SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.09503 · ICML · 2021

Memory-Efficient Pipeline-Parallel DNN Training

Deepak Narayanan, Amar Phanishayee, Kaiyu Shi, Xie Chen, Matei Zaharia

arXiv · PDF · Open in the Atlas

Code that ran

2 of 10 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; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 0 failures of our instrument; 8 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2006.09503.json · badge: /api/ran/2006.09503.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/pipedream community
Papers with Code archive
2 ran · 8 no recorded run
FunctionStatusWhere it lives
Counter Ran
ran on a synthesized input
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("3c265f2d8c92f3ee")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. 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 threading

class Counter:

    def __init__(self, initial_count):
        self.count = initial_count
        self.cv = threading.Condition()

    def decrement(self):
        self.cv.acquire()
        self.count -= 1
        self.cv.notify_all()
        self.cv.release()

    def wait(self):
        self.cv.acquire()
        while self.count > 0:
            self.cv.wait()
        self.cv.release()


INIT_ARGS = {
    'initial_count': 3
}

ARGS = {}

METHOD = 'decrement'

counter = Counter(**INIT_ARGS)
counter.decrement()
counter.decrement()
counter.decrement()
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Queue Ran
ran on a synthesized input
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("111f03fa2d08d9c4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. 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 threading
import time

INIT_ARGS = {}

# For Queue, we need to call methods that add and remove tensors
# The class doesn't have a single "forward" or "__call__" method
# We'll test the add and remove methods

# Create a test tensor to add
test_tensor = torch.randn(2, 4)

# Since this class is about thread-safe queue operations, we need to demonstrate
# both add and remove. We'll use add as the primary method to test.
METHOD = 'add'
ARGS = {'tensor': test_tensor}

# Note: To properly test this class, you would typically:
# 1. Create an instance: queue = Queue()
# 2. Call queue.add(tensor) to add items
# 3. Call queue.remove() to retrieve items
# But since the harness expects a single method call, we're specifying 'add'
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
CommunicationHandler Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("89aff7109a54c076")
How it ran

no invocation recorded.

ModulesWithDependencies Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("3ae8ca5ecd61e9dc")
How it ran

no invocation recorded.

RuntimeStats Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("7cdedcfbbde8c89f")
How it ran

no invocation recorded.

StageRuntime Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("2973f073a38c500d")
How it ran

no invocation recorded.

_recv Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("2ab17ec9dae4b8e0")
How it ran

no invocation recorded.

_send Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("c577cb25127f2018")
How it ran

no invocation recorded.

recv_helper_thread Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("2d8868cdaa7fe53f")
How it ran

no invocation recorded.

send_helper_thread Not yet run
unverified (harvested, no recorded run)
msr-fiddle/pipedream/runtime/runtime.py
code served (permissive licence) · get_code("57414cce8342765c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 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 10 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 2026-11-21.

Repositories linked to this paper

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

Abstract

Many state-of-the-art ML results have been obtained by scaling up the number of parameters in existing models. However, parameters and activations for such large models often do not fit in the memory of a single accelerator device; this means that it is necessary to distribute training of large models over multiple accelerators. In this work, we propose PipeDream-2BW, a system that supports memory-efficient pipeline parallelism. PipeDream-2BW uses a novel pipelining and weight gradient coalescing strategy, combined with the double buffering of weights, to ensure high throughput, low memory footprint, and weight update semantics similar to data parallelism. In addition, PipeDream-2BW automatically partitions the model over the available hardware resources, while respecting hardware constraints such as memory capacities of accelerators and interconnect topologies. PipeDream-2BW can accelerate the training of large GPT and BERT language models by up to 20× with similar final model accuracy.

For agents

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

get_harvested_code_for_paper("2006.09503")
get_code_for_paper("2006.09503")
have("2006.09503")

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

curl https://syntology.ai/api/ran/2006.09503.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.09503.svg)](https://syntology.ai/paper/2006.09503)

Connect an agent — have() is free.