Frequently Asked Questions
Everything you need to know about Epochly.
Getting Started
What is Epochly?
Epochly is a drop-in Python performance overlay. Install it with pip, run your existing scripts — no code changes needed. It automatically accelerates CPU-bound workloads using JIT compilation, GPU acceleration, and multicore parallelism.
How do I install Epochly?
Run pip install epochly and then run your Python scripts as usual. Epochly activates automatically. Use epochly.stats() to see what it's doing.
Where does it run?
Linux, Windows, and macOS. x86-64 and ARM64 architectures. Python 3.9 through 3.13.
How do I disable it instantly?
Set EPOCHLY_DISABLE=1 as an environment variable, or uninstall with pip uninstall epochly. Either way, your code returns to normal Python immediately.
Safety & Correctness
Is Epochly safe for production?
Yes. Epochly uses progressive enhancement — it monitors first, optimizes only after verifying stability, and automatically falls back to standard Python if anything looks wrong. No data corruption, ever.
Will it break my code?
No. Epochly monitors first and only optimizes when safe. If issues are detected, it falls back to standard Python automatically.
What about security and compliance?
Epochly runs entirely on your infrastructure with no data exfiltration. Enterprise tier is on our roadmap and will add Epochly Sandbox for isolated execution, SSO, audit logs, and on-premise deployment.
Can I run Epochly in air-gapped environments?
Epochly runs locally with no external network requirements. Enterprise tier is on our roadmap and will include formal on-premise deployment support and compliance documentation.
What data does it collect?
Execution patterns and performance metrics for optimization. No source code or data values. Telemetry is configurable and can be disabled entirely.
Performance
How much faster will my code run?
It depends on your workload. Numerical Python loops see up to 193x speedup via JIT. Large array operations (10M+ elements) see up to 70x on GPU. Parallel CPU-bound work sees 8-12x. Workloads already optimized (vectorized NumPy, C extensions) won't see gains — Epochly detects this and yields gracefully.
How does this compare to rewriting in Rust/C++?
Epochly works today with no code changes. A rewrite can take months and requires ongoing maintenance. Many teams use Epochly to get immediate gains while evaluating longer-term options.
What workloads does Epochly NOT help with?
I/O-bound work (network, disk), already-vectorized NumPy, code already using GPU, Numba/Cython-compiled code, and sub-millisecond workloads where optimization overhead exceeds the benefit. Epochly detects these cases and stays out of the way.
Pricing & Plans
What does the free tier include?
Community tier is free forever and includes Levels 0-3 optimization (up to 4 parallel cores), basic telemetry, and community support. No GPU acceleration.
What happens after my Pro trial ends?
After 30 days, your subscription begins automatically at the plan you selected. Cancel anytime before the trial ends to avoid being billed — your account will revert to Community tier (4 cores, no GPU acceleration). No data is lost.
How do I start a Pro trial?
Visit the pricing page and click "Get Pro". You’ll enter your payment details but won’t be charged for 30 days. Pro features activate immediately. Cancel anytime before the trial ends.
Do I need a credit card to try Pro?
The web checkout requires a payment method, but you won’t be charged during the 30-day trial period. Cancel anytime before it ends to avoid being billed.
Can I use Epochly in production?
Yes. Community tier is production-ready for workloads that fit within 4 cores. For unlimited scaling and GPU acceleration, upgrade to Pro.
What does Enterprise include?
Enterprise includes everything in Pro, plus fleet-wide Lens dashboard, unlimited alerts, 13-month data retention, RBAC with audit logs, and priority support. Contact us to discuss your requirements.
Is there a team plan?
Pro supports individual teams. For fleet-wide visibility, RBAC, and extended retention, contact us about Enterprise.
Still have questions?
We're here to help. Reach out and we'll get back to you.