> Lets see AI take the AV2 spec and Dav1d code and try to make a working high performance AV2 decoder.
That sounds like one of these high-risk, high-reward things that are great for people / projects / companies who have nothing to lose, but is not a great baseline strategy for an established market player. AV2 is here with support from aomedia and its members. AV2 will be used, and we need a production-grade decoder regardless of where AI is at, so it makes much more conservative business sense to use established approaches (language: c/asm, devteam: ffmpeg/dav1d) as a starting point. While that's happening, we can dabble in AI and other risky stuff and see if it helps. If so, great, and if not, nothing lost.
dav2d author here - the 5x number is just where we currently are, it's not the theoretical limit. We're hopeful that a significant amount of the increase we observe in dav2d relative to dav1d is in math code, which should be easier to optimize using hand-written assembly or other algorithmic optimizations. If that holds true, the practical slowdown once everything has been optimized may be substantially less, possibly 2x.
> Isn't this all referring to the "R&D tax credit"?
No. The terminology is confusing. Section 174, specifically point 3, classifies software development as R&E as of 2022 and requires amortization over 5-15 years. This prevents software engineering salaries as a standard business tax deduction. This has nothing to do with the R&D tax credit (section 41).
Amortizing makes sense for things that have a huge up-front cost and value over time, like - indeed - a robot, or a laptop, or a house. But salaries are not that, they are a continuous cost. I can take a loan on a house (mortgage), and laptops can be bought on 0% payment plans over 3 years. But how would that work for salaries? I'm forced to take out loans to pay the duplicate tax over salaries (and other minor expenses) over the amortization period. That's the insane bit here.
That sounds like one of these high-risk, high-reward things that are great for people / projects / companies who have nothing to lose, but is not a great baseline strategy for an established market player. AV2 is here with support from aomedia and its members. AV2 will be used, and we need a production-grade decoder regardless of where AI is at, so it makes much more conservative business sense to use established approaches (language: c/asm, devteam: ffmpeg/dav1d) as a starting point. While that's happening, we can dabble in AI and other risky stuff and see if it helps. If so, great, and if not, nothing lost.