Does VREM.VX trap on divide by zero?
No. The official V extension defines the divide-by-zero result: the remainder is the dividend, and the instruction itself does not raise a divide-by-zero trap.
VREM.VX
RISC-V VREM.VX Instruction Details
Instruction ManualOPMVX (OP-V)vector-scalar signed integer remainder; writes each remainder to vd, and divide by zero does not trap.
Instruction Syntax
vrem.vx vd, vs2, rs1[, v0.t]
Operand Breakdown
Syntax: vrem.vx vd, vs2, rs1[, v0.t]; omit the mask for vm=1, include v0.t for vm=0. 32-bit OP-V: opcode=1010111, funct3=110 (OPMVX), funct6=100011.
vs2[i] is the dividend; x[rs1] supplies the scalar divisor: XLEN>SEW takes low SEW bits; XLEN<SEW sign-extends even for unsigned arithmetic; x0 supplies zero. Compute signed SEW remainders. A zero divisor returns the dividend; a nonzero remainder has the dividend's sign, and minimum divided by -1 has remainder zero.
Active body satisfies vstart≤i<vl with the mask enabled. Inactive/tail follow vma/vta: retain or agnostic, without a fixed value to rely on. vl=0 updates no destination elements.
Same-width sources/destination may overlap. Masked vd cannot be v0; v0 cannot provide both the EEW=1 mask and SEW data. Group starts align to EMUL, EMUL≤8.
Require V 1.0, enabled vector state and valid supported vtype/SEW. The animation fixes m1 and vstart=0; VLEN=4096. Steps are not processor cycles.
VVector IntegerDivide/Remainder
Instruction Behavior
VREM.VX is a RISC-V V extension vector-scalar signed integer remainder instruction. Each active element uses vs2 as the dividend and the x[rs1] scalar value as the divisor, computes the SEW-width remainder, and writes vd; zero divisors and signed overflow cases use the official specified results without a divide-by-zero trap.
VREM.VX Decode And Execute Animation
Starts from OP-V encoding fields, then shows how VREM computes each signed remainder and applies the official divide-by-zero and signed-overflow special cases.
Instruction input
vrem.vx
Require V 1.0, enabled vector state and valid supported vtype/SEW; fix LMUL=m1 and vstart=0. Fix VLEN=4096 and VLMAX=4096/16=256; equal-width sources/destination may overlap. The signed quotient rounds toward zero; a nonzero remainder has the dividend’s sign. MIN/-1 has remainder zero. A zero divisor returns the dividend. XLEN>SEW takes low scalar bits; XLEN<SEW sign-extends even for unsigned arithmetic; x0 supplies zero. Masked vd cannot be v0; v0 cannot supply both mask and SEW data. Enter nonnegative element bit patterns. VL=0 leaves vd unchanged. Only body is shown; no definite inactive/tail values. Teaching stages are not processor cycles.
Execution Context
OP-V encodingvrem.vx
funct6
100011
vm
0
vs2
01000
rs1
01011
funct3
110
vd
00100
opcode
1010111
lane
0
1
2
3
4
5
6
7
v8
0x0064
0xffd6
0x8000
0x007f
0xffff
0x0001
0x0064
0xffd6
%
%
%
%
%
%
%
%
x11
0xffff
0xffff
0xffff
0xffff
0xffff
0xffff
0xffff
0xffff
v0.t
1
0
1
1
1
0
1
1
v4
...
-
...
...
...
-
...
...
Current step
Show OP-V 32-bit encoding fields
vrem.vx uses OP-V encoding. The animation places fixed fields, register fields, vm, and the rs1 field in one encoding strip.
This animation shows OP-V field decode, active-element reads, signed remainder, mask control, and vd writeback. These are teaching stages; they do not model pipelines, caches, or timing.
Quick Understanding & Search Notes
VREM.VX does not raise a divide-by-zero exception. For active elements within the current vl, it reads vs2 as the dividend and the x[rs1] scalar value as the divisor, writes the signed remainder, and with vm=0 only body elements whose v0.t bit is 1 execute.
32-bit OP-V: opcode=1010111, funct3=110 (OPMVX), funct6=100011.
vs2 is the dividend and the divisor comes from the x[rs1] scalar value; each lane is computed independently with no cross-lane combination.
Operands are interpreted as signed two's-complement SEW integers, and the result is the remainder.
When the divisor is zero, the remainder result is the dividend SEW-bit pattern.
For signed minimum divided by -1 overflow, VDIV writes the dividend and VREM writes zero.
With vm=0, v0.t is the execution mask; masked-off and tail elements follow the current vtype policies.
The signed quotient rounds toward zero; a nonzero remainder has the dividend’s sign. XLEN>SEW takes low scalar bits; XLEN<SEW sign-extends even for unsigned arithmetic; x0 supplies zero. Require enabled vector state and valid supported vtype/SEW; vl=0 leaves vd unchanged.
Vector Execution Context
Syntax: vrem.vx vd, vs2, rs1[, v0.t]; omit the mask for vm=1, include v0.t for vm=0. 32-bit OP-V: opcode=1010111, funct3=110 (OPMVX), funct6=100011.
vs2[i] is the dividend; x[rs1] supplies the scalar divisor: XLEN>SEW takes low SEW bits; XLEN<SEW sign-extends even for unsigned arithmetic; x0 supplies zero. Compute signed SEW remainders. A zero divisor returns the dividend; a nonzero remainder has the dividend's sign, and minimum divided by -1 has remainder zero.
Active body satisfies vstart≤i<vl with the mask enabled. Inactive/tail follow vma/vta: retain or agnostic, without a fixed value to rely on. vl=0 updates no destination elements.
Same-width sources/destination may overlap. Masked vd cannot be v0; v0 cannot provide both the EEW=1 mask and SEW data. Group starts align to EMUL, EMUL≤8.
Require V 1.0, enabled vector state and valid supported vtype/SEW. The animation fixes m1 and vstart=0; VLEN=4096. Steps are not processor cycles.
Common Usage Scenarios
Element-wise signed remainder
Assume enabled V/VS, valid m1, supported SEW, VLEN≥128 and XLEN=64, vl=4, vstart=0, v0.mask[0..3]=1111 and initialized sources. SEW=8, a1=3, v12=[249(-7),7,128(-128),5] → v8=[255(-1),1,254(-2),2]. Inactive/tail follow vma/vta; vl=0 leaves vd unchanged.
Masked vector remainder
Assume enabled V/VS, valid m1, supported SEW, VLEN≥128 and XLEN=64, vl=4, vstart=0, v0.mask[0..3]=1111 and initialized sources. SEW=8, a1=3, v12=[249(-7),7,128(-128),5] → v8=[255(-1),1,254(-2),2]. Inactive/tail follow vma/vta; vl=0 leaves vd unchanged.
Pre-Use Checklist
Syntax Check
- Syntax: vrem.vx vd, vs2, rs1[, v0.t]; omit the mask for vm=1, include v0.t for vm=0. 32-bit OP-V: opcode=1010111, funct3=110 (OPMVX), funct6=100011.
- vs2[i] is the dividend; x[rs1] supplies the scalar divisor: XLEN>SEW takes low SEW bits; XLEN<SEW sign-extends even for unsigned arithmetic; x0 supplies zero. Compute signed SEW remainders. A zero divisor returns the dividend; a nonzero remainder has the dividend's sign, and minimum divided by -1 has remainder zero.
Semantic Check
- Active body satisfies vstart≤i<vl with the mask enabled. Inactive/tail follow vma/vta: retain or agnostic, without a fixed value to rely on. vl=0 updates no destination elements.
- Same-width sources/destination may overlap. Masked vd cannot be v0; v0 cannot provide both the EEW=1 mask and SEW data. Group starts align to EMUL, EMUL≤8.
- Require V 1.0, enabled vector state and valid supported vtype/SEW. The animation fixes m1 and vstart=0; VLEN=4096. Steps are not processor cycles.
Pitfalls / Common Confusions
A zero divisor does not trap; software that needs an error path must test divisors explicitly.
The signed minimum divided by -1 case is architecturally defined: divide writes the dividend, and remainder writes zero.
The second source comes from x[rs1] and is formed as a SEW-width element by the .vx scalar rule; there is no scalar-divided-by-vector reverse form.
FAQ
How is VREM.VX related to VDIV?
VREM.VX writes the remainder; the paired VDIV form writes the quotient for the same divide relation. Both use the same divide-by-zero and signed-overflow special-case rules.