So how do these stack up against running that same workload (simulations) on a powerful GPU? Often a sim requires running the same operation over thousands of d

So how do these stack up against running that same workload (simulations) on a powerful GPU? Often a sim requires running the same operation over thousands of different cells: it seems really suited to the “warp” style of parallelism, where all cores in a set run the same instruction in lockstep. Or SIMD, but that’s usually a bit more effort to port. LLVM has a spirv backend these days, so does POWER actually make sense for these workloads outside of places where it’s difficult to port your codebase?