Kprobe weird results

vbimport

#1

Today I tried scannig the same disc G04 silver top changing the read speed and turning RT chart on and off and I got pretty disappointed with KP(1.1.23 and 29).
First thnig the disc was burned and read with a 411@811 Z4 with firmware HS0K; all scans were make ECC sum 8 and for time reasons I scanned the first 10000 block at the beginning of the disc.

Scan#1: 2x, RT chart OFF, PI=45, PO=4.5 averages

Scan#2: 2x, RT chart ON, PI=34, PO=3.3 avg

Scan#3:4x, RT chart OFF, PI=33, PO=0.44 avg

Scan#4:4x, RT ON, PI=18, PO=0.2 avg

On all scans I tracked down the CPU usage and was well below the limits(4-30%).

IMHO KP has some problem of consistency and reliability. If I scan a disc at a lower speed I expect better results, but thisd is not he case(PO over 4!!!)
I’m not convinced either that scannig with RTC on gives lower errors becase of high CPU usage. I’m not new to programming and I know that in the digital world if a CPU can’t cope up with the software, the algorithm hangs or falls behind, it cannot happen that the algorithm tries to make calculations quicker and so it makes errors. Either it runs or it doesn’t.
Another thing that makes me think that is a software fault is that with RTC off the graph are flat, while with RTC on there are variation, as it should be ( we know that some part of a disc get written better of others).

I did another test with ECC set to one and guess what, turning realtime char on or off doesn’t make any differece, the graphs are very similar and so are the values:
RTC ON: PI=4.192 PO=0.067
RTC OFF: PI=4.343 PO=0.072
And this results with RTC on were obtaind with a CPU usage over 70%, this means that it is not a problem of CPU usage, as I stated before, but a software fault.

If my result are confirmed by someone else I suggest to change the default settings for the scannig af DVDs to ECC SUM 1 and maximum PI=35 and max PO=4 to be within the specs(this are obvoiusly conservative values on the PI side)

Comments and further investigation is apreciated.
Bye


#2

This is nothing new. KProbe will give varying results, and it has been known to vary with other factors, such as FW of the reader, etc. These variations seem to be fairly minor. But it’s an order of magnitude, you say. Well, an order of magnitude at such low values don’t really matter that much.


#3

sirluke79
Your results are completely normal, every scan varies from the next. This is a function of the drive itself, and has nothing to do with Kprobe. You are not measuring an absolute value on the disc, but the variable interaction between disc and drive. Temperature changes alone can account for the minor differences.


#4

“Your results are completely normal, every scan varies from the next”
It’s not a matter of small variations between a scan and the next one. The variations are systematic, directly correlated with the parameter variation.
And a variation between 18 and 33 is 83%, which is not small.

“Well, an order of magnitude at such low values don’t really matter that much.”
An order of magnitude is never a small variation. If I scanned a Princo instead of a Ritek an order of magnitude would have made the difference between a disc within the specs or completely out. Besides, a disc with an average PO of 4 sucks, IMHO.

Again I think the solution is to change the scanning parameters to what suggested in the first post, otherwise keeping a database of plots that don’t have any meaning and cannot be compared is completely useless.


#5

variation between 18 and 33 is 83%, which is not small

Yes, actually, it is small. Functionally, the scans represent identical discs. (they will perform the same). In the lower rance of error counts, differences several times that great are equally insignificant.

Media such as Ritek, which is “average to below average” in quality will often produce varying scan results such as you have seen. The worse the media, the greater the variations.

But jumping to a conclusing that the scans are “meaningless” based on results with one type of media is, well, silly.


#6

Why do you say that Riteks G04 are “average or below average” media. Most of the scans show PI under 10 avg and PO well below one avg.