For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo struct
QuantizeDynamicScaledFloat8RowBounded
struct QuantizeDynamicScaledFloat8RowBounded
Registers the mo.quantize_dynamic_scaled_float8.row_bounded graph op.
Same numerics as mo.quantize_dynamic_scaled_float8, but the rows it
touches stop at a count the GPU publishes rather than at the height of the
input tensor. The EP MoE down projection needs this: its activation buffer
is sized for the worst-case dispatch, the dispatch kernel writes the live
row count into row_offsets, and the host never learns that count.
A separate symbol rather than an operand on the shared op, because the graph compiler's RMS-norm and all-reduce fusion patterns match the shared op by its two-operand signature.
Implemented traits
Methods
execute
static def execute[input_type: DType, scales_type: DType, output_type: DType, //, group_size_or_per_token: Int, target: StringSpan[ImmStaticOrigin]](output: ManagedTensorSlice[IOSpec[_, _].Output, static_spec=output.static_spec], scales: ManagedTensorSlice[IOSpec[_, _].Output, static_spec=scales.static_spec], input: ManagedTensorSlice[IOSpec[_, _].FusedInput, static_spec=input.static_spec], scale_ub: Float32, row_offsets: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=row_offsets.static_spec], ctx: DeviceContext)