P087 Hardware heterogeneity and technology change
Does any verification protocol run unmodified across accelerators from more than one vendor, and how much work does porting to a new hardware generation take?
Hardware heterogeneity and portability of protocols.
Terms: hardware heterogeneity: a regime having to work across accelerators from different vendors and generations. ICI (inter-chip interconnect): the high-speed links joining accelerators (e.g. on Google TPUs). flexHEG: a specific proposed open hardware-governance design.
Status
Posed
Posed by
- Harack et al. (2025), Verification for International AI Governance (Oxford AIGI):
hardware stacks diverge from one another due to competition among hardware and software providers
(§1.5.3, p.36, L1977–1978) (source states the problem but not the quantity the question asks for)
Also raised in: Petrie, Aarne, Ammann & Dalrymple (2025), Overview of Flexible Hardware-Enabled Guarantees (flexHEG Part 1); Petrie & Aarne (2025), Technical Options for Flexible Hardware-Enabled Guarantees (flexHEG Part 2); Canary Institute research.
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P087.1 | Does any proposed verification protocol run unmodified on accelerators from multiple vendors (e.g. NVIDIA and AMD or Huawei)? | implied by a source, not stated as open |
| P087.2 | How many engineer-months does porting a verification protocol to a new accelerator generation take? | open |
| P087.3 | Does any tap design work on bidirectional (circulator) optical links at TPU ICI rates, resolving the tension that the most replay-friendly stack is the least tappable? | open |
| P087.4 | Has an integrated flexHEG prototype (enclosure + guarantee processor) been demonstrated on a frontier data-center accelerator with frontier workloads, on short notice? | open |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P087.5 | Must components central to verification be purpose-built and co-designed by rival parties? |
Possibly relevant work found by automated search
Found by AI agents searching for this problem's sub-questions, mostly by keyword and citation in the OpenAlex scholarly database and on the web; a second AI pass reviewed each one, and we spot-checked some by hand. They seemed reasonably related, as far as one can tell from outside each paper's field. Some come from neighbouring fields, such as arms control and financial audit, and bear on the problem by analogy. Not reviewed against the status above; a paper listed here does not change that status. 928 of 1052 quotes checked word for word against the paper's text.
For P087.1: Does any proposed verification protocol run unmodified on accelerators from multiple vendors (e.g. NVIDIA and AMD or Huawei)?
Data
- problems.csv — one row per problem
- problem_status.csv — status, scope and evidence per problem
- problem_sources.csv — the posed-by quote and other documents per problem
- leaves.csv — all sub-questions, including those without a checkable form
- leaf_sources.csv — every source row behind each sub-question