My Rac on my machine was around 95,000 and actually climing slowly
after all the problems weve had on the wu side my RAC has fallen to 68,000 and doesnt seem to want to rise particularly quickly if at all

has anybody else noticed this ?

the work units seem to take the same time so is there something differant about them or the coversion to credit ?

Ian

MW started a new Validation system which employs Wingmen to validate results, in addition to other server side checks that also take place in parrallel - in a similar way to what you saw at Collatz. The validator here does not require every WU to have a wingman to check it - I believe its 50% at random I think.

The result of that is WUs will go into pendings until a wingman confirms the result. Initially therefore pendings will fill up slowly until the number of WUs going into pendings are balanced by the number going out having been validated. The number in pendings can be checked via your account page.

It will even off shortly, and your RAC will start to climb back up once the number leaving pendings exceed those going in. Its a short term one off statistical quirk, give it 10 days or so and you will be back to normal having "filled" a "buffer" in pendings.

Regards
Zy

It started for me on the 15th March when my RAC dropped 30% overnight. I thought it was a problem with my PC's. I even altered the project split to give Milkyway more cruch time from 50% to 75% over SETI and my Milkyway RAC still kept dropping.

I hope it starts to turn around soon.

The project did have a couple of problems that even recently still effect wu output when i started this thread my RAC was at 95,000 and shot down to end up at 68,000 now i am doing more work constantly the RAC has begun to rise again it's at 141,000 at time of writing and still rising slowly as before
as long as work units are produced this trend should continue for both of us

best regards
Ian

I swapped one of my GPUs from Collatz to Milkyway in early April. The aim was to see what RAC the GPU would settle at. During April, it met 3 server side problems getting new work, and twice it has raised it's output to 132K and 135K before falling. This time the RAC is rising towards 130K for the third time.

I wonder how much further it might rise before settling to an average level?

---

Go away, I was asleep

I swapped one of my GPUs from Collatz to Milkyway in early April. The aim was to see what RAC the GPU would settle at. During April, it met 3 server side problems getting new work, and twice it has raised it's output to 132K and 135K before falling. This time the RAC is rising towards 130K for the third time.

I wonder how much further it might rise before settling to an average level?

It won't, at least based on my experience ...

It has to do with strict FIFO, the fact that MW does not issue more than x tasks to you, the Resource Scheduler and Work Fetch modules are bent (along with RR Sim, but that is another story), and the timing of stat exports and export reads ...

If you look at my 60 day chart though I have the same GPU project in play on 5 systems (which means any one project should be running on at least one system at all time, though GPU Grid is falling because I am running out of Nvidia cards); ...

You can see an average line in there however ... but it is very annoying ...

Oh, it has been rising, and will rise some more because over the last two weeks I have been migrating from Win XP-32 to Win7-64, added more FreeHAL tasks now I have more memory, and a new H5870 and another today (Its on the UPS truck for delivery, "WEST SACRAMENTO, CA, US 04/30/2010 5:53 A.M. OUT FOR DELIVERY") ...