SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.13143 · ICML · 2024

Integrated Hardware Architecture and Device Placement Search

Amar Phanishayee, Jakub Tarnawski, Divya Mahajan, Irene Wang

arXiv · PDF · Open in the Atlas

Code that ran

4 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 1 failure of our instrument; 12 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2407.13143.json · badge: /api/ran/2407.13143.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
msr-fiddle/phaze canonical (classifier label)
classifier basis: classifier_2025
4 ran · 12 no recorded run
FunctionStatusWhere it lives
create_core_config Ran
ran, honoured the contract we drafted
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("ec69c754f196e872")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned per_core_config. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from collections import namedtuple, OrderedDict

torch.manual_seed(0)

# Define a stand-in config class that has the required attributes
class CoreConfig:
    def __init__(self):
        self.num_tc = 4
        self.num_vc = 8
        self.width = 16
        self.depth = 32
        self.width_vc = 24
        self.GLB_Buffer = 512

cc = CoreConfig()

ARGS = {
    'cc': cc,
    'core_type': 'TC'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type per_core_config — a per_core_config output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a per_core_config output the recorded ndim 1 and dtype family int describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_config_across_models Ran
ran on a synthesized input
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("245dff1bccbff4b8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

class MockOutput:
    def __init__(self, throughput):
        self.throughput = throughput

# Set up the global explored_acc_configs dictionary
explored_acc_configs = {
    "config_1": {
        "model_a": MockOutput(throughput=100.5),
        "model_b": MockOutput(throughput=95.3),
        "model_c": MockOutput(throughput=102.1),
    },
    "config_2": {
        "model_a": MockOutput(throughput=98.2),
        "model_b": MockOutput(throughput=99.8),
    },
    "config_3": {
        "model_a": MockOutput(throughput=105.0),
        "model_b": MockOutput(throughput=103.5),
        "model_c": MockOutput(throughput=104.2),
        "model_d": MockOutput(throughput=106.1),
    },
}

ARGS = {}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_out_filename Ran
ran; our contract draft was wrong, not the code
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("73fb0cda316ba202")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

# Define the model lists as they appear in the function
bert_models = ['bertbase', 'bertlarge', 'megatronbert']
gpt_models = ['gpt2', 'megatrongpt3', 'megatrongpt2-xl', 'megatrongpt2-54', 'megatrongpt2-72']
llama_models = ['llama2']
megatron_language_models = ['megatronbert', 'megatrongpt3', 'megatrongpt2-xl', 'megatrongpt2-54', 'megatrongpt2-72']
opt_models = ['opt']
language_models = bert_models + gpt_models + opt_models + llama_models

# Test with a megatron language model to exercise the full parameter set
ARGS = {
    'model_name': 'megatronbert',
    'ext': 'txt',
    'micro_batch_size': 8,
    'tmp_width': 2,
    'sequence_length': 512,
    'module_type': 'encoder'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_num_cores Ran
ran, honoured the contract we drafted
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("a2b8a47f1a78cbfb")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

class Config:
    def __init__(self):
        self.num_vc = 8
        self.num_tc = 16

config = Config()

ARGS = {
    "config": config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
device_placement Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("88f36ee3e7356cdc")
How it ran

no invocation recorded.

extract_op_graph_ilp Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("726c74abb2a2fd17")
How it ran

no invocation recorded.

get_configs_to_explore Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("045095c47c56ff37")
How it ran

no invocation recorded.

get_next_acc_config Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("53484ad987afc3c9")
How it ran

no invocation recorded.

ilp Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("527a9f1d884b10d7")
How it ran

no invocation recorded.

latency_per_layer Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("1185447c7929792e")
How it ran

no invocation recorded.

optimize_latency Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("54176a7933685d40")
How it ran

no invocation recorded.

run_solver Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("0f6fdba25526d18b")
How it ran

no invocation recorded.

schedule_using_ilp_x_values Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("2f8e76fed12d4260")
How it ran

no invocation recorded.

solve Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("48468011bfdc665a")
How it ran

no invocation recorded.

sort_acc_configs_by_area Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("b3f18a53011bc087")
How it ran

no invocation recorded.

write_explored_configs_to_file Not yet run
unverified (harvested, no recorded run)
msr-fiddle/phaze/Solver/solver.py
code served (permissive licence) · get_code("5594ad139855a245")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Distributed execution of deep learning training involves a dynamic interplay between hardware accelerator architecture and device placement strategy. This is the first work to explore the co-optimization of determining the optimal architecture and device placement strategy through novel algorithms, improving the balance of computational resources, memory usage, and data distribution. Our architecture search leverages tensor and vector units, determining their quantity and dimensionality, and on-chip and off-chip memory configurations. It also determines the microbatch size and decides whether to recompute or stash activations, balancing the memory footprint of training and storage size. For each explored architecture configuration, we use an Integer Linear Program (ILP) to find the optimal schedule for executing operators on the accelerator. The ILP results then integrate with a dynamic programming solution to identify the most effective device placement strategy, combining data, pipeline, and tensor model parallelism across multiple accelerators. Our approach achieves higher throughput on large language models compared to the state-of-the-art TPUv4 and the Spotlight accelerator search framework. The entire source code of PHAZE is available at https://github.com/msr-fiddle/phaze.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2407.13143")
get_code_for_paper("2407.13143")
have("2407.13143")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2407.13143.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2407.13143.svg)](https://syntology.ai/paper/2407.13143)

Connect an agent — have() is free.