pto.vpack¶
pto.vpack is part of the Data Rearrangement instruction set.
Summary¶
Pack two wide vectors into one narrower vector.
Mechanism¶
pto.vpack narrows the two source vectors and concatenates the narrowed halves into one destination vector. The exact narrowing mode is controlled by %part and the selected target profile.
Syntax¶
PTO Assembly Form¶
vpack %dst, %src0, %src1, %part
AS Level 1 (SSA)¶
%result = pto.vpack %src0, %src1, %part : !pto.vreg<NxT_wide>, !pto.vreg<NxT_wide>, index -> !pto.vreg<2NxT_narrow>
Inputs¶
| Operand | Type | Description |
|---|---|---|
| %src0 | !pto.vreg<NxT_wide> |
First wide source vector |
| %src1 | !pto.vreg<NxT_wide> |
Second wide source vector |
| %part | index |
Packing mode or submode selector |
Expected Outputs¶
| Result | Type | Description |
|---|---|---|
| %result | !pto.vreg<2NxT_narrow> |
Packed narrow vector built from both sources |
Side Effects¶
This operation has no architectural side effect beyond producing its destination values. It does not implicitly reserve buffers, signal events, or establish memory fences.
Constraints¶
Constraints
- Packing is a narrowing conversion; values that do not fit the destination width follow the truncation or saturation behavior of the selected form.
- Lowering MUST preserve the destination ordering between the first and second source halves.
Exceptions¶
Exceptions
- The verifier rejects illegal operand shapes, unsupported element types, and attribute combinations that are not valid for the selected instruction set or target profile.
- Any additional illegality stated in the constraints section is also part of the contract.
Target-Profile Restrictions¶
Target-Profile Restrictions
- A5 is the most detailed concrete profile in the current manual; CPU simulation and A2/A3-class targets may support narrower subsets or emulate the behavior while preserving the visible PTO contract.
- Code that depends on an instruction-set-specific packing, selector, or permutation mode should treat that dependency as target-profile-specific unless the manual states cross-target portability explicitly.
Examples¶
for (int i = 0; i < N; i++) {
dst[i] = truncate(src0[i]);
dst[N + i] = truncate(src1[i]);
}
Related Ops / Instruction Set Links¶
- Instruction set overview: Data Rearrangement
- Previous op in instruction set: pto.vperm
- Next op in instruction set: pto.vsunpack