Module zve64x_fixed_point_helpers 

Opaque helpers for Zve64x extension

Functions

avg_add

Averaging signed add: (vs2 + src) >> 1 with rounding per vxrm.

avg_addu

Averaging unsigned add: (vs2 + src) >> 1 with rounding per vxrm.

avg_sub

Averaging signed subtract: (vs2 - src) >> 1 with rounding per vxrm.

avg_subu

Averaging unsigned subtract: (vs2 - src) >> 1 with rounding per vxrm.

check_narrowing_sew

Verify that the destination SEW is valid for narrowing (must be at most 32 in Zve64x).

check_vs2_narrowing_alignment

Check that vs2 for a narrowing instruction is aligned to 2 * group_regs and fits in [0,32).

nclip

Narrowing signed clip: read a 2*SEW signed element from vs2, shift right arithmetically with rounding, saturate to signed SEW range.

nclipu

Narrowing unsigned clip: read a 2*SEW element from vs2, shift right by shamt with rounding, saturate to unsigned SEW range, set vxsat on clamp.

read_wide_element_u64^⚠

Read a 2*SEW-wide element as u64 from the double-width source register group of a narrowing instruction.

rounded_sra

Perform a rounded arithmetic right shift of val (sign-extended to SEW) by shift bits.

rounded_srl

Perform a rounded right shift of val by shift bits (logical / unsigned).

sat_add

Saturating signed add: vs2 + src, clamped to [-(2^(SEW-1)), 2^(SEW-1) - 1].

sat_addu

Saturating unsigned add: vs2 + src, clamped to [0, 2^SEW - 1].

sat_sub

Saturating signed subtract: vs2 - src, clamped to [-(2^(SEW-1)), 2^(SEW-1) - 1].

sat_subu

Saturating unsigned subtract: vs2 - src, clamped to [0, 2^SEW - 1].

smul

Fractional multiply with rounding and saturation: vsmul.