Em Sex, 2006-04-14 às 01:33 -0700, Trent Piepho escreveu: > So anyway, here is my version of linear interpolation. It has a simpler > formula in the code, and much better rounding behavior, as you can see in the > plot here: http://www.speakeasy.org/~xyzzy/ber-snr.png Sounds good. I'll replace the corresponding parts at my code. May I include your Signed-off-by at the patch? > > static int ber_lut[][2] = { > { 1.00 * 2047, 4798 }, > { 0.99 * 2047, 5380 }, > { 0.98 * 2047, 6032 }, > { 0.95 * 2047, 8502 }, > { 0.90 * 2047, 15066 }, > { 0.85 * 2047, 26696 }, > { 0.80 * 2047, 47306 }, > { 0.75 * 2047, 83826 }, > { 0.70 * 2047, 148539 }, > { 0.65 * 2047, 263210 }, > { 0.60 * 2047, 466406 }, > { 0.40 * 2047, 4598482 }, > { 0.30 * 2047, 14439082 } }; There is one remaining stuff: From the tests I did, BER next to 4798 generate a MPEG stream with errors on almost every block. So, the curve (that were just the closest curve from the previous values) is not representative, from users perspective, in terms of quality. I think we should just drop the old values and create a more realistic curve, since, from "Meaningful reporting of SNR", it seems that snr is being used as a "quality indicator" without any linkage with the real SNR. I guess the values of the curve are associated to SNR by a curve map. Unfortunately, the public datasheet doesn't show this curve. Also, the more relevant part to final user seems to be that ones from BER varying from 1 to 5500, with is mapped as 100% (this is the values from the original curve). Going to userspace, very few variations on dB means high quality losses, so, userspace applications need to do some conversions to make it valuable to the user. Just showing SNR makes no sense to users. Cheers, Mauro. _______________________________________________ linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
