Adywebb, I can fully understand your motivation, your efforts are very commendable

Hi wgbjr
Your comments are always welcome, feel free to post here anytime

It would appear that the aforementioned member has a couple of HTT machines dedicated to the WCG and do little else but process work units
So as you suggested, this could have something to do with him showing an overall increase

Adywebb, the aforementioned member has two identical machines
Let's hope he will test your theory and enlighten us, regarding his findings for the benefit of everyone

You certainly do go the extra mile Sir!

You sound like David Autumns, you realise that there is no cure for your crunching addiction

I am looking forward to hearing the results of your findings, as I am sure are many other members who frequent the Forums

Graham, both BOINC and UD run at the lowest priority on a Windows box. Thus, when I was making the switch, I had both running and saw both with just under 50 percent in task manager. I would exepect on any compter that has a single CPU, that is what you are going to see. If you muck with the priority for one, that will simply leave the other one idle as it will not see any free cycles. I'm no hardware guru by ANY stretch of the imagination so if these HTT chips are simply single CPUs that "fool" windows, I would expect that you'd really see the same thing there. On a true multi-CPU box, you could if fact run both BOINC and UD keeping them on separate CPUs. In that case, it's just like having multiple machines. Given how many machines hit the upper limits on the UD benchmarks, I can't imagine a multi-CPU box that wouldn't. In that case, running UD on any of the CPUs would mean being constrained by that. In other words, you'd be better off running BOINC. So just run BOINC on all the CPUs. If you look at the lower end machines not constrained by the UD benchmark limits, it may very well be that there is a scenario that concludes it is better to run UD on that machine.

It's way out

After just 1 days worth of crunching on this machine I bagged 1784 points in 24 hours. As you all know with UD 29.4 *24 = 705.6

I make that 2.53* the points of UD sorry folks to be a party pooper but that's just not right.

I've hurtled past Emilpalm yesterday after a week or so of just hanging onto his/her coat tails and I've had one machine with me on the trip to Glasgow off line.

I have to say it needs sorting

In the interests of accuracy of defining the correction factor I need to gather more evidence over the next couple of days

I'll let you know

Dave

Cheers Dave

Ah I see it is a bit of social engineering

The value (i.e. speed of crunching) being the same on both platforms (per processor or core)

For a level playing field though a correction factor (which can be done as we saw for the Linux Boinc) needs applying

UD is easier to install for the newbie as you don't need to go get your account code and it's prettier which might appeal more to your casual leave it on the screensaver cruncher (all of which adds processor time to the project which is the goal)

btw can anyone explain why the Boinc agent uploads a couple of files after each result

17/02/2006 03:52:14|World Community Grid|Started upload of faah0179_atv_x6711_00_1_0
17/02/2006 03:52:14|World Community Grid|Started upload of faah0179_atv_x6711_00_1_1

could it be that we are getting 2 results per work unit hence twice the points?

Me again