as blah: Whence WCG is able to produce variable sized tasks, fit to accommodate all, the HCMD2 approach, but pre-sized, rather than cut-off at, many more will be able to eclectically participate on almost any WCG science. How that will work out... we'll see. If a task requires 6 hours, but the device is not on for more than half an hour a day [even if it's a Mercedes amongst computer], than doubt much can be done about that. It would get into options like at Rosetta: Give me 1, 2, 3, 6 hour running tasks. From that point on I propose WCG does away with with any result statistics... a completely meaningless measure, which it to good extent already is. If only credits [points] were a fair representation of work calculations performed [which they are not IMHO, anywhere], runtime and results truly become a redundant measure.

Why anyone would complain about work getting completed faster, is beyond words that can be expressed. Just because one's computer cannot complete wu's as fast as the next person has no bearing on WCG as a whole. It could be the slowest computer that computes a wu that holds the key to a cure. No one knows. All I know it the more wu's get completed, the closer we are at finding a cure. No one particular computer is more important than the other. We all share a goal. That's finding a cure. Period!

To emphasize, *Nothing* was "slowed down", the tasks were packed with more work per unit, the sum of the work units still producing the same amount of end product. Extending HCC [2 images in 1 WU] is a natural consequence to reduce the load on the scheduler and make room for more things to come in the future [not wait on the future, see things tilt, and only then act].

That said, lunch time hints at 1.9 million total results for the day and things from the outside appearance, still holding up. Whether knreed behind the scenes is not already on double dose of tranquilizer v.v. the pace things are running... don't know. Making the HCC double sized will massively reduce scheduler/distributer/transitioner/feeder loads to ensure states such as keeping the cache queues filled... prime directive.