Function cublas_fp8_proj 

pub fn cublas_fp8_proj(
    gpu: &dyn GpuBackend,
    ptg_quant_k: Fp8ActQuant,
    scale_kmajor_k: KernelHandle,
    act_bf16: DevicePtr,
    act_fp8_scratch: DevicePtr,
    act_scale_scratch: DevicePtr,
    act_scale_kmajor_scratch: DevicePtr,
    fp8w: &Fp8Weight,
    out: DevicePtr,
    m: u32,
    n: u32,
    k: u32,
    stream: u64,
) -> Result<()>

2026-09-25: Quantize the activation with per_token_group_quant_fp8 (FP8 plus one scale per token and 128 of K), then run cublas_fp8_proj_prequant. The FP8 weight and its 128x128 block scales go to cuBLASLt as they are, with no dequant.

The scratch buffers need the extents listed on cublas_fp8_proj_prequant.