[XLA:GPU] cuBlas supports complex floats, use gemm instead of our O(n^3) implementation
Also increase test coverage for C64 a bit. PiperOrigin-RevId: 207297946
Loading
Please sign in to comment
Also increase test coverage for C64 a bit. PiperOrigin-RevId: 207297946