For NVIDIA TESLA V100 I know a hashcat performance for SHA1: it is about 17000 MH/s. Performance index for V100 is FP32 5120, FP64 2560. But what about supercomputers? For example Summit supercomputer from USA is 200 petaFLOPS in peak. Is any was to calculate it's hash per second performance for SHA1 for example?
The number of NVIDIA TESLA V100 in the Summit supercomputer is documented, and it's reasonable to assume that most of its hashcat performance for SHA-1 would come from that: according to this source on the Power9 CPU ISA, there is no SHA-1 instructions for the Power CPUs, and then these benchmarks give an order of magnitude on other CPUs.
Isn't that enough for answering the question?
Caveat: a truly powerful specialized SHA-1 cracker would use ASICs, like bitcoin miners do, for even more performance.