iverilog/vvp/opcodes.txt

/*
 * Copyright (c) 2001-2003 Stephen Williams (steve@icarus.com)
 *
 *  $Id: opcodes.txt,v 1.78 2007/06/12 02:36:58 steve Exp $
 */



EXECUTABLE INSTRUCTION OPCODES

Instruction opcodes all start with a % character and have 0 or more
operands. In no case are there more than 3 operands. This chapter
describes the specific behavior of each opcode, in enough detail
(I hope) that its complete effect can be predicted.

General principles of Arithmetic:

The binary arithmetic instruction in general take three parameters,
the left operand, the right operand, and the base. The left operand is
replaced with the result, which is the same width as the left and
right operands.


* %add <bit-l>, <bit-r>, <wid>

This instruction adds the right vector into the left vector, the
vectors having the width <wid>. If any of the bits of either vector
are x or z, the result is x. Otherwise, the result is the arithmetic
sum.

See also the %sub instruction.


* %add/wr <bit-l>, <bit-r>

This is the real valued version of the %add instruction. The arguments
are word indices of the operands. The right operand is added into the
left operand.

See also the %sub/wr instruction.


* %addi <bit-l>, <imm>, <wid>

This instruction adds the immediate value (no x or z bits) into the
left vector. The imm value is limited to 16 significant bits, but it
is zero extended to match any width.

* %and <bit-l>, <bit-r>, <wid>

Perform the bitwise AND of the two vectors, and store the result in
the left vector. Each bit is calculated independent of other bits. AND
means the following:

	0 and ? --> 0
	? and 0 --> 0
	1 and 1 --> 1
	otherwise   x

* %assign/av <array-label>, <delay>, <bit>

The %assign/av instruction assigns a vector value to a word in the
labeled array. The <delay> is the delay in simulation time to the
assignment (0 for non-blocking assignment) and the <bit> is the base
of the vector to write.

The width of the word is retrieved from index register 0.

The base of a part select is retrieved from index register 1.

The address of the word in the memory is from index register 3. The
address is canonical form.

* %assign/mv <memory-label>, <delay>, <bit> (DEPRECATED)

the %assign/mv instruction assigns a vector value to a word in the
labeled memory. The <delay> is the delay in simulation time to the
assignment (0 for non-blocking assignment) and the <bit> is the base
of the vector to write.

The width of the word is retrieved from index register 0.

The base of a part select is retrieved from index register 1.

The address of the word in the memory is from index register 3. The
address is canonical form.

* %assign/v0 <var-label>, <delay>, <bit>
* %assign/v0/d <var-label>, <delayx>, <bit>

The %assign/v0 instruction is a vector version of non-blocking
assignment. The <delay> is the number of clock ticks in the future
where the assignment should be schedule, and the <bit> is the base of
the vector to be assigned to the destination. The vector width is in
index register 0.

The %assign/v0/d variation puts the delay instead into an integer
register that is given by the <delayx> value. This should not be 0, of
course, because integer 0 is taken with the vector width.

The <var-label> references a .var object that can receive non-blocking
assignments. For blocking assignments, see %set/v.

* %assign/v0/x1 <var-label>, <delay>, <bit>
* %assign/v0/x1/d <var-label>, <delayx>, <bit>

This is similar to the %assign/v0 instruction, but adds the index-1
index register with the canonical index of the destination where the
vector is to be written. This allows for part writes into the vector.

* %assign/wr <vpi-label>, <delay>, <index>

This instruction causes a non-blocking assign of the indexed value to
the real object addressed by the <vpi-label> label.

* %assign/x0 <var-label>, <delay>, <bit> (OBSOLETE -- See %assign/v0x)

This does a non-blocking assignment to a functor, similar to the
%assign instruction. The <var-label> identifies the base functor of
the affected variable, and the <delay> gives the delay when the
assignment takes place. The delay may be 0. The actual functor used is
calculated by using <var-label> as a base, and indexing with the
index[0] index register. This supports indexed assignment.

The <bit> is the address of the thread register that contains the bit
value to assign.


* %blend <bit-l>, <bit-r>, <wid>

This instruction blends the bits of a vector into the destination in a
manner like the expression (x ? <a> : <b>). The truth table is:

	1  1 --> 1
	0  0 --> 0
	z  z --> z
	x  x --> x
	.... --> x

In other words, if the bits are identical, then take that
value. Otherwise, the value is x.

* %breakpoint

This instruction unconditionally breaks the simulator into the
interactive debugger. The idea is to stop the simulator here and give
the user a chance to display the state of the simulation using
debugger commands.

This may not work on all platforms. If run-time debugging is compiled
out, then this function is a no-op.

* %cassign/v <var-label>, <bit>, <wid>

Perform a continuous assign of a constant value to the target
variable. This is similar to %set, but it uses the cassign port
(port-1) of the signal functor instead of the normal assign, so the
signal responds differently. See "VARIABLE STATEMENTS" in the
README.txt file.


* %cmp/u <bit-l>, <bit-r>, <wid>
* %cmp/s <bit-l>, <bit-r>, <wid>

These instructions perform a generic comparison of two vectors of equal
size. The <bit-l> and <bit-r> numbers address the least-significant
bit of each vector, and <wid> is the width. If either operand is 0,
1, 2 or 3 then it is taken to be a constant replicated to the selected
width.

The results of the comparison go into bits 4, 5, 6 and 7:

	4: eq  (equal)
	5: lt  (less than)
	6: eeq (case equal)

The eeq bit is set to 1 if all the bits in the vectors are exactly the
same, or 0 otherwise. The eq bit is true if the values are logically
the same. That is, x and z are considered equal. In other words the eq
bit is the same as ``=='' and the eeq bit ``===''.

The lt bit is 1 if the left vector is less then the right vector, or 0
if greater then or equal to the right vector. It is the equivalent of
the Verilog < operator. Combinations of these three bits can be used
to implement all the Verilog comparison operators.

The %cmp/u and %cmp/s differ only in the handling of the lt bit. The
%cmp/u does an unsigned compare, whereas the %cmp/s does a signed
compare. In either case, if either operand contains x or z, then lt
bit gets the x value.

* %cmpi/s <bit-l>, <immr>, <wid>
* %cmpi/u <bit-l>, <immr>, <wid>

These instructions are similar to the %cmp instructions above, except
that the right hand operand is an immediate value. This is a positive
number that the vector is compared with.

* %cmp/wr <bit-l>, <bit-r>

[compare real values.]

* %cmp/ws <bit-l>, <bit-r>
* %cmp/wu <bit-l>, <bit-r>

[compare signed/unsigned integer words.]

* %cmp/z <bit-l>, <bit-r>, <wid>
* %cmp/x <bit-l>, <bit-r>, <wid>

These instructions are for implementing the casez and casex
comparisons. These work similar to the %cmp/u instructions, except
only an eq bit is calculated. These comparisons both treat z values in
the left or right operand as don't care positions. The %cmp/x
instruction will also treat x values in either operand as don't care.

Only bit 4 is set by these instructions.


* %cvt/ir <bit-l>, <bit-r>
* %cvt/ri <bit-l>, <bit-r>
* %cvt/vr <bit-l>, <bit-r>, <wid>

Copy a word from r to l, converting it from real to integer (ir) or
integer to real (ri) in the process. The source and destination may
be the same word address, leading to a convert in place.

The %cvt/vr opcode converts a real word <bit-r> to a thread vector
starting at <bit-l> and with the width <wid>. Non-integer precision is
lost in the conversion.

* %deassign <var-label>

Deactivate and disconnect a procedural continuous assignment to a
variable. The <var-label> identifies the affected variable.

* %delay <delay>

This opcode pauses the thread, and causes it to be rescheduled for a
time in the future. The <amount> is the number of the ticks in the
future to reschedule, and is >= 0. If the %delay is zero, then the
thread yields the processor for another thread, but will be resumed in
the current time step.

* %delayx <idx>

This is similar to the %delay opcode, except that the parameter
selects an index register, which contains the actual delay. This
supports run-time calculated delays.

* %disable <scope-label>

This instruction terminates threads that are part of a specific
scope. The label identifies the scope in question, and the threads are
the threads that are currently within that scope.


* %div <bit-l>, <bit-r>, <wid>
* %div/s <bit-l>, <bit-r>, <wid>

This instruction arithmetically divides the <bit-l> vector by the
<bit-r> vector, and leaves the result in the <bit-l> vector. IF any of
the bits in either vector are x or z, the entire result is x.

The %div/s instruction is the same as %div, but does signed division.


* %div/wr <bit-l>, <bit-r>

This opcode divides the left operand by the right operand. If the
right operand is 0, then the result is NaN.


* %force/v <label>, <bit>, <wid>

Force a constant value to the target variable. This is similar to %set
and %cassign/v, but it uses the force port (port-2) of the signal
functor instead of the normal assign port (port-0), so the signal
responds differently. See "VARIABLE STATEMENTS" and "NET STATEMENTS"
in the README.txt file.

* %force/x0 <label>, <bit>, <wid>

Force a constant value to part target variable. This is similar to
%set/x instruction, but it uses the force port (port-2) of the signal
functor instead of the normal assign port (port-0), so the signal
responds differently. See "VARIABLE STATEMENTS" and "NET STATEMENTS"
in the README.txt file.

* %fork <code-label>, <scope-label>

This instruction is similar to %jmp, except that it creates a new
thread to start executing at the specified address. The new thread is
created and pushed onto the child stack.  It is also marked runnable,
but is not necessarily started until the current thread yields.

The %fork instruction has no effect other than to push a child thread.

See also %join.


* %inv <bit>, <wid>

Perform a bitwise invert of the vector starting at <bit>. The result
replaces the input. Invert means the following, independently for each
bit:

	0  --> 1
	1  --> 0
	x  --> x
	z  --> x


* %ix/get <idx>, <bit>, <wid>
* %ix/get/s <idx>, <bit>, <wid>

This instruction loads a thread vector starting at <bit>, size <wid>,
into the index register <idx>. The <bit> is the lsb of the value in
thread bit space, and <wid> is the width of the vector.

The function converts the 4-value bits into a binary number, without
sign extension. If any of the bits of the vector is x or z, then the
index register gets the value 0. The %ix/get/s is the same, except
that it assumes the source vector is sign extended to fit the index
register.

The function also writes into bit 4 a 1 if any of the bits of the
input vector are x or z. This is a flag that the 0 value written into
the index register is really the result of calculating from unknown
bits.

	4: unknown value
	5: (reserved)
	6: (reserved)

* %ix/getv <functor-label>, <bit>

This instruction is like the %ix/get instruction, except that is reads
directly from a functor label instead of from thread bits. It sets
bits 4/5/6 just line %ix/get.

* %ix/load <idx>, <value>

This instruction loads an immediate value into the addressed index
register. The index register holds numeric values, so the <value> is a
number. The idx value selects the index register, and may be 0, 1, 2
or 3. This is different from %ix/get, which loads the index register
from a value in the thread bit vector.


* %ix/add <idx>, <value>
* %ix/sub <idx>, <value>
* %ix/mul <idx>, <value>

This instruction adds, subtracts, or multiplies an immediate value to
the addressed index register. The index register holds numeric values,
so the <value> is a number. The <idx> value selects the index register,
and may be 0, 1, 2 or 3.


* %jmp <code-label>

The %jmp instruction performs an unconditional branch to a given
location. The parameter is the label of the destination instruction.

* %jmp/[01xz] <code-label>, <bit>

This is a conditional version of the %jmp instruction. In this case,
a single bit (addressed by <bit>) is tested. If it is one of the
values in the part after the /, the jump is taken. For example:

	%jmp/xz T_label, 8;

will jump to T_label if bit 8 is x or z.

* %join

This is the partner to %fork. This instruction causes the thread to
wait for the top thread in the child stack to terminate, then
continues. It has no effect in the current thread other than to wait
until the top child is cleared.

It is an error to execute %join if there are no children in the child
stack. Every %join in the thread must have a matching %fork that
spawned off a child thread.

If the matching child instruction is still running, a %join suspends
the calling thread until the child ends. If the child is already
ended, then the %join does not block or yield the thread.

* %load/av <bit>, <array-label>, <wid>

This instruction loads a word from the specified array. The word
address is in index register 3. The width should match the width of
the array word.

* %load/avx.p <bit>, <array-label>, <index>

This instruction is similar ro %load/av, but it loads only a single
bit, and the <index> is the selector for the bit to use. If <index> is
out of range, then x is loaded.

* %load/mv <bit>, <memory-label>, <wid>

this instruction loads a word from the specified memory. The word
address is in index register 3. The width should match the width of
the memory word.

* %load/nx <bit>, <vpi-label>, <idx>

This instruction load a value from a .net object bit. Since .net
objects don't really exist (they are only named indirection into the
netlist) this instruction indexes into the .net list of bits.

* %load/v <bit>, <functor-label>, <wid>

This instruction loads a vector value from the given functor node into
the specified thread register bit. The functor-label can refer to a
.net, a .var or a .functor with a vector output. The entire vector,
from the least significant up to <wid> bits, is loaded starting at
thread bit <bit>. It is an OK for the width to not match the vector
width at the functor. If the <wid> is less then the width at the
functor, the the most significant bits are dropped.

* %load/vp0 <bit>, <functor-label>, <wid>

This instruction is the same as %load/v above, except that it also
adds the integer value is index register 0 into the loaded value. The
addition is a verilog-style add, which means that if any of the input
bits are X or Z, the entire result is turned into a vector of X bits.

* %load/wr <bit>, <vpi-label>

This instruction reads a real value from the vpi-like object to a word
register.

* %load/x <bit>, <functor-label>, <idx>
* %load/x.p <bit>, <functor-label>, <idx>

This is an indexed load. It uses the contents of the specified index
register to select a bit from a vector functor at <functor-label>. The
bit is pulled from the indexed bit of the addressed functor and loaded
into the destination thread bit. If the indexed value is beyond the
width of the vector, then the result is X.

The %load/x.p is the same, but when the operation is done, it
increments the specified index register. This provides a basic
auto-increment feature.

* %loadi/wr <bit>, <mant>, <exp>

This opcode loads an immediate value, floating point, into the word
register selected by <bit>. The mantissa is an unsigned integer value,
up to 32 bits, that multiplied by 2**(<exp>-0x1000) to make a real
value. The sign bit is OR-ed into the <exp> value at bit 0x4000, and
is removed from the <exp> before calculating the real value.

If <exp>==0x3fff and <mant> == 0, the value is +inf.
If <exp>==0x7fff and <mant> == 0, the value is -inf.
If <exp>--0x3fff and <mant> != 0, the value is NaN.

* %mod   <bit-l>, <bit-r>, <wid>
* %mod/s <bit-l>, <bit-r>, <wid>

This instruction calculates the modulus %r of the left operand, and
replaces the left operand with the result. The <wid> gives the width
of the left and the right vectors, and the left vector is completely
replaced with the result.

The /s form does signed %.

* %mod/wr <bit-l>, <bit-r>

This opcode is the real-valued modulus of the two real values.

* %mov <dst>, <src>, <wid>
* %mov/wr <dst>, <src>
* %movi <dst>, <value>, <wid>

This instruction copies a vector from one place in register space to
another. The destination and source vectors are assumed to be the same
width and non-overlapping. The <dst> may not be 0-3, but if the <src>
is one of the 4 constant bits, the effect is to replicate the value
into the destination vector. This is useful for filling a vector.

The %movi variant moves a binary value, LSB first, into the
destination vector.

* %mul <bit-l>, <bit-r>, <wid>

This instruction multiplies the left vector by the right vector, the
vectors having the width <wid>. If any of the bits of either vector
are x or z, the result is x. Otherwise, the result is the arithmetic
product.


* %mul/wr <bit-l>, <bit-r>

This opcode multiplies two real words together. The result replaces
the left operand.


* %muli <bit-l>, <imm>, <wid>

This instruction is the same as %mul, but the second operand is an
immediate value that is padded to the width of the result.


* %nand <dst>, <src>, <wid>

Perform the bitwise NAND of the two vectors, and store the result in
the left vector. Each bit is calculated independent of other bits. NAND
means the following:

	0 and ? --> 1
	? and 0 --> 1
	1 and 1 --> 0
	otherwise   x


* %nor <dst>, <src>, <wid>

Perform the bitwise nor of the vectors. Each bit in the <dst> is
combined with the corresponding bit in the source, according to the
truth table:

	1 nor ? --> 0
	? nor 1 --> 0
	0 nor 0 --> 1
	otherwise  x


* %nor/r <dst>, <src>, <wid>

The %nor/r instruction is a reduction nor. That is, the <src> is a
vector with width, but the result is a single bit. The <src> vector is
not affected by the operation unless the <dst> bit is within the
vector. The result is calculated before the <dst> bit is written, so
it is valid to place the <dst> within the <src>.

The actual operation performed is the inverted or of all the bits in
the vector.


* %or <dst>, <src>, <wid>

Perform the bitwise or of the vectors. Each bit in the <dst> is
combined with the corresponding bit in the source, according to the
truth table:

	1 or ? --> 1
	? or 1 --> 1
	0 or 0 --> 0
	otherwise  x


* %or/r <dst>, <src>, <wid>

This is a reduction version of the %or opcode. The <src> is a vector,
and the <dst> is a writable scalar. The <dst> gets the value of the
or of all the bits of the src vector.


* %release/net <functor-label>
* %release/reg <functor-label>

Release the force on the signal that is represented by the functor
<functor-label>.  The force was previously activated with a %force/v
statement.  If no force was active on this functor the statement does
nothing. The %release/net sends to the labeled functor the release
command with net semantics: the unforced value is propagated to the
output of the signal after the release is complete. The %release/reg
sends the release command with reg semantics: the signal holds its
forced value until another value propagates through.

* %set/v <var-label>, <bit>, <wid>

This sets a vector to a variable, and is used to implement blocking
assignments. The <var-label> identifies the variable to receive the
new value. Once the set completes, the value is immediately available
to be read out of the variable. The <bit> is the address of the thread
register that contains the LSB of the vector, and the <wid> is the
size of the vector. The width must exactly match the width of the
signal.

* %set/av <array-label>, <bit>, <wid>

This sets a thread vector to an array word. The <array-label>
addresses an array device, and the <bit>,<wid> describe a vector to be
written. Index register 3 contains the address of the word within the
array.

The base of a part select is retrieved from index register 1. The
width is implied from the <wid> that is the argument. This is the part
*within* the word.

The address (in canonical form) is precalculated and loaded into index
register 3. This is the address of the word within the array.

* %set/mv <memory-label>, <bit>, <wid>

This sets a thread vector to a memory word. The <memory-label>
addresses a memory device, and the <bit>,<wid> describe a vector to be
written. Index register 3 contains the address of the word within the
memory.

The base of a part select is retrieved from index register 1.

The address (in canonical form) is precalculated and loaded into index
register 3.


* %set/wr <vpi-label>, <bit>

This instruction writes a real word to the specified VPI-like object.

* %set/x0 <var-label>, <bit>, <wid>

This sets the part of a signal vector, the address calculated by
using the index register 0 to index the base within the vector of
<var-label>. The destination must be a signal of some sort. Otherwise,
the instruction will fail.

The addressing is canonical (0-based) so the compiler must figure out
non-zero offsets, if any. The width is the width of the part being
written. The other bits of the vector are not touched.

The index may be signed, and if less then 0, the beginning bits are
not assigned. Also, if the bits go beyond the end of the signal, those
bits are not written anywhere.


* %shiftl/i0 <bit>, <wid>

This instruction shifts the vector left (towards more significant
bits) by the amount in index register 0. The <bit> is the address of
the lsb of the vector, and <wid> the width of the vector. The shift is
done in place. Zero values are shifted in.

* %shiftr/i0 <bit>, <wid>
* %shiftr/s/i0 <bit>, <wid>

This instruction shifts the vector right (towards the less significant
bits) by the amount in the index register 0. The <bit> is the address
of the lsb of the vector, and <wid> is the width of the vector. The
shift is done in place.

%shiftr/i0 is an unsigned down shift, so zeros are shifted into the
top bits. %shiftr/s/i0 is a signed shift, so the value is sign-extended.

* %sub <bit-l>, <bit-r>, <wid>

This instruction arithmetically subtracts the right vector out of the
left vector. It accomplishes this by adding to the left vector 1 plus
the 1s complement of the right vector. The carry value is dropped, and
the result, placed in <bit-l>, is the subtraction of <bit-r> from the
input <bit-l>. Both vectors have the same width. If any bits in either
operand are x, then the entire result is x.

See also the %add instruction.

* %subi <bit-l>, <imm>, <wid>

This instruction arithmetically subtracts the immediate value from the
left vector. The <imm> value is a 16bit unsigned value zero-extended to
the <wid> of the left vector. The result replaces the left vector.

See also the %addi instruction.


* %sub/wr <bit-l>, <bit-r>

This instruction operates on real values in word registers. The right
indexed value is subtracted from the left indexed value, and the
result placed in the left index.


* %vpi_call <name> [, ...]

This instruction makes a call to a system task that was declared using
VPI. The operands are compiled down to a vpiHandle for the call. The
instruction contains only the vpiHandle for the call. See the vpi.txt
file for more on system task/function calls.


* %vpi_func <name>, <dst>, <wid> [, ...]

This instruction is similar to %vpi_call, except that it is for
calling system functions. The difference here is the <dst> and <wid>
parameters that specify where the return value is to go. The normal
means that the VPI code uses to write the return value causes those
bits to go here.


* %wait <functor-label>

When a thread executes this instruction, it places itself in the
sensitive list for the addressed functor. The functor holds all the
threads that await the functor. When the defined sort of event occurs
on the functor, a thread schedule event is created for all the threads
in its list and the list is cleared.

* %xnor <dst>, <src>, <wid>

This does a bitwise exclusive nor (~^) of the <src> and <dst> vector,
and leaves the result in the <dst> vector. xnor is this:

	0 xnor 0 --> 1
	0 xnor 1 --> 0
	1 xnor 0 --> 0
	1 xnor 1 --> 1
	otherwise    x


* %xor <dst>, <src>, <wid>

This does a bitwise exclusive or (^) of the <src> and <dst> vector,
and leaves the result in the <dst> vector. xor is this:

	0 xnor 0 --> 0
	0 xnor 1 --> 1
	1 xnor 0 --> 1
	1 xnor 1 --> 0
	otherwise    x


/*
 * Copyright (c) 2001-2003 Stephen Williams (steve@icarus.com)
 *
 *    This source code is free software; you can redistribute it
 *    and/or modify it in source code form under the terms of the GNU
 *    General Public License as published by the Free Software
 *    Foundation; either version 2 of the License, or (at your option)
 *    any later version.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    GNU General Public License for more details.
 *
 *    You should have received a copy of the GNU General Public License
 *    along with this program; if not, write to the Free Software
 *    Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
 */