Skip to content

Research

Every result the lab publishes, each with its area. Each page states the problem, what was shown, who we expect would care, what the result does not show yet, and the lab’s file that states it.

  • ChipletOS · Chip packagingA fast coupling model, graded by outside solversOn every sample set FastCap graded, and on the one slice of layouts Palace graded, the lab’s fast model of how electrical signals couple between the vertical wires of a chip package came closer to those two outside physics programs than the lab’s own pair-by-pair sum did, for one property, the charge they can store (capacitance). That sum is a baseline the lab defined, and the lab’s own records say full-wave field solvers do not use it.
  • AxiomLimit · AI-cluster networkingFixed-size memory for reordering AI network dataIn the lab’s own network simulator, a network-card design that keeps track of out-of-order data in a small, fixed amount of memory, compared on the same simulated grid with the lab’s model of STrack, a published fixed-size design, and with a second bounded-state design in the lab’s simulator.
  • ChipletOS · Chip printingChecked brightness ranges for chip printsA check of how a chip pattern will print that gives, for every point of the image, a range for its brightness. On every test mask the lab’s detailed simulation fell inside the range, apart from a rounding error in one internal step that the range leaves out.
  • OrbitalProof · AI agentsCrash recovery for AI assistants’ multi-step changesA transaction layer meant to finish or undo an AI assistant’s multi-step change after a crash, for plans it accepts before they run. At every crash point the lab chose, it recovered the change whole or not at all; killed at random moments, some recoveries were not, and the lab has not fixed that yet.
  • ChipletOS · Chip packagingThe capacitance flat models leave outThe lab’s own three-dimensional solver measured how much of one simplified chip-package connection’s capacitance (its charge-storing capacity) flat, two-dimensional models leave out.
  • OrbitalProof · AI-cluster networkingAn Ultra Ethernet recovery design, proved in a modelA machine-checked proof, in the lab’s one small model, that its design for the loss-recovery step of Ultra Ethernet’s ordered-delivery mode (the step that deals with data that has gone missing) never passes data on out of sequence order, or twice; it is about ordering, not delivery.
  • ChipletOS · Chip packagingPair-by-pair coupling estimates overstate the worst case, in a modelA proof, over every layout in one narrow family in a simplified model, that adding up coupling one pair at a time always overstates the worst case by at least a stated minimum.
  • OrbitalProof · AI agentsTool combinations that leak a secretA run the lab reports as exhaustive, in a test world the lab built, over every sequence of up to three AI-assistant tool calls, finding the smallest combinations that leak a stored secret.
  • AxiomLimit · AI inferenceWhere deletion receipts can be fooledIn a small simulated setup, six concrete ways a signed deletion receipt can check out while the data is still recoverable or was never deleted.
  • ChipletOS · Chip printingSigning off some tiles near a chip-mask edit, in simulationA method that signs off, in the lab’s own simulator, a small piece of a chip-mask design (a tile) next to a small edit as still valid without recomputing it. It applies when a bound the lab computes on the edit’s effect fits inside the tile’s safety margin; this site does not name what checks that bound.
  • OrbitalProof · Wi-Fi securityAn automated attack search against quantum-safe Wi-Fi sign-inAn automated search, given no starting attack sequence and no hint of which combination might work, tried combinations of moves from the lab’s own list of known attacks against the lab’s hardened models of quantum-resistant Wi-Fi sign-in designs, and, within a small fixed budget, found no combined attack that got through.
  • OrbitalProof · Wi-Fi securityA memory ceiling for post-quantum Wi-Fi, proved in a modelIn the lab’s model of its admission-control module (the code that decides which sign-ins an access point accepts), a machine-checked proof that an access point’s memory for half-received quantum-resistant sign-ins stays capped, however many devices an attacker connects.
  • AxiomLimit · AI inferenceThe least bookkeeping a shared AI cache needs, in a modelA machine-checked proof that a shared AI cache which never leaks between customers and always reuses what it may must remember at least one bit for every pairing of a customer with a cached item, where any item can be shared with any group of customers (the lab’s “dense rights model”).
  • ChipletOS · Chip printingA near-constant-size sign-off record for a photomaskA sign-off record for a whole photomask that stayed a little over a kilobyte in size across every mask size the lab tried; its size does not buy assurance.

More

How we show numbers

Every number on this site links to the file it comes from. How each result is checked

  • We never show a number before its file has loaded.
  • A question we have not checked yet is marked as unchecked.
  • A check that found nothing says so.
  • A file with no value for a question says so.
  • A number whose file is missing or has changed is not shown.
  • Two files that disagree about what a number describes are both flagged.
  • A number from too few samples shows its sample size.
  • Two files that give different values are both shown.
  • A file we cannot publish is listed by its fingerprint only.
  • A measurement more than a week old shows its age.
  • A question that does not apply to a page is left off it.
  • A measurement whose program failed is shown as failed.