If you are using BOINC, points are awarded based on everybody's average. I wonder whether that is the problem. If you are using the UD agent, there can be some very minor differences between points and statistics, depending on the source.

Yours sounds like a larger difference than usual but there are only 2 results so far. Since you are a fairly new member and it takes several days to generate a result for you, I'm not sure whether the technicians have enough information to comment.

They may ask for more if they want to track it.

Points are also influenced by your CPU type, amount of memory, and amount of virtual storage (default 1 Gb ... change to 10 Gb in the device profile for best results if you have that much disk drive space available. It isn't all used for the current projects.)

I got the spead sheet from the forum only.I dont remember exactly from which thread i downloaded it.Your replies were somewhat understandable for me.Ill have to go through them again once or twice to get a good grasp of what you have said in there.
Thanks anyway.

I have found that the benchmarks for both UD and BOINC tend to undervalue my desktop machine, and overvalue my Laptop machine, compared to the expectations from other std benchmarks such as SISOFT SANDRA.

If faster machines are in general undervalued (I dont know this), then BOINC has the advantage of awarding the quorum average number of points. So in my case, BOINC always awards my fast Desktop more points than it claims. My slower laptop is usually awarded less than it claims.

Benchmark programs can be pretty complicated things in themselves. I think the benchmark routines in UD and BOINC are probably fairly simple (obviously the main effort needs to go into the project, not the benchmark). But with BOINC the quorum average idea probably irons out the differences. In any case, crunching is not just about points, is it?

I have just benchmarked both my machines. My Laptop, a Pentium M 1.86GHz 512MB gets 1619 double precisision MIPS and 3368 Integer MIPS. My desktop, a 3GHz Pentium D only gets 1506 double precisision MIPS and 2484 Integer MIPS for each core !! As I said, other benchmarks tell me there is nothing wrong with my Desktop.

I think another factor may be how busy the pc is at the moment UD/BOINC performs a benchmark - if the cpu is busy at that moment with other tasks, a low result may be obtained which will undervalue the next result. I schedule BOINC to suspend at the same time I have scheduled an antivirus/antispyware scan to start.

Based on my statistical analysis of 500 entries...

H3 = Actual CPU
I3 = Actual Memory
J3 = Actual Storage
K3 = Actual Network

Overall PC performance =
ROUND(ROUND(H3*0.3846,0)+ROUND(I3*0.2451,0)+ROUND(J3*0.044,0)+ROUND(K3*0.3446,0),0)

Assumptions:
Not a good sample as it comes from one PC only
CPU variation is mostly close to 170+.
Variation of memory is only done for 512M and 2GB.
Variation of Storage is done by varying the extremes on webpage
Network is not varied at all (so, it's really a constant offset).

With the above assumptions, Actual vs Estimate is 99% at +-2 points


D3 = Overall PC Performance (this is provided "after" your points are uploaded, or u can use the number above)
E3 = Runtime in seconds
Points = ROUND(D3*E3/19025.7,0)

Actual vs Estimate is 92% at +-2 points
& 61% at 0 points variation.

No idea what the constants mean...