CtrlK
BlogDocsLog inGet started
Tessl Logo

nemo-mbridge-perf-megatron-fsdp

Operational guide for enabling Megatron FSDP in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.

59

Quality

68%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/nemo-mbridge-perf-megatron-fsdp/SKILL.md
SKILL.md
Quality
Evals
Security

Megatron FSDP Skill

For stable background and recommendation level, see:

  • @docs/training/megatron-fsdp.md
  • @skills/nemo-mbridge-perf-megatron-fsdp/card.yaml

Enablement

Minimal Megatron FSDP override in Bridge:

cfg.dist.use_megatron_fsdp = True
cfg.ddp.use_megatron_fsdp = True
cfg.ddp.data_parallel_sharding_strategy = "optim_grads_params"
cfg.ddp.average_in_collective = False
cfg.checkpoint.ckpt_format = "fsdp_dtensor"

Example recipe fixup:

cfg = llama3_8b_pretrain_config()
cfg.dist.use_megatron_fsdp = True
cfg.ddp.use_megatron_fsdp = True
cfg.ddp.data_parallel_sharding_strategy = "optim_grads_params"
cfg.ddp.average_in_collective = False
cfg.checkpoint.ckpt_format = "fsdp_dtensor"
cfg.checkpoint.save = "/tmp/fsdp_ckpts"
cfg.checkpoint.load = None

Performance harness note:

python scripts/performance/launch.py --use_megatron_fsdp true

Code Anchors

Bridge config definition:

use_megatron_fsdp: bool = False
"""Use Megatron's Fully Sharded Data Parallel. Cannot be used together with use_torch_fsdp2."""

use_torch_fsdp2: bool = False
"""Use the torch FSDP2 implementation. FSDP2 is not currently working with Pipeline Parallel.
It is still not in a stable release stage, and may therefore contain bugs or other
potential issues."""

Bridge validation:

if self.dist.use_megatron_fsdp and self.dist.use_torch_fsdp2:
    raise ValueError(...)
...
assert not self.dist.use_tp_pp_dp_mapping, "use_tp_pp_dp_mapping is not supported with Megatron FSDP"
...
assert self.checkpoint.ckpt_format == "fsdp_dtensor", (
    "Megatron FSDP only supports fsdp_dtensor checkpoint format"
)

Runtime wrapper selection:

if use_megatron_fsdp:
    DP = FullyShardedDataParallel
elif use_torch_fsdp2:
    DP = TorchFullyShardedDataParallel
else:
    DP = DistributedDataParallel
...
DP(
    config=get_model_config(model_chunk),
    ddp_config=ddp_config,
    module=model_chunk,
    ...
    pg_collection=pg_collection,
)

Perf harness overrides:

recipe.ddp.use_megatron_fsdp = True
recipe.ddp.data_parallel_sharding_strategy = "optim_grads_params"
recipe.ddp.keep_fp8_transpose_cache = False
recipe.ddp.average_in_collective = False
...
recipe.checkpoint.load = None

Pitfalls

  1. Public recipes often expose use_megatron_fsdp but still default to ckpt_format="torch_dist". If save/load is enabled, switch to fsdp_dtensor.
  2. use_torch_fsdp2 exists, but on the validated branch Bridge still fails before training because _ddp_wrap passes pg_collection.
  3. CPU offloading is only valid when pipeline_model_parallel_size == 1 and activation recomputation is disabled.
  4. Upstream warns that FSDP and TP/CP can want different CUDA_DEVICE_MAX_CONNECTIONS settings on Hopper and earlier.
  5. Megatron FSDP and FSDP2 are mutually exclusive.

Verification

Use the existing 2-GPU functional smoke test:

CUDA_VISIBLE_DEVICES=0,1 uv run python -m torch.distributed.run --nproc_per_node=2 \
  -m pytest tests/functional_tests/training/test_megatron_fsdp.py::TestMegatronFSDP::test_fsdp_pretrain_basic -v -s

Success criteria:

  • Pytest reports 1 passed
  • The log shows finite loss at the last iteration
  • The run finishes without a checkpoint format assertion
Repository
NVIDIA/skills
Path

skills/nemo-mbridge-perf-megatron-fsdp/SKILL.md

Commit

17d96f1

Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.

Claim skill