Load ZA array vector
This instruction performs a contiguous load of bytes to a ZA array vector from the memory address generated by a 64-bit scalar base plus an optional immediate offset multiplied by the current vector length in bytes. The ZA array vector is selected by the sum of the vector select register and the same immediate offset, modulo the number of bytes in a Streaming SVE vector.
This instruction is unpredicated.
The load is performed as contiguous byte accesses, with no endian conversion and no guarantee of single-copy atomicity larger than a byte. However, if alignment is checked, then the base register must be aligned to 16 bytes.
This instruction does not require the PE to be in Streaming SVE mode, and it is expected that this instruction will not experience a significant slowdown due to contention with other PEs that are executing in Streaming SVE mode.
Variants: FEAT_SME (PROFILE_A)
31 | 30 | 29 | 28 | 27 | 26 | 25 | 24 | 23 | 22 | 21 | 20 | 19 | 18 | 17 | 16 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
1 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |||||||||||
op | Rv | Rn | off4 |
---|
LDR ZA[<Wv>, <offs>], [<Xn|SP>{, #<offs>, MUL VL}]
if !IsFeatureImplemented(FEAT_SME) then EndOfDecode(Decode_UNDEF); constant integer n = UInt(Rn); constant integer v = UInt('011':Rv); constant integer offset = UInt(off4);
CheckSMEAndZAEnabled(); constant integer SVL = CurrentSVL; constant integer dim = SVL DIV 8; bits(64) base; constant integer moffs = offset * dim; bits(SVL) result; constant bits(32) vbase = X[v, 32]; constant integer vec = (UInt(vbase) + offset) MOD dim; constant boolean contiguous = TRUE; constant boolean nontemporal = FALSE; constant boolean tagchecked = n != 31; constant AccessDescriptor accdesc = CreateAccDescSME(MemOp_LOAD, nontemporal, contiguous, tagchecked); if IsFeatureImplemented(FEAT_TME) && TSTATE.depth > 0 then FailTransaction(TMFailure_ERR, FALSE); if n == 31 then CheckSPAlignment(); base = SP[64]; else base = X[n, 64]; bits(64) addr = AddressAdd(base, moffs, accdesc); constant boolean aligned = IsAligned(addr, 16); if !aligned && AlignmentEnforced() then constant FaultRecord fault = AlignmentFault(accdesc, addr); AArch64.Abort(fault); for e = 0 to dim-1 Elem[result, e, 8] = AArch64.MemSingle[addr, 1, accdesc, aligned]; addr = AddressIncrement(addr, 1, accdesc); ZAvector[vec, SVL] = result;
If PSTATE.DIT is 1, the timing of this instruction is insensitive to the value of the data being loaded or stored.