Bootstrap files are now loaded very slowly. A chunk is read, then deserialized, then processed. Allthese operations are independent, and they can be parallelized efficiently in a pipeline, where every thread with these jobs will poll the others on whether new data is available.
Keep in mind that blocks have to be read approximately in order. Otherwise the orphans can go too far and be blocked as "too far in the future".
Bootstrap files are now loaded very slowly. A chunk is read, then deserialized, then processed. Allthese operations are independent, and they can be parallelized efficiently in a pipeline, where every thread with these jobs will poll the others on whether new data is available.
Keep in mind that blocks have to be read approximately in order. Otherwise the orphans can go too far and be blocked as "too far in the future".