These are not trivial algorithms to implement, and the other factorization records require months of work from implementation experts. It's not an easy task, and theory work stands independently of implementation effort.
Still, if this new algorithm could threaten 1024 bit RSA using 10,000 computers for 10,000 days after a 10,000x speed up from optimisation, it should be able to solve the RSA-896 factoring challenge with a single computer for a single day without optimisation, shouldn't it?
After all, 2^896 is 38 orders of magnitude smaller than 2^1024.
After all, 2^896 is 38 orders of magnitude smaller than 2^1024.