Activations, accumulation, and checkpointing
Use activations, accumulation, and checkpointing to move the gpu training production brief toward a defensible release.
Training repeatedly OOMs and misses its nightly completion window.
This lesson isolates activations, accumulation, and checkpointing as one decision inside that system. The people affected are product users and operators; the learning data must carry event time, availability time, ownership, and version; and the operating envelope is 24 GB device and nightly completion.
Choose whether and how to use FP16/BF16 at a declared prediction cutoff.
Measure decision utility alongside calibration, slice reliability, and system latency—not model score alone.
Training repeatedly OOMs and misses its nightly completion window. An unsafe release must degrade to a named baseline or the last known-good version.
Build the mental model before the machinery.
The core move is to treat activations, accumulation, and checkpointing as a contract between data, a computation, and an action. Record hardware/software, shapes, precision, seeds, throughput method, peak memory, and cost per sample. The implementation becomes easier to debug once you can state which inputs exist, which state is learned, what output means, and what must remain invariant after serialization.
FP16/BF16
Define it in a hand-checkable form and name the prediction-time inputs.
loss scaling
Connect it to the production metric and identify what it cannot guarantee.
activation memory
Stress it with a slice, a temporal boundary, and a failure-safe alternative.
profilers
Stress it with a slice, a temporal boundary, and a failure-safe alternative.
Lessons 1, 2, 3 in this course.
Name every symbol. Check every shape.
Roofline bound is the central invariant for this lesson. The formula is useful only when its inputs match the production cutoff and its output maps to an action.
Roofline bound
| Symbol | Meaning / shape / unit |
|---|---|
P_peak | peak compute |
I | arithmetic intensity |
B_mem | memory bandwidth |
Open derivation and numerical substitution
Start from the production quantity being optimized, substitute the observed values with their declared units, then isolate the model-controlled term. Preserve shape annotations at each step so broadcasting or aggregation cannot silently change the result.
- Write the named inputs: P_peak, I, B_mem.
- Substitute one small, hand-checkable batch before vectorizing.
- Calculate an independent reference value and compare within a declared tolerance.
# equation → code contract
inputs = validate_shapes_and_units(batch)
value = compute_c12(inputs)
assert is_finite(value)Calculate it small. Shape it realistically. Break it on purpose.
A result you can reproduce on paper
Compare one large matrix multiplication with thousands of tiny kernel launches.
- Write every input and unit.
- Substitute values into the roofline bound equation above.
- Compare the result to one simple baseline and explain the direction of the difference.
The same reasoning under real constraints
Profile a decoder trainer, overlap input transfer, enable mixed precision, tune shapes, and checkpoint selected activations.
The production record includes the data snapshot, transformation state, artifact identity, cutoff, score, decision, and the version of the policy that consumed it.
The attractive result you should reject
Unsynchronized timing reports impossible speedups while retained tensors keep computation graphs alive and leak memory.
Diagnostic: replay the smallest failing slice from immutable inputs, then compare each boundary rather than retuning the model.
Change one assumption and make the tradeoff visible.
This lab runs predefined TypeScript only. It never executes learner code. Use the slider, numeric input, reset, live text, or table—the computation is the same.
GPU memory ledger
Change microbatch size and calculate a simplified training-memory budget.
Assumption: 3.2 GB weights+grads+optimizer, 0.34 GB activations per sample, 24 GB device, 10% reserve.
Open nonvisual data table
| Item | Computed state | Interpretation |
|---|---|---|
| block 1 | 1.09 GB activations | resident |
| block 2 | 2.18 GB activations | resident |
| block 3 | 3.26 GB activations | resident |
| block 4 | 4.35 GB activations | resident |
| block 5 | 5.44 GB activations | resident |
Trace the complete operating path.
- 01
Validate and version FP16/BF16.
- 02
Compute activations, accumulation, and checkpointing from prediction-time-safe inputs.
- 03
Persist model, feature, and configuration identities together.
- 04
Serve or materialize behind explicit 24 GB device and nightly completion.
- 05
Join telemetry to mature outcomes and retain a rollback path.
Observability
Join service health, input quality, prediction distributions, slice behavior, and mature outcomes by exact version.
Cost
Measure storage, preprocessing, compute, queueing, and human review under a representative arrival pattern.
Failure modes
Unsynchronized timing reports impossible speedups while retained tensors keep computation graphs alive and leak memory. Add a detector, owner, mitigation, and stop condition for this class of failure.
Alternatives
Compare a rule, a simpler statistical baseline, and a different system boundary before adding model complexity.
Explain the contract, not just the vocabulary.
Single-GPU optimization plan
Profile and optimize a correctness-locked trainer without sacrificing model quality.
- Profiler diagnosis
- Mixed precision parity
- Memory plan
- Throughput/cost report
Read primary material with a purpose.
Return to the opening failure.
Record hardware/software, shapes, precision, seeds, throughput method, peak memory, and cost per sample.
For this lesson, the release evidence is a hand-checked formal result, deterministic simulation output, a ≥80% checkpoint, the production rubric, and a named fallback. The course resolves when the system can produce a profiled, mixed-precision, memory-budgeted training loop.