A computation of D(9) using FPGA Supercomputing — mentioned by vlovich123

Tue, 27 Jun 2023 14:59:51 GMT

> The problem: The calculation of these terms on normal processors is slow and also the use of GPUs as currently the fastest hardware accelerator technology for many AI applications is not efficient for this algorithm. > https://arxiv.org/pdf/2304.03039.pdf At this rate, a single FPGA processes about 5.2 α values per second, taking 47’000 FPGA hours to compute D(9) and > https://arxiv.org/pdf/2304.00895.pdf The only thing left to say is that we run the algorithm on Nvidia A100 GPUs. 5311 GPU hours and 4257682565 matrix multiplications later, we got the following value for the ninth Dedekind…

Paper and discussion history. A mention is not an endorsement.

A computation of D(9) using FPGA Supercomputing — mentioned by bazzargh

Tue, 27 Jun 2023 13:07:15 GMT

the van Hirtum paper now refers to that one as confirmation of its result https://arxiv.org/abs/2304.03039 beaten to the punch by a couple of days!

Paper and discussion history. A mention is not an endorsement.

A computation of D(9) using FPGA Supercomputing · HackerPapers discussions

A computation of D(9) using FPGA Supercomputing — mentioned by vlovich123

A computation of D(9) using FPGA Supercomputing — mentioned by bazzargh