Functions — github.com/NVlabs/GEM

CodeHub MCPcopy Create free account

hub / github.com/NVlabs/GEM / functions

Functions85 in github.com/NVlabs/GEM

⨍ Functions 85 ◇ Types & classes 29

↓ 65 callersMethoditer

src/bin/cuda_test.rs:126

↓ 12 callersMethodfor_each_input

Enumerate all related aigpin inputs for this endpoint group. The enumerated inputs may have duplicates.

↓ 12 callersMethodinto_iter

src/bin/cuda_test.rs:95

↓ 11 callersMethodget_endpoint_group

(&self, endpt_id: usize)

↓ 11 callersMethodis_empty

src/bin/cuda_test.rs:121

↓ 11 callersMethoditer

src/bin/flatten_test.rs:382

↓ 9 callersMethodadd_and_gate

(&mut self, a: usize, b: usize)

↓ 8 callersMethodtopo_traverse_generic

( &self, endpoints: Option<&Vec<usize>>, is_primary_input: Option<&IndexSet<usize>>,

↓ 7 callersMethodnum_endpoint_groups

↓ 6 callersFunctionbuild_staged_aigs

Given the level split points, return a list of split stages. For example, given [10, 20], will return a list like this: [(0, 10, stage0_10), (10, 20,

src/staging.rs:188

↓ 6 callersFunctiondfs_topo

(netlistdb: &NetlistDB, topo_vis: &mut Vec<bool>, topo_instack: &mut Vec<bool>, topo: &mut Vec<usize>, pinid:

src/bin/naive_sim.rs:291

↓ 6 callersMethodquery_permute_with_pin_iv

returns permutation id, invert bit, and setzero bit

src/flatten.rs:295

↓ 5 callersMethodadd_aigpin

(&mut self, driver: DriverType)

↓ 5 callersMethodplace_sram_duplicate

places a sram_duplicate bit.

src/flatten.rs:304

↓ 5 callersFunctionset_bit_in_u32

(v: &mut u32, pos: u32, bit: u8)

src/flatten.rs:219

↓ 4 callersMethodtrace_clock_pin

given a clock pin, trace back to clock root and return its enable signal (with invert bit). if result is 0, that means the pin is dangled. if an erro

↓ 3 callersMethodget_endpoint_group

Get the virtual endpoint group with an index. This mimics the interface given by a raw AIG.

src/staging.rs:35

↓ 3 callersMethodget_or_place_output_with_activation

returns a final local position for a data output bit with given pin_iv and clken_iv. if is not already placed, we will place it as well as place the

src/flatten.rs:330

↓ 3 callersMethoditer

src/bin/naive_sim.rs:110

↓ 3 callersFunctionplace_bit

( aig: &AIG, hier: &mut Vec<Vec<usize>>, hier_visited_nodes_count: &mut IndexMap<usize

↓ 3 callersMethodplace_clken_datainv

places a writeout bit's clock enable and data invert.

src/flatten.rs:313

↓ 2 callersFunctioncompute_lvl1_necessary_nodes

compute the set of nodes that must be implemented in level 1 in addition to the current hierarchy. the necessary_level1 nodes can only come from leve

↓ 2 callersFunctioncompute_reverse_level

compute the maximum number of steps needed from this node to reach an endpoint node. during this path, except the starting point, no node should alre

↓ 2 callersMethodhash

(&self, state: &mut H)

src/bin/cuda_test.rs:101

↓ 2 callersMethodis_empty

src/bin/flatten_test.rs:377

↓ 2 callersMethodnum_endpoint_groups

Get the number of endpoint groups that should be fulfilled with this staged AIG. This mimics the interface given by a raw AIG.

src/staging.rs:28

↓ 2 callersFunctionprocess_partitions

Given an initial clustering solution of endpoints, generate and map a refined solution. The refined solution will have smaller number of partitions a

↓ 2 callersMethodto_edges

src/repcut.rs:182

↓ 1 callersFunctionbuild_flattened_script_v1

( aig: &AIG, stageds: &[&StagedAIG], parts_in_stages: &[&[Partition]], num_blocks: usize, inpu

src/flatten.rs:686

↓ 1 callersFunctionbuild_one_boomerang_stage

build a single boomerang stage given the current inputs and outputs.

↓ 1 callersMethodbuild_script

( &self, aig: &AIG, part: &Partition, input_map: &IndexMap<usize, u32>, staged_io_map:

src/flatten.rs:467

↓ 1 callersMethoddfs_netlistdb_build_aig

recursively add aig pins for netlistdb pins for sequential logics like DFF and RAM, 1. their netlist pin inputs are not patched, 2. their aig pin inp

↓ 1 callersFunctionfind_top_scope

( items: &'i [ScopeItem], top_scope: &'_ str )

src/bin/flatten_test.rs:407

↓ 1 callersFunctionfind_top_scope

( items: &'i [ScopeItem], top_scope: &'_ str )

src/bin/cuda_test.rs:151

↓ 1 callersFunctionfind_top_scope

( items: &'i [ScopeItem], top_scope: &'_ str )

src/bin/naive_sim.rs:135

↓ 1 callersMethodhash

(&self, state: &mut H)

src/bin/flatten_test.rs:357

↓ 1 callersFunctionhash_of

src/bin/flatten_test.rs:59

↓ 1 callersMethodinit_afters_writeouts

( &mut self, aig: &AIG, staged: &StagedAIG, part: &Partition )

src/flatten.rs:229

↓ 1 callersMethodinto_iter

src/bin/flatten_test.rs:351

↓ 1 callersMethodinto_iter

src/bin/naive_sim.rs:79

↓ 1 callersMethodis_empty

src/bin/naive_sim.rs:105

↓ 1 callersMethodmake_inputs_outputs

( &mut self, aig: &AIG, staged: &StagedAIG, part: &Partition, input_ma

src/flatten.rs:371

↓ 1 callersFunctionmap_global_read_to_rounds

( inputs_taken: &BTreeMap<u32, u32> )

src/flatten.rs:110

↓ 1 callersFunctionmatch_scope_path

Try to match one component in a scope. If succeed, returns the remaining scope (can be None itself indicating all paths matched). If fails, return Non

src/bin/flatten_test.rs:391

↓ 1 callersFunctionmatch_scope_path

Try to match one component in a scope. If succeed, returns the remaining scope (can be None itself indicating all paths matched). If fails, return Non

src/bin/cuda_test.rs:135

↓ 1 callersFunctionmatch_scope_path

Try to match one component in a scope. If succeed, returns the remaining scope (can be None itself indicating all paths matched). If fails, return Non

src/bin/naive_sim.rs:119

↓ 1 callersMethodnum_vertices

Make an edge list. (weight, node indices)

src/repcut.rs:178

↓ 1 callersMethodpartition

Run mt-kahypar to partition this hypergraph.

src/repcut.rs:215

↓ 1 callersMethodpopcount

src/repcut.rs:44

↓ 1 callersFunctionprocess_partitions_from_hgr_parts_file

Read a cluster solution from hgr.part.xx file. Then call [process_partitions].

↓ 1 callersFunctionpurge_bit

( aig: &AIG, hier: &mut Vec<Vec<usize>>, hier_visited_nodes_count: &mut IndexMap<usize

↓ 1 callersFunctionrun_par

Call builtin partitioner.

src/bin/cut_map_interactive.rs:19

↓ 1 callersFunctionsimulate_block_v1

CPU prototype partition executor for script version 1.

src/bin/flatten_test.rs:66

↓ 1 callersFunctionsimulate_block_v1

CPU prototype partition executor for script version 1.

src/bin/cuda_test.rs:170

↓ 1 callersMethodto_endpoint_subset

build a staged AIG by horizontal splitting given a subset of endpoints. return built StagedAIG. the endpoints are given as a slice of endpoint group

src/staging.rs:63

Methodbuild_one

build one partition given a set of endpoints to realize. if the resource is overflowed, None will be returned. see [Partition] for resource constrain

src/repcut.rs:25

(aig: &AIG, vis: &mut IndexSet<usize>, ret: &mut Vec<usize>, is_primary_input: Option<&IndexSet<usize>>, u: us

Methoddirection_of

( &self, macro_name: &CompactString, pin_name: &CompactString, pin_idx: Option<isize>

src/aigpdk.rs:25

src/bin/flatten_test.rs:372

src/bin/cuda_test.rs:116

src/bin/naive_sim.rs:100

(&self, f: &mut fmt::Formatter<'_>)

src/repcut.rs:256

build a flattened script. `init_parts` give the partitions to flatten. it is better sorted in advance in descending order of #layers for better duty

src/flatten.rs:862

Methodfrom_full_aig

build a staged AIG that consists of all levels.

src/staging.rs:45

Methodfrom_netlistdb

(netlistdb: &NetlistDB)

Methodfrom_split

build a staged AIG by vertical splitting at the given level id. return built StagedAIG. the active middle nodes at split can be obtained from the Sta

src/staging.rs:96

Methodfrom_staged_aig

(aig: &AIG, staged: &StagedAIG)

src/repcut.rs:57

(&self, state: &mut H)

src/bin/naive_sim.rs:85

src/bin/cut_map_interactive.rs:64

src/bin/flatten_test.rs:425

src/bin/cuda_test.rs:431

src/bin/naive_sim.rs:153

src/bin/boomerang_test.rs:38

src/bin/cuda_dummy_test.rs:44

src/bin/level_test.rs:22

src/bin/repcut_test.rs:31

src/bin/flatten_test.rs:335

src/bin/cuda_test.rs:79

src/bin/naive_sim.rs:63

(cur: CompactString)

src/bin/flatten_test.rs:367

(cur: CompactString)

src/bin/cuda_test.rs:111

(cur: CompactString)

src/bin/naive_sim.rs:95

( &self, macro_name: &CompactString, pin_name: &CompactString )

src/aigpdk.rs:72