I have a search algorithm that is several times faster than the @57fe https://bitcointalk.org/index.php?topic=5166284.msg52582231#msg52582231 algorithm in single-threaded mode. According to my calculations, in skilled hands, you can keep within a month on a good server in the cloud to calculate 110 tasks. Server costs will reduce the reward by about half. This is more interesting than simple mining.
The algorithm can only be shifted to FPGA. Modular arithmetic does not imply the use of ASIC.

If there are interested people, I can post the evidence in the video.

Python does not assume that code is closed when used by other people, so I can not publish the source code, and the prototype algorithm is implemented in python.

Sorry for my English, but I'm not a native speaker of this language