Show HN: Qumulator – quantum circuit simulator, 1000 qubits, no GPU

A “1,000-qubit” miracle? The comments immediately hit the brakes

TLDR: Qumulator says it can handle some 1,000-qubit quantum simulations on normal cloud CPUs, which is a huge claim in a field known for impossible hardware demands. Commenters were impressed but quickly argued the fine print: is this a true all-purpose leap, or just a clever shortcut for special cases?

A new Show HN post is making a big, shiny promise: simulate quantum circuits with up to 1,000 qubits on ordinary cloud computers, with no GPU and no futuristic lab gear. On paper, that sounds like sci-fi. In practice, the community reaction was a classic internet combo of “whoa” and “okay, but what’s the catch?”

The biggest mood in the comments was curious skepticism. One camp was impressed by the flex — 1,000 qubits, fast runtimes, tiny memory, and a simple API you can call over the internet. That’s the sort of claim that makes people sit up straight. But the hottest pushback came fast: is this really a general-purpose simulator, or is it only fast for the kinds of quantum problems that happen to be easier to compress? In plain English, commenters were asking whether this is a universal breakthrough or a very smart shortcut wearing a superhero cape.

That tension became the real drama. One commenter wanted the inevitable showdown: how does this stack up against rivals like Bluequbit? Another cut right to the heart of the hype, basically saying the 1,000-qubit number only counts when the problem has a clever loophole. It’s a familiar tech-launch storyline: bold benchmark drops, then the comments turn into MythBusters. No huge flame war yet, but the vibe is definitely “cool demo, now show the receipts.”

Key Points

  • Qumulator is presented as a cloud API and Python SDK for simulating quantum circuits and related systems on standard classical hardware without GPUs or quantum hardware.
  • The article claims a 1,000-qubit circuit at depth 3 can run in under 1 second using 1 MB of memory, and a 105-qubit Willow-layout circuit at depth 5 can complete in under 0.5 seconds.
  • Its proprietary KLT Engine is described as automatically selecting among internal representations such as tensor networks, Gaussian covariance matrices, nexus graphs, and full statevectors based on problem structure.
  • Benchmarks are reported for multiple tasks, including Bell tests, Heisenberg chains, photonic hafnians, random circuit sampling, discrete time crystals, a holographic wormhole, anyon braiding, a Kitaev chain, QUBO optimization, and a Kuramoto BEC system.
  • The SDK supports installation via pip, optional integrations with Qiskit and Cirq, API key creation, demo commands, Python usage examples, and OpenQASM 2/3 input.

Hottest takes

"how does it work and compare to other simulators" — stubbi
"doesn't simulate any general purpose quantum circuit with 1000 qubits" — alyxya
"only ones where there's a more efficient strategy" — alyxya
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.