Skip to content

DSLX Built-In Functions and Standard Library

This page documents the DSLX built-in functions (i.e. functions that don't require an import and are available in the top level namespace) and standard library modules (i.e. which are imported with an unqualified name like import std).

Generally built-in functions have some capabilities that cannot be easily done in a user-defined function, and thus do not live in the standard library.

Built-ins

This section describes the built-in functions provided for use in the DSL that do not need to be explicitly imported.

Some of the built-ins have exclamation marks at the end; e.g. assert! -- exclamation marks in DSLX indicate built-in functions that have special facilities that normal functions are not capable of. This is intended to help the user discern when they're calling "something that will behave like a normal function" versus "something that has special abilities beyond normal functions, e.g. special side-effects".

Note

A brief note on "Parallel Primitives": the DSL is expected to grow additional support for use of high-level parallel primitives over time, adding operators for order-insensitive reductions, scans, groupings, and similar. By making these operations known to the compiler in their high level form, we potentially enable optimizations and analyses on their higher level ("lifted") form. As of now, map is the sole parallel-primitive-oriented built-in.

array_size

array_size returns the number of elements in a given array-typed argument.

fn array_size<T: type, N: u32>(x: T[N]) -> u32

#[test]
fn test_array_size() {
    assert_eq(array_size(u32[3]:[0, 1, 2]), u32:3);
    assert_eq(array_size(u8[1]:[42]), u32:1);
}

widening_cast, checked_cast

widening_cast and checked_cast cast bits-type values to bits-type values with additional checks compared to casting with as.

widening_cast will report a static error if the type casted to is unable to respresent all values of the type casted from (ex. widening_cast<u5>(s3:0) will fail because s3:-1 cannot be respresented as an unsigned number).

checked_cast will cause a runtime error during dslx interpretation if the value being casted is unable to fit within the type casted to (ex. checked_cast<u5>(s3:0) will succeed while checked_cast<u5>(s3:-1) will cause the dslx interpreter to fail.

Currently both widening_cast and checked_cast will lower into a normal IR cast and will not generate additional assertions at the IR or Verilog level.

fn widening_cast<sN[N]>(value: uN[M]) -> sN[N]; where N > M
fn widening_cast<sN[N]>(value: sN[M]) -> sN[N]; where N >= M
fn widening_cast<uN[N]>(value: uN[M]) -> uN[N]; where N >= M

fn checked_cast<xN[M]>(value: uN[N]) -> xN[M]
fn checked_cast<xN[M]>(value: sN[N]) -> xN[M]

smulp and umulp

smulp and umulp perform signed and unsigned partial multiplications. These operations return a two-element tuple with the property that the sum of the two elements is equal to the product of the original inputs. Performing a partial multiplication allows for a pipeline stage in the middle of a multiply. These operations have the following signatures:

fn smulp<N>(lhs: sN[N], rhs: sN[N]) -> (sN[N], sN[N])
fn umulp<N>(lhs: uN[N], rhs: uN[N]) -> (uN[N], uN[N])

map

map, similarly to other languages, executes a transformation function on all the elements of an original array to produce the resulting "mapped' array. For example: taking the absolute value of each element in an input array:

import std;

fn main(x: s3[3]) -> s3[3] {
  let y: s3[3] = map(x, std::abs);
  y
}

#[test]
fn main_test() {
  let got: s3[3] = main(s3[3]:[-1, 1, 0]);
  assert_eq(s3[3]:[1, 1, 0], got)
}

Note that map is special, in that we can pass it a callee as if it were a value. As a function that "takes" a function as an argument, map is a special built-in -- in language implementor parlance it is a higher order function.

Implementation note: Functions are not first class values in the DSL, so the name of the function must be referred to directly.

Note

Novel higher order functions (e.g. if a user wanted to write their own map) cannot currently be written in user-level DSL code.

zip

zip places elements of two same-sized arrays together in an array of 2-tuples.

Its signature is: zip(lhs: T[N], rhs: U[N]) -> (T, U)[N].

#[test]
fn test_zip_array_size_1() {
    const LHS = u8[1]:[42];
    const RHS = u16[1]:[64];
    const WANT = [(u8:42, u16:64)];
    assert_eq(zip(LHS, RHS), WANT);
}

#[test]
fn test_zip_array_size_2() {
    assert_eq(zip(u32[2]:[1, 2], u64[2]:[10, 11]), [(u32:1, u64:10), (u32:2, u64:11)]);
}

array_rev

array_rev reverses the elements of an array.

#[test]
fn test_array_rev() {
  assert_eq(array_rev(u8[1]:[42]), u8[1]:[42]);
  assert_eq(array_rev(u3[2]:[1, 2]), u3[2]:[2, 1]);
  assert_eq(array_rev(u3[3]:[2, 3, 4]), u3[3]:[4, 3, 2]);
  assert_eq(array_rev(u4[3]:[0xf, 0, 0]), u4[3]:[0, 0, 0xf]);
  assert_eq(array_rev(u3[0]:[]), u3[0]:[]);
}

const_assert!

Performs a check on a constant expression that only fails at compile-time (not at runtime).

Note that this can only be applied to compile-time-constant expressions.

#[test]
fn test_const_assert() {
  const N = u32:42;
  const_assert!(N >= u32:1<<5);
}

Keep in mind that all const_assert!s in a function evaluate, similar to static_assert in C++ -- they are effectively part of the type system, so you cannot suppress them by putting them in a conditional:

fn f() {
  if false {
    const_assert!(false);  // <-- still fails even inside the "if false"
  }
}

clz, ctz

DSLX provides the common "count leading zeroes" and "count trailing zeroes" functions:

  let x0 = u32:0x0FFFFFF8;
  let x1 = clz(x0);
  let x2 = ctz(x0);
  assert_eq(u32:4, x1);
  assert_eq(u32:3, x2)

decode

Converts a binary-encoded value into a one-hot value. For an operand value of n``interpreted as an unsigned number, then-th result bit and only then`-th result bit is set. Has the following signature:

fn decode<uN[W]>(x: uN[N]) -> uN[W]

The width of the decode operation may be less than the maximum value expressible by the input (2**N - 1). If the encoded operand value is larger than the number of bits of the result, the result is zero.

Example usage: dslx/tests/decode.x.

See also the IR semantics for the decode op.

encode

Converts a one-hot value to a binary-encoded value of the "hot" bit of the input. If the n-th bit and only the n-th bit of the operand is set, the result is equal to the value n as an unsigned number. Has the following signature:

fn encode(x: uN[N]) -> uN[ceil(log2(N))]

If multiple bits of the input are set, the result is equal to the logical or of the results produced by the input bits individually. For example, if bit 3 and bit 5 of an encode input are set the result is equal to 3 | 5 = 7.

Example usage: dslx/tests/encode.x.

See also the IR semantics for the encode op.

enumerate

Decorates elements of an array with indices. Has the following signature:

fn enumerate<T: type, N: u32>(x: T[N]) -> (u32, T)[N]

one_hot

Converts a value to one-hot form. Has the following signature:

fn one_hot<N: u32, NP1:u32={N+1}>(x: uN[N], lsb_is_prio: bool) -> uN[NP1]

When lsb_is_prio is true, the least significant bit that is set becomes the one-hot bit in the result. When it is false, the most significant bit that is set becomes the one-hot bit in the result.

When all bits in the input are unset, the additional bit present in the output value (MSb) becomes set.

Example usage: dslx/tests/one_hot.x.

See also the IR semantics for the one_hot op.

one_hot_sel

Produces the result of 'or'-ing all case values for which the corresponding bit of the selector is enabled. In cases where the selector has exactly one bit set (it is in one-hot form) this is equivalent to a match.

fn one_hot_sel(selector: uN[N], cases: xN[N][M]) -> uN[N]

Evaluates each case value and ors each case together if the corresponding bit in the selector is set. The first element of cases is included if the LSB is set, the second if the next least significant bit and so on. If no selector bits are set this evaluates to zero. This function is not generally used directly though the compiler will when possible synthesize the equivalent code from a match expression.

Note

This is included largely for testing purposes and for bespoke 'intrinsic-style programming' use cases.

priority_sel

Implements a priority selector. Has the following signature:

fn priority_sel(selector: uN[N], cases: xN[M][N], default_value: xN[M]) -> xN[M]

That is, the selector is N bits, and we give N cases to choose from of M-bit values of arbitrary signedness.

If the selector is zero, the default value is returned.

Example usage: dslx/tests/priority_sel.x.

range

Returns an array filled with the sequence from inclusive START to exclusive LIMIT. Note that START and LIMIT must be compile-time-constant values (AKA "constexpr"). It is compile-time-checked by the type system that LIMIT >= START.

fn range(START: const uN, LIMIT: const uN) -> uN[{LIMIT-START}]

signex

Casting has well-defined extension rules, but in some cases it is necessary to be explicit about sign-extensions, if just for code readability. For this, there is the signex built-in.

To invoke the signex built-in, provide it with the operand to sign extend (lhs), as well as the target type to extend to: these operands may be either signed or unsigned. Note that the value of the right hand side is ignored, only its type is used to determine the result type of the sign extension.

#[test]
fn test_signex() {
  let x = u8:0xff;
  let s: s32 = signex(x, s32:0);
  let u: u32 = signex(x, u32:0);
  assert_eq(s as u32, u)
}

Note that both s and u contain the same bits in the above example.

slice

Array-slice built-in operation. Note that the "want" argument is not used as a value, but is just used to reflect the desired slice type. (Prior to constexprs being passed to built-in functions, this was the canonical way to reflect a constexpr in the type system.) Has the following signature:

fn slice<T: type, N, M, S>(xs: T[N], start: uN[M], want: T[S]) -> T[S]

rev

rev is used to reverse the bits in an unsigned bits value. The LSb in the input becomes the MSb in the result, the 2nd LSb becomes the 2nd MSb in the result, and so on.

// (Dummy) wrapper around reverse.
fn wrapper<N: u32>(x: bits[N]) -> bits[N] {
  rev(x)
}

// Target for IR conversion that works on u3s.
fn main(x: u3) -> u3 {
  wrapper(x)
}

// Reverse examples.
#[test]
fn test_reverse() {
  assert_eq(u3:0b100, main(u3:0b001));
  assert_eq(u3:0b001, main(u3:0b100));
  assert_eq(bits[0]:0, rev(bits[0]:0));
  assert_eq(u1:1, rev(u1:1));
  assert_eq(u2:0b10, rev(u2:0b01));
  assert_eq(u2:0b00, rev(u2:0b00));
}

bit_slice_update

bit_slice_update(subject, start, value) returns a copy of the bits-typed value subject where the contiguous bits starting at index start (where 0 is the least-significant bit) are replaced with value. The bit-width of the returned value is the same as the bit-width of subject. Any updated bit indices which are out of bounds (if start + bit-width(value) >= bit-width(subject)) are ignored.

Example usage: dslx/tests/bit_slice_update.x.

bit-wise reductions: and_reduce, or_reduce, xor_reduce

These are unary reduction operations applied to a bits-typed value:

  • and_reduce: evaluates to bool:1 if all bits of the input are set, and 0 otherwise.
  • or_reduce: evaluates to bool:1 if any bit of the input is set, and 0 otherwise.
  • xor_reduce: evaluates to bool:1 if there is an odd number of bits set in the input, and 0 otherwise.

These functions return the identity element of the respective operation for trivial (0 bit wide) inputs:

#[test]
fn test_trivial_reduce() {
  assert_eq(and_reduce(bits[0]:0), true);
  assert_eq(or_reduce(bits[0]:0), false);
  assert_eq(xor_reduce(bits[0]:0), false);
}

update

update(array, index, new_value) returns a copy of array where array[index] has been replaced with new_value, and all other elements are unchanged. index can either be a scalar index or a tuple when updating nested element(s) of multi-dimensional array. Note that this is not an in-place update of the array, it is an "evolution" of array. It is the compiler's responsibility to optimize by using mutation instead of copying, when it's safe to do. The compiler makes a best effort to do this, but can't guarantee the optimization is always made.

#[test]
fn test_update() {
  assert_eq(update([u8:1, u8:2, u8:3], u2:2, u8:42), [u8:1, u8:2, u8:42]);
  assert_eq(update([[u8:1, u8:2], [u8:3, u8:4]],
                   (u1:1, u1:0), u8:42), [[u8:1, u8:2], [u8:42, u8:4]]);
}

assert_eq, assert_lt

In a unit test pseudo function all valid DSLX code is allowed. To evaluate test results DSLX provides the assert_eq primitive (we'll add more of those in the future). Here is an example of a divceil implementation with its corresponding tests:

fn divceil(x: u32, y: u32) -> u32 {
  (x-u32:1) / y + u32:1
}

#[test]
fn test_divceil() {
  assert_eq(u32:3, divceil(u32:5, u32:2));
  assert_eq(u32:2, divceil(u32:4, u32:2));
  assert_eq(u32:2, divceil(u32:3, u32:2));
  assert_eq(u32:1, divceil(u32:2, u32:2));
}

assert_eq cannot currently be synthesized into equivalent Verilog. Because of that it is recommended to use it within test constructs (interpretation) only.

zero!<T>

DSLX has a macro for easy creation of zero values, even from aggregate types. Invoke the macro with the type parameter as follows:

struct MyPoint {
  x: u32,
  y: u32,
}

enum MyEnum : u2 {
  ZERO = u2:0,
  ONE = u2:1,
}

#[test]
fn test_zero_macro() {
  assert_eq(zero!<u32>(), u32:0);
  assert_eq(zero!<MyPoint>(), MyPoint{x: u32:0, y: u32:0});
  assert_eq(zero!<MyEnum>(), MyEnum::ZERO);
}

The zero!<T> macro can also be used with the struct update syntax to initialize a subset of fields to zero. In the example below all fields except foo are initialized to zero in the struct returned by f.

struct MyStruct {
  foo: u1,
  bar: u2,
  baz: u3,
  bat: u4,
}

fn f() -> MyStruct {
  MyStruct{foo: u1:1, ..zero!<MyStruct>()}
}

all_ones!<T>

Similar to zero!<T>, DSLX has a macro for easy creation of all-ones values, even from aggregate types. Invoke the macro with the type parameter as follows:

struct MyPoint {
  x: u32,
  y: u32,
}

enum MyEnum : u2 {
  ZERO = u2:0,
  ONE = u2:1,
  THREE = u2:3,
}

#[test]
fn test_all_ones_macro() {
  assert_eq(all_ones!<u32>(), u32:0xFFFFFFFF);
  assert_eq(all_ones!<MyPoint>(), MyPoint{x: u32:0xFFFFFFFF, y: u32:0xFFFFFFFF});
  assert_eq(all_ones!<MyEnum>(), MyEnum::THREE);
}

The all_ones!<T> macro can also be used with the struct update syntax to initialize a subset of fields to zero. In the example below all fields except foo are initialized to zero in the struct returned by f.

struct MyStruct {
  foo: u1,
  bar: u2,
  baz: u3,
  bat: u4,
}

fn f() -> MyStruct {
  MyStruct{foo: u1:1, ..all_ones!<MyStruct>()}
}

trace_fmt!

DSLX supports printf-style debugging via the trace_fmt! builtin, which allows dumping values to stdout. For example:

// Note: to see `trace_fmt!` output you need to be seeing `INFO` level logging,
// enabled by adding the '--alsologtostderr' flag to the command line (among
// other means). For example:
// bazel run -c opt //xls/dslx:interpreter_main  /path/to/dslx/file.x -- --alsologtostderr

fn shifty(x: u8, y: u3) -> u8 {
  trace_fmt!("x: {:x} y: {}", x, y);
  // Note: y looks different as a negative number when the high bit is set.
  trace_fmt!("y as s8: {}", y as s2);
  x << y
}

#[test]
fn test_shifty() {
  assert_eq(shifty(u8:0x42, u3:4), u8:0x20);
  assert_eq(shifty(u8:0x42, u3:7), u8:0);
}

would produce the following output, with each trace being annotated with its corresponding source position:

[...]
[ RUN UNITTEST  ] test_shifty
I0607 10:11:02.897455  810431 shifty.x:6] x: 42 y: 4
I0607 10:11:02.897462  810431 shifty.x:8] y as s8: 0
I0607 10:11:02.908182  810431 shifty.x:6] x: 42 y: 7
I0607 10:11:02.908208  810431 shifty.x:8] y as s8: -1
[            OK ]
[...]

The following formats are available, given an s20 input of 0xbeef:

Format string Description Example
{} Default 48879
{:d} Signed decimal 48879
{:u} Unsigned decimal 48879
{:x} Plain hex beef
{:#x} Hex with 0x prefix 0xbeef
{:0x} Hex with leading zeros 0_beef
{:b} Plain binary 1011111011101111
{:#b} Binary with 0b prefix 0b1011_1110_1110_1111
{:0b} Binary with leading zeros 0000_1011_1110_1110_1111

The number of digits for zero-padding is determined by the actual type of the input, e.g., a u16 pads up to 16 binary digits or 4 hexadecimal digits.

Note

trace! currently also exists as a builtin but is in the process of being removed, as it provided the user with only a "global flag" way of specifying the desired format for output values -- trace_fmt! is more powerful.

fail! / assert!: assertion failure

The fail! builtin indicates a path that should not be reachable in practice. Its general signature is:

fail!(label: u8[N], fallback_value: T) -> T

The assert! builtin is similar to fail, but takes a predicate, and does not produce a value:

assert!(predicate: bool, label: u8[N]) -> ()

These can be thought of as "fatal assertions", and convert to Verilog/SytemVerilog assertions in generated code.

Note

XLS hopes to permit users to optionally insert fatal-error-signaling hardware that correspond to these operations. See https://github.com/google/xls/issues/1352

fail! indicates a control path that should not be reachable, assert! gives a predicate that should always be true when the statement is reached.

If triggered, these raise a fatal error in simulation (e.g. via a JIT-execution failure status or a Verilog assertion when running in RTL simulation).

Assuming fail! will not be triggered minimizes its cost in synthesized form. In this situation, when it is "erased", it acts as the identity function, providing the fallback_value. This allows XLS to keep well defined semantics even when fatal assertion hardware is not present.

Example assert!: if there is a function that should never be invoked with the value 42 as a precondition:

fn main(x: u32) -> u32 {
  assert!(x != u32:42, "x_never_forty_two");
  x - u32:42
}

Example fail!: if only these two enum values shown should be possible (say, as a documented precondition for main):

enum EnumType: u2 {
  FIRST = 0,
  SECOND = 1,
}

fn main(x: EnumType) -> u32 {
  match x {
    EnumType::FIRST => u32:0,
    EnumType::SECOND => u32:1,
    // This should not be reachable.
    // But, if we synthesize hardware, under this condition the function is
    // well-defined to give back zero.
    _ => fail!("unknown_EnumType", u32:0),
  }
}

The fail!("unknown_EnumType", u32:0) above indicates that a) that match arm should not be reached (and if it is in the JIT or RTL simulation it will cause an error status or assertion failure respectively), but b) provides a fallback value to use (of the appropriate type) in case it were to happen in synthesized gates which did not insert fatal-error-indicating hardware.

The associated label (e.g. the first argument to fail!) must be a valid Verilog identifier and is used for identifying the failure when lowered to SystemVerilog. At higher levels in the stack, it's unused.

cover!

Note

Currently, cover! has no effect in RTL simulators supported in XLS open source (i.e. iverilog). See google/xls#436.

The cover! builtin tracks how often some condition is satisfied. It desugars into SystemVerilog cover points. Its signature is:

cover!(<name>, <condition>);

Where name is a function-unique literal string identifying the coverpoint and condition is a boolean element. When condition is true, a counter with the given name is incremented that can be inspected upon program termination. Coverpoints can be used to give an indication of code "coverage", i.e. to see what paths of a design are exercised in practice. The name of the coverpoint must begin with either a letter or underscore, and its remainder must consist of letters, digits, underscores, or dollar signs.

gate!

The gate! built-in is used for operand gating, of the form:

let gated_value = gate!(<pass_value>, <value>);

This will generally use a special Verilog macro to avoid the underlying synthesis tool doing boolean optimization, and will turn gated_value to 0 when the predicate pass_value is false. This can be used in attempts to manually avoid toggles based on the gating predicate.

It is expected that XLS will grow facilities to inserting gating ops automatically, but manual user insertion is a practical step in this direction. Additionally, it is expected that if, in the resulting Verilog, gating occurs on a value that originates from a flip flop, the operand gating may be promoted to register-based load-enable gating.

join: sequencing I/O tokens

The join builtin "joins together" a variable number of tokens into one token -- this resulting token represents that an I/O operation happens "no earlier than" all the given tokens. That is, it establishes a >= event ordering with respect to all the parameter tokens.

join(token...) -> token

This is useful to a user that wants to sequence their I/O operations, e.g. ensuring that two earlier I/O operations happen before a later I/O operation can begin.

    let (token0, value0) = recv(chan0);
    let (token1, value1) = recv(chan1);
    let joined = join(token0, token1);
    let token2 = send(joined, chan2, another_value);

Note

this routine can only be used in the body of a proc.

send: send a value on a channel

The send builtin sends a value on a channel, taking an I/O sequencing token and producing a new I/O sequencing token (which can be used to order communication events).

send(tok: token, chan<T> out, value: T) -> token

send_if: conditionally send a value on a channel

send_if(tok: token, chan<T> out, predicate: bool, value: T) -> token

The send_if builtin does a send on a channel as described in [send][#send], but only attempts to do so if the given predicate is true.

recv: (blocking) receive of a value from a channel

The recv builtin does a "blocking" recv of a value from a channel -- it is blocking in the sense that the current activation of the proc cannot complete until the recv has been performed. The token that the recv produces can be used to force a sequencing of the recv with respect to other I/O operations.

recv(tok: token, c: chan<T> in) -> (token, T)

recv_if: conditional (blocking) receive of a value from a channel

The recv_if builtin does a blocking receive as described in [recv][#recv], but only attempts to do so if the given predicate is true.

recv_if(tok: token, c: chan<T> in, predicate: bool, default_value: T) -> (token, T)

recv_non_blocking: non-blocking receive of a value from a channel

Performs a non-blocking receive from channel c -- if the channel is empty the default_value is returned as the result, and the bool in the result indicates whether the value originated from the channel (i.e. true means the value came from the channel).

recv_non_blocking(tok: token, c: chan<T> in, default_value: T) -> (token, T, bool)

Note

non-blocking operations make a block latency sensitive and can no longer be described as pure "Kahn Process Networks", which means that the design's correctness is more sensitive to the chosen schedule, and thus design verification should occur on the scheduled design.

recv_if_non_blocking: conditional non-blocking receive of a value from a channel

As recv_non_blocking is above, but with an additional predicate that indicates whether we should attempt to do the nonblocking receive from the channel. If this predicate is false, the default value will be provided and the returned boolean will be false.

recv_if_non_blocking(tok: token, c: chan<T> in, predicate: bool, default_value: T) -> (token, T, bool)

import std: DSLX standard library routines

Bits Type Properties

std::?_min_value

pub fn unsigned_min_value<N: u32>() -> uN[N]
pub fn signed_min_value<N: u32>() -> sN[N]

Returns the minimum signed or unsigned value contained in N bits.

std::?_max_value

pub fn unsigned_max_value<N: u32>() -> uN[N];
pub fn signed_max_value<N: u32>() -> sN[N];

Returns the maximum signed or unsigned value contained in N bits.

std::sizeof

pub fn sizeof<S: bool, N: u32>(x : xN[S][N]) -> u32

Returns the number of bits (sizeof) of unsigned or signed bit value. The signature above is parameterized on the signedness of the input type.

Bit Manipulation Functions

std::to_signed & std::to_unsigned

pub fn to_signed<N: u32>(x: uN[N]) -> sN[N]
pub fn to_unsigned<N: u32>(x: sN[N]) -> uN[N] {

Convenience helper that converts an unsigned bits argument to its same-sized signed type, or visa-versa. This is morally equivalent to (but slightly more convenient than) a pattern like:

    x as sN[std::sizeof(x)]

std::lsb

pub fn lsb<N: u32>(x: uN[N]) -> u1

Extracts the LSb (Least Significant bit) from the value x and returns it.

std::msb

pub fn msb<N: u32>(x: uN[N]) -> u1

Extracts the MSb (Most Significant bit) from the value x and returns it.

std::convert_to_bits_msb0

pub fn convert_to_bits_msb0<N: u32>(x: bool[N]) -> uN[N]

Converts an array of N bools to a bits[N] value.

Note well: the boolean value at index 0 of the array becomes the most significant bit in the resulting bit value. Similarly, the last index of the array becomes the least significant bit in the resulting bit value.

import std;

#[test]
fn convert_to_bits_test() {
  assert_eq(u3:0b001, std::convert_to_bits_msb0(bool[3]:[false, false, true]));
  assert_eq(u3:0b100, std::convert_to_bits_msb0(bool[3]:[true, false, false]));
}

There's always a source of confusion in these orderings:

  • Mathematically we often indicate the least significant digit as "digit 0"
  • But, in a number as we write the digits from left-to-right on a piece of paper, if you made an array from the written characters, the digit at "array index 0" would be the most significant bit.

So, it's somewhat ambiguous whether "index 0" in the array would become the least significant bit or the most significant bit. This routine uses the "as it looks on paper" conversion; e.g. [true, false, false] becomes 0b100.

std::convert_to_bools_lsb0

pub fn fn convert_to_bools_lsb0<N:u32>(x: uN[N]) -> bool[N]

Convert a "word" of bits to a corresponding array of booleans.

Note well: The least significant bit of the word becomes index 0 in the array.

std::mask_bits

pub fn mask_bits<X: u32>() -> bits[X]

Returns a value with X bits set (of type bits[X]).

std::concat3

pub fn concat3<X: u32, Y: u32, Z: u32, R: u32 = X + Y + Z>(x: bits[X], y: bits[Y], z: bits[Z]) -> bits[R]

Concatenates 3 values of arbitrary bitwidths to a single value.

std::rotr, std::rotl

pub fn rotr<N: u32>(x: bits[N], y: bits[N]) -> bits[N]

Rotate x right by y bits.

pub fn rotl<N: u32>(x: bits[N], y: bits[N]) -> bits[N]

Rotate x left by y bits.

std::popcount

pub fn popcount<N: u32>(x: bits[N]) -> bits[N]

Counts the number of bits in x that are '1'.

std::extract_bits

pub fn extract_bits<FROM_INCLUSIVE: u32, TO_EXCLUSIVE: u32, FIXED_SHIFT: u32,
                    N: u32>(x: bits[N]) -> bits[std::max(u32:0, TO_EXCLUSIVE - FROM_INCLUSIVE)] {
    let x_extended = x as uN[std::max(std::sizeof(x) + FIXED_SHIFT, TO_EXCLUSIVE)];
    (x_extended << fixed_shift)[FROM_INCLUSIVE:TO_EXCLUSIVE]
}

Extracts a bit-slice from x shifted left by fixed_shift. This function behaves as-if x as resonably infinite precision so that the shift does not drop any bits and that the bit slice will be in-range.

If to_exclusive <= from_excsuive, the result will be a zero-bit bits[0].

std::vslice

pub fn vslice<MSB: u32, LSB: u32>(x: bits[IN]) -> bits[OUT]

Similar to extract_bits above, but corresponds directly to the "part-select" Verilog syntax. That is:

y <= x[3:0];

Corresponds to:

let y: u4 = vslice<u32:3, u32:0>(x);

This is useful when porting code literally as a way to avoid transcription errors. For new code the DSLX first-class slicing syntax (either range-slicing or width-slicing) is preferred.

Mathematical Functions

std::bounded_minus_1

pub fn bounded_minus_1<N: u32>(x: uN[N]) -> uN[N]

Returns the value of x - 1 with saturation at 0.

std::abs

pub fn abs<BITS: u32>(x: sN[BITS]) -> sN[BITS]

Returns the absolute value of x as a signed number.

std::is_pow2

pub fn is_pow2<N: u32>(x: uN[N]) -> bool

Returns true when x is a non-zero power-of-two.

std::?add

pub fn uadd<N: u32, M: u32, R: u32 = umax(N,M) + 1>(x: uN[N], y: uN[M]) -> uN[R]
pub fn sadd<N: u32, M: u32, R: u32 = umax(N,M) + 1>(x: sN[N], y: sN[M]) -> sN[R]

Returns sum of x (N bits) and y (M bits) as a umax(N,M)+1 bit value.

std::?mul

pub fn umul<N: u32, M: u32, R: u32 = N + M>(x: uN[N], y: uN[M]) -> uN[R]
pub fn smul<N: u32, M: u32, R: u32 = N + M>(x: sN[N], y: sN[M]) -> sN[R]

Returns product of x (N bits) and y (M bits) as an N+M bit value.

std::iterative_div

pub fn iterative_div<N: u32, DN: u32 = N * u32:2>(x: uN[N], y: uN[N]) -> uN[N]

Calculate x / y one bit at a time. This is an alternative to using the division operator '/' which may not synthesize nicely.

std::div_pow2

pub fn div_pow2<N: u32>(x: bits[N], y: bits[N]) -> bits[N]

Returns x / y where y must be a non-zero power-of-two.

std::mod_pow2

pub fn mod_pow2<N: u32>(x: bits[N], y: bits[N]) -> bits[N]

Returns x % y where y must be a non-zero power-of-two.

std::ceil_div

pub fn ceil_div<N: u32>(x: uN[N], y: uN[N]) -> uN[N]

Returns the ceiling of (x divided by y).

std::round_up_to_nearest

pub fn round_up_to_nearest(x: u32, y: u32) -> u32

Returns x rounded up to the nearest multiple of y.

std::round_up_to_nearest_pow2_?

Returns x rounded up to the nearest multiple of y, where y is a known positive power of 2. This functionality is the same as std::round_up_to_nearest but optimized when y is a power of 2.

std::?pow

pub fn upow<N: u32>(x: uN[N], n: uN[N]) -> uN[N]
pub fn spow<N: u32>(x: sN[N], n: uN[N]) -> sN[N]

Performs integer exponentiation as in Hacker's Delight, Section 11-3. Only non-negative exponents are allowed, hence the uN parameter for spow.

std::clog2

pub fn clog2<N: u32>(x: bits[N]) -> bits[N]

Returns ceiling(log2(x)), with one exception: When x = 0, this function differs from the true mathematical function: clog2(0) = 0 where as ceil(log2(0)) = -infinity

This function is frequently used to calculate the number of bits required to represent x possibilities. With this interpretation, it is sensible to define clog2(0) = 0.

Example: clog2(7) = 3.

std:flog2

pub fn flog2<N: u32>(x: bits[N]) -> bits[N]

Returns floor(log2(x)), with one exception:

When x=0, this function differs from the true mathematical function: flog2(0) = 0 where as floor(log2(0)) = -infinity

This function is frequently used to calculate the number of bits required to represent an unsigned integer n to define flog2(0) = 0, so that flog(n)+1 represents the number of bits needed to represent the n.

Example: flog2(7) = 2, flog2(8) = 3.

std::?max

pub fn smax<N: u32>(x: sN[N], y: sN[N]) -> sN[N]
pub fn umax<N: u32>(x: uN[N], y: uN[N]) -> uN[N]

Returns the maximum of two integers.

std::?min

pub fn smin<N: u32>(x: sN[N], y: sN[N]) -> sN[N]
pub fn umin<N: u32>(x: uN[N], y: uN[N]) -> uN[N]

Returns the minimum of two unsigned integers.

std::uadd_with_overflow

pub fn uadd_with_overflow<V: u32>(x: uN[N], y: uN[M]) -> (bool, uN[V])

Returns a 2-tuple indicating overflow (boolean) and a sum (x + y) as uN[V]. An overflow occurs if the result does not fit within a uN[V].

std::umul_with_overflow

pub fn umul_with_overflow<V: u32>(x: uN[N], y: uN[M]) -> (bool, uN[V])

Returns a 2-tuple indicating overflow (boolean) and a product (x * y) as uN[V]. An overflow occurs if the result does not fit within a uN[V].

Misc Functions

Signed comparison - std::{sge, sgt, sle, slt}

pub fn sge<N: u32>(x: uN[N], y: uN[N]) -> bool
pub fn sgt<N: u32>(x: uN[N], y: uN[N]) -> bool
pub fn sle<N: u32>(x: uN[N], y: uN[N]) -> bool
pub fn slt<N: u32>(x: uN[N], y: uN[N]) -> bool

Explicit signed comparison helpers for working with unsigned values, can be a bit more convenient and a bit more explicit intent than doing casting of left hand side and right hand side.

std::find_index

pub fn find_index<BITS: u32, ELEMS: u32>( array: uN[BITS][ELEMS], x: uN[BITS]) -> (bool, u32)

Returns (found, index) given an array and the element to find within the array.

Note that when found is false, the index is 0 -- 0 is provided instead of a value like -1 to prevent out-of-bounds accesses from occurring if the index is used in a match expression (which will eagerly evaluate all of its arms), to prevent it from creating an error at simulation time if the value is ultimately discarded from the unselected match arm.

std::distinct

pub fn distinct<COUNT: u32, N: u32, S: bool>(items: xN[S][N][COUNT], valid: bool[COUNT]) -> bool

Returns whether all the items are distinct (i.e. there are no duplicate items) after the valid mask is applied.

import std;

#[test]
fn test_distinct_with_invalid() {
    let items = u8[4]:[1, 2, 3, 1];
    let valid = bool[4]:[true, true, true, false];
    assert_eq(std::distinct(items, valid), true);
}

import acm_random

Port of ACM random number generator to DSLX.

DO NOT use acm_random.x for any application where security -- unpredictability of subsequent output and previous output -- is needed. ACMRandom is in NO WAY a cryptographically secure pseudorandom number generator, and using it where recipients of its output may wish to guess earlier/later output values would be very bad.

acm_random::rng_deterministic_seed

pub fn rng_deterministic_seed() -> u32

Returns a fixed seed for use in the random number generator.

acm_random::rng_new

pub fn rng_new(seed: u32) -> State

Create the state for a new random number generator using the given seed.

acm_random::rng_next

pub fn rng_next(s: State) -> (State, u32)

Returns a pseudo-random number in the range [1, 2^31-2].

Note that this is one number short on both ends of the full range of non-negative 32-bit integers, which range from 0 to 2^31-1.

acm_random::rng_next64

pub fn rng_next(s: State) -> (State, u64)

Returns a pseudo random number in the range [1, (2^31-2)^2].

Note that this does not cover all non-negative values of int64, which range from 0 to 2^63-1. The top two bits are ALWAYS ZERO.