ssm_h_state_f16_to_f32

spark_model::layers::ops

Function ssm_h_state_f16_to_f32 

Source 
pub fn ssm_h_state_f16_to_f32(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    src: DevicePtr,
    dst: DevicePtr,
    n: u64,
    stream: u64,
) -> Result<()>
Expand description

One-shot FP16 -> FP32 widening of one layer’s SSM h-state (ATLAS_SSM_H_FP16). n is the FP32 ELEMENT count of the destination.

Kernel: ssm_h_state_f16_to_f32(src, dst, n). Grid-stride; src and dst must not alias.