Lawrence,

Thank you for the link in your reply. After reading the information in the link, it appears I failed to be as clear and precise in my prior post as I really needed to be.

Please permit me to try again with my position.

The observation was not for any particular project as I believe the post you referenced presented.

In my post, I had intended to supply sufficient information where the methodology, not necessarily a technique, could be designed and possibly implemented such that all WU's could fall into separate queues specifically to be dispatched to machines which return WU results within a 24 hour period.

This method could easily be used for high priority dispatching of WU's within a specific queue of WU's for a particular project. There appears to already be a mechanism in place for the express purpose of having the necessary WU's and the requisite validation WU's to be sent out for immediate processing forming quick turnaround. My suggestion, while similar is, I believe, just a tad different. As I see it, some set of WU's and their associated validation WU's would all be placed into this queue for nominal WU completion and return for validation to be completed within a 24 hr period.

There are some benefits for this type of implementation. Many folks, will stop carrying long queues of WU's that need to be completed. In other words, I suspect many will reduce their number of "Additional Days" of WU depth.

As I indicated before, I can only suspect this has an opportunity to change the complexion of how the entire WGC behaves in complete through put of WU's.

Here's my oldest WU in pending validation:



Here's my most recent WU in pending validation:



Now; if there were a queue for machines that returned WU's typically within 24 hrs were available and all the WU's listed in the WU status as shown in the above; I suspect the throughput of the WU's in general would;

• show an overall improvement in WCG throughput (being defined as all WU's in a batch being verified quickly)
  
• provide results back for analysis faster (presuming some things)
  
• almost eliminate the "pending validation" for machines that respond within 24 hrs
  

Personally, I have my system set up so that it only starts to look for additional work just about the time a WU is about to complete, something like about 7 minutes of WU completion.

Does that present a better explanation about what I would like to see?

Thanks
---Barney

If the system can reward the quick turnaround users with Betas. . . why couldn't the same system muster those same users together to get work turning faster?

I believe the crunching experience is enhanced with work turning in two days instead of two weeks.

Or take a trick from GPU grid. They give a 25% point bonus if the WU is returned within 24 hours... might not make everyone lower their buffer, but my buffer would definitely drop to less than 24 hours.

and I have a feeling it would make enough people change to make a noticeable impact...

Otis11

And that in and of itself is the real crux of the matter.

As my images above easily demonstrated; its not about "being patient" it certainly revolves around the demonstrable benefit in WU throughput.

It's not about the badges, it is about getting WU's:

• Processed quickly
  
• Returned quickly
  
• Validated quickly
  

So those with the big brains that can use the information for the purpose of mankind. (that as it is these days).

Seemingly it does not benefit anyone for WU's that are completed within hours of initial dispatching sitting and awaiting validation by others for 2+ weeks. Awaiting validation within a 24 hour period seems to be reasonable. WCG must know which machines return WU's within the shortest time periods. So creating a specialized queue containing the entire WU batch and putting that entire batch within the "fast Queue" seems a reasonable and responsible thing to request for numerous reasons.

Now, I'm certainly not suggesting that if machines suddenly start moving over, they be immediately placed into list of fast responders. This is something that requires a consistent track record, I don't know, say 10 consecutive days of returning results within 24 hrs?

Turn your machine off whilst you go on holiday for say a week; and you must again certify your machine as a fast responder? These are just thoughts and not truly intended to be implementation perspectives. The masters that run WCG have much better insight to this than I, a simple mortal.

When we had single core systems, that could hardly get a WU returned in 24 hours to 48 hours, the 7 / 12 / 14 / day return validation methodology appears to have been a sound and reasonable decision which ensured the most WU's were returned completed which eliminated copious re-work of WU's.

Today, many are returning WU results using unbelievably fast, multi-core CPU's tied to systems with almost incomprehensible quantities of memory; HDD space; and actual CPU processing capacity not available only a few years prior. These technological advances have been the single biggest contributor to being able to return many WU's within a 24 hour period of time. In my mind's eye, as it relates to not having a "queue for 24 hour turnaround" seems strikingly similar to one justifying using a howitzer as the appropriate tool to kill unwanted mice.

As an aside, while I can't say this with any real certainty, one could argue the point, should WCG implement a "Fast Queue" response for each and every project, additional projects regardless of size could benefit in reduced time to completion of the complete project.

BTW, I just looked at my PV queue.

When I started this thread, I was at a PV queue depth of 45. I'm quickly approaching a PV queue depth of 60. Should the results continue to pile up like this, some possibility exists my simple little machine could have a PV queue depth of over 100 before long. Surely WCG can do something about getting those of us in PV purgatory out of this mess.

Wouldn't it be simply amazing if by the adoption and implementation of this seemingly simple change, a project could get all it's WU's performed say 6 / 7 months earlier than initially anticipated and a cure for something could be found in time to save someone that would have otherwise been lost? Dreamy eyed, Pie in the Sky, perhaps... but what if it worked out to be the one it saved was you, one of yours or one of your friends, would it still be "impracticable?"

So wadaya say WCG guys can you do something to encourage a seemingly simple behavioral change within the community?

(Hey Camera guy, you gettin' all this? Opps sorry, I thought this was a Sham-Wow Commercial! )