elasticai.creator.ir2verilog

Package Contents

Classes

Ir2Verilog
Shape
PluginLoader	PluginLoader for Ir2Verilog passes.
PluginLoader	PluginLoader for Ir2Verilog passes.
Node	Extending ir.core.Node to an hdl specific node.
Edge
TemplateDirector	Director for building verilog templates.
VerilogTemplate

Functions

type_handler
type_handler_iterable

Data

factory
factory
Code

API

class elasticai.creator.ir2verilog.Ir2Verilog

Initialization

__call__(root: elasticai.creator.hdl_ir.DataGraph, registry: elasticai.creator.hdl_ir.Registry, default_root_name='root') → collections.abc.Iterable[elasticai.creator.ir2verilog.ir2verilog.Code]

register_static(name: str | None, fn: collections.abc.Callable[[], str]) → collections.abc.Callable[[], str]

register(name: str | None, fn: elasticai.creator.hdl_ir.TypeHandler) → elasticai.creator.hdl_ir.TypeHandler

override(name: str | None, fn: elasticai.creator.hdl_ir.TypeHandler) → elasticai.creator.hdl_ir.TypeHandler

elasticai.creator.ir2verilog.type_handler(name: str | None, fn: elasticai.creator.hdl_ir.NonIterableTypeHandler) → elasticai.creator.hdl_ir.NonIterableTypeHandler

elasticai.creator.ir2verilog.type_handler_iterable(name: str | None, fn: elasticai.creator.hdl_ir.TypeHandler) → elasticai.creator.hdl_ir.TypeHandler

class elasticai.creator.ir2verilog.Shape(*values: int)

Initialization

values are interpreted as one of the following:

• width

• depth, width

• depth, width, height

Usually width is kernel_size, depth is channels

classmethod from_tuple(values: elasticai.creator.hdl_ir.ShapeTuple | list[int]) → elasticai.creator.hdl_ir.Shape

to_tuple() → elasticai.creator.hdl_ir.ShapeTuple

to_list() → list[int]

__getitem__(item)

size() → int

ndim() → int

property depth: int

__eq__(other)

property width: int

property height: int

__repr__() → str

elasticai.creator.ir2verilog.factory

‘IrFactory(…)’

class elasticai.creator.ir2verilog.PluginLoader(lowering: elasticai.creator.ir2verilog.ir2verilog.Ir2Verilog)

Bases: elasticai.creator.plugin.PluginLoaderBase

PluginLoader for Ir2Verilog passes.

Initialization

filter_plugin_dicts(plugins: collections.abc.Iterable[dict[str, Any]]) → collections.abc.Iterable[dict[str, Any]]

load_symbol(symbol: elasticai.creator.ir2verilog.ir2verilog.PluginSymbol) → None

get_symbols(specs: collections.abc.Iterable[elasticai.creator.ir2verilog.ir2verilog.PluginSpec]) → collections.abc.Iterable[elasticai.creator.ir2verilog.ir2verilog.PluginSymbol]

class elasticai.creator.ir2verilog.PluginLoader(lowering: elasticai.creator.ir2verilog.ir2verilog.Ir2Verilog)

Bases: elasticai.creator.plugin.PluginLoaderBase

PluginLoader for Ir2Verilog passes.

Initialization

filter_plugin_dicts(plugins: collections.abc.Iterable[dict[str, Any]]) → collections.abc.Iterable[dict[str, Any]]

load_symbol(symbol: elasticai.creator.ir2verilog.ir2verilog.PluginSymbol) → None

get_symbols(specs: collections.abc.Iterable[elasticai.creator.ir2verilog.ir2verilog.PluginSpec]) → collections.abc.Iterable[elasticai.creator.ir2verilog.ir2verilog.PluginSymbol]

elasticai.creator.ir2verilog.factory

‘IrFactory(…)’

class elasticai.creator.ir2verilog.Node

Bases: elasticai.creator.ir.ir_v2.Node, typing.Protocol

Extending ir.core.Node to an hdl specific node.

This node contains all knowledge that we need to create and use an instance of an hdl component. However, this becomes a little bit complicated because vhdl differentiates between

Attributes:

implementation:: The name of the implementation, e.g., entity name in vhdl or module name for verilog, will be used to derive the architecture name. E.g., if the implementation is "adder", we will instantiate the entity work.adder(rtl). CAUTION: This behaviour is subject to change. Future versions might require the full entity name

property implementation: str

property input_shape: elasticai.creator.hdl_ir.Shape

property output_shape: elasticai.creator.hdl_ir.Shape

class elasticai.creator.ir2verilog.Edge

Bases: elasticai.creator.ir.ir_v2.Edge, typing.Protocol

property src_dst_indices: tuple[tuple[int | str, int | str], ...]

elasticai.creator.ir2verilog.Code: TypeAlias

None

class elasticai.creator.ir2verilog.TemplateDirector

Director for building verilog templates.

Most methods correspond to verilog language constructs. Building the final template can be expensive! Typically you only want to do this once. E.g.:

from string import Template
from elasticai.creator.ir2verilog import type_handler, Code, TemplateDirector, Implementation

class _ExplodingTemplate:
    def substitute(self, params):
        raise RuntimeError("Template not initialized!")

_template = None

@type_handler
def fir_filter(impl: Implementation) -> Code:
    global _template
    if _template is None:
        _template = (
            TemplateDirector()
            .define_scoped_switch("USE_EXT_WEIGHTS", False)
            .define_scoped_switch("USE_EXT_MAC", False)
            .parameter("BITWIDTH")
            .parameter("LENGTH")
            .localparam("FILT_COEFFS")
            .localparam("NUM_MULT_UNIT")
            .add_module_name()
            .set_prototype(res.read_text(package_path, "verilog/filter_fir_full.v"))
            .build()
        )
    return _template.substitute(impl.attributes)

Initialization

reset() → Self

set_prototype(prototype: str) → Self

parameter(name: str) → Self

localparam(name: str) → Self

replace_module_of_instance(module_name: str, new_name: str) → Self

replace_instance_name(module_name: str, new_name: str) → Self

define_scoped_switch(name: str, default: bool) → Self

Add a switch for a define that is scoped to the module name.

The switch will be prefixed with the value that users provide as module_name to the render call.

Parameters:

• name – String with name of the switch/define name

• default – Setting switch for defining output state (True=set, False=undefine)

add_module_name() → Self

build() → elasticai.creator.ir2verilog.templates.VerilogTemplate

class elasticai.creator.ir2verilog.VerilogTemplate(template: string.Template, defines: dict[str, bool], module_name: dict[str, str])

Initialization

substitute(params: Mapping[str, str | bool]) → str