Function quantize_bf16_to_fp8 

pub fn quantize_bf16_to_fp8(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    input: DevicePtr,
    output: DevicePtr,
    row_scales: DevicePtr,
    n: u32,
    k: u32,
    stream: u64,
) -> Result<()>

2026-09-25: Quantize a BF16 weight [N, K] to FP8 E4M3 [N, K] with one f32 scale per row [N], one CTA per row (kernel quantize_bf16_to_fp8 in kernels/gb10/common/dense_gemv_fp8w.cu). DenseWeight::quantize_to_fp8 calls it; the resulting Fp8DenseWeight feeds the row-scaled FP8 kernels such as fp8_gemm_n128_row_scaled.