Valerii Zapodovnikov

BTW, you copied link very badly: https://discuss.pixls.us/t/port-the-lmmse-gamma-step-to-the-other-demosaicers/28233 Do you think that https://discuss.pixls.us/t/port-the-lmmse-gamma-step-to-the-other-demosaicers/28233 with infamous big LinearizationTable and the fact it is linear transfer means ON1 already uses it? IMHO, you...

> -fno-tree-pta Does this give any cool acceleration for that file? Or is this just a bug in gcc?

Just be careful, it is apparently very simple to totally code it wrong https://mobile.twitter.com/Cuboxel/status/1570043613828026369

>does not give much improvement But there are at least some 1 bit corrections if you "debug" with printf()? Did you check that changing any [only] one bit in the...

@carlesfernandez When is it going to get merged? I prefer this to @koloboxer variant, my friend told me his PPP glonass code follows ICD in Hamming correction. P.S. BTW, I...

> My experiments show that this case just means that all information bits are valid and there is a single error in `C_Sigma` So the words "multiple errors" is a...

Everywhere when parity bit does not pass you just remove the string. Why? Because even right out noise can pass this test then! And you even did not check it...

> enabling ECC does not give noticeable performance improvement, compared to 'main' branch. You did not test in bad environment, did you? When ionosphere is chaotic. Or when there is...

> skyview was blocked by trees and buildings most of time. Multipath really depends on whether you have a smart phase array like in a phone (bad, requires AI to...

How many times 1 bit was corrected?