I don't get this, before casp my rac was climbing like mad. Now in casp even with getting 8-10 points over the claimed credit my RAC is starting to flat line.

Boinc won't report for the whole day if i don't budge it along. not sure if this is the cause for the flat line as time is longer than the points gained?

I run about 6 tasks in 10 hrs on a dual core on a 4 hour run per task, but don't understand why Boinc won't report more often. Also before casp, I think Boinc was reporting more often, since I would come home and the tasks waiting to report were only 2 or 4 at max.

I don't get this, before casp my rac was climbing like mad. Now in casp even with getting 8-10 points over the claimed credit my RAC is starting to flat line.

Boinc won't report for the whole day if i don't budge it along. not sure if this is the cause for the flat line as time is longer than the points gained?

I run about 6 tasks in 10 hrs on a dual core on a 4 hour run per task, but don't understand why Boinc won't report more often. Also before casp, I think Boinc was reporting more often, since I would come home and the tasks waiting to report were only 2 or 4 at max.

You don't need BOINC to report any more than once per day - your RAC will even out to the same score regardless (well almost), but I think 1050 RAC is about right for your CPU anyway (unless it's overclocked?). My E6420 got around that at 3.2GHz.

The different jobs do get different amounts of credit on different platforms - I notice it more on my AMD machine.

---

I don't get this, before casp my rac was climbing like mad. Now in casp even with getting 8-10 points over the claimed credit my RAC is starting to flat line.

Boinc won't report for the whole day if i don't budge it along. not sure if this is the cause for the flat line as time is longer than the points gained?

I run about 6 tasks in 10 hrs on a dual core on a 4 hour run per task, but don't understand why Boinc won't report more often. Also before casp, I think Boinc was reporting more often, since I would come home and the tasks waiting to report were only 2 or 4 at max.

You don't need BOINC to report any more than once per day - your RAC will even out to the same score regardless (well almost), but I think 1050 RAC is about right for your CPU anyway (unless it's overclocked?). My E6420 got around that at 3.2GHz.

The different jobs do get different amounts of credit on different platforms - I notice it more on my AMD machine.

How much does overclocking affect credit? I thought it would give you more if you OC'd a bit. I am boosting my cpu from 3ghz (default) to 3.158 (350mhz x 9.0) according to the ASUS AI Suite. I could push more but my operating temp would climb higher than what I read is ok for the cpu.

How much does overclocking affect credit?

Since most BOINC tasks are CPU bound, increasing the CPU by a certain percentage will usually lead to a similar gain in RAC (usually a fraction less for a variety of factors, but in the main the point is valid).

However, if you're only after credit, you're better off with other projects with optimized apps or 64-bit gains, etc - these will gain you a lot more RAC than overclocking.

Still, Rosetta is a good project, so any increase in crunching power you care to send its way will be appreciated, I'm sure : )

---

Crunching Rosetta as a member of the Whirlpool BOINC Teams

RAC, obviously is based on credit earned over time, and if you speed up your machine, then you will earn credit at a higher rate.

It is sort of like comparing your food intake to your body weight. Body weight being an average. And food intake varying rather dramatically depending on the timeframe over which you study it. If I guage food intake over 4 hour periods of time, I'm going to see some zeros while you are sleeping, and some very large numbers when you eat a big meal. If you skip a meal, you may be starving... but won't have lost any noticible weight yet.

To further the analogy, running the current CASP work, (as compared to earlier tasks that were being sent out) is like changing the foods that you eat. And so the relative effects on body weight over time depend on how many calaories per serving the new food has as compared to the previous food.

To take it even further, one body may react differently to the new food then another. Different people have different reactions to sugars, different rates of metabolism, etc. And this is very similar to how different machines may earn credit with any new type of tasks.

...but isn't credit supposed to be concistant, and based on the number of operations my computer does?? yes. But it's just an estimate. And, if you were a dietician, you would say the difference in reaction to the new foods between two people was quite predictable. And if you had perfect information about the computers, and Rosetta work being done at the time, I suppose a difference in reaction between Intel and AMD (as one example) would be predictable as well. In fact, just by external observation, many have noted that the size of the machine's L2 cache makes a considerable difference in the rate of work it can be expected to complete.

The RAC chart, once well-established, takes considerable time to move 1%... as does body weight. ...I need a doughnut!

---

Rosetta Moderator: Mod.Sense