|
Download FAQ.md from elyanlabs/ram-coffers: direct link, hf CLI and curl.
- Browser
- Download file 3.75 kB
-
https://huggingface.co/elyanlabs/ram-coffers/resolve/main/FAQ.md
- Command line
-
hf download hf://elyanlabs/ram-coffers/FAQ.md
-
curl -L -o FAQ.md https://huggingface.co/elyanlabs/ram-coffers/resolve/main/FAQ.md
3.75 kB
| # RAM Coffers FAQ | |
| This FAQ answers common onboarding questions for contributors and operators who | |
| are reading RAM Coffers as part of the broader Proof of Physical AI and | |
| RustChain ecosystem. | |
| ## Is RAM Coffers a token treasury? | |
| No. In this repository, "coffers" refers to NUMA-aware memory banks for LLM | |
| inference. RAM Coffers is an inference optimization and research codebase, not a | |
| custodial wallet, exchange, bank, or treasury contract. | |
| The project is connected to RustChain because the same physical hardware that | |
| runs RAM Coffers can also participate in Proof of Antiquity. That means the | |
| hardware can do useful AI work and provide physical-device proof, but this repo | |
| does not hold user funds or define payout policy by itself. | |
| ## How does RAM Coffers differ from traditional token treasuries? | |
| Traditional token treasuries usually focus on managing balances, grants, votes, | |
| or spending from an on-chain or organizational pool. | |
| RAM Coffers focuses on hardware-local memory routing: | |
| - Coffer placement maps model knowledge to NUMA nodes or cache tiers. | |
| - Resonance routing selects the relevant memory bank for a query. | |
| - POWER8-specific primitives such as DCBT, VSX, and vcipher reduce memory and | |
| attention cost during inference. | |
| The "value" being organized here is physical compute and memory locality, not a | |
| pooled token balance. | |
| ## What is the minimum RTC amount for treasury participation? | |
| This repository does not define a minimum RTC amount for participation. RAM | |
| Coffers documents the inference side of the stack; RTC participation rules live | |
| in the RustChain ecosystem and may change over time. | |
| For current RTC onboarding, mining, wallet, and bounty details, check the | |
| RustChain documentation and active bounty board rather than relying on this FAQ | |
| as a payment policy source. | |
| ## How are allocation decisions made? | |
| For RAM Coffers, allocation decisions are technical rather than financial: | |
| - Model shards are assigned to coffers based on memory capacity and domain role. | |
| - Inference requests route to coffers by resonance matching. | |
| - Threads and prefetch hints are chosen to keep hot paths close to the relevant | |
| POWER8 NUMA node. | |
| For contribution rewards or bounty allocation, maintainers evaluate public | |
| issues, pull requests, reviews, and validation evidence in the relevant GitHub | |
| repositories. | |
| ## Can I withdraw my treasury contribution? | |
| RAM Coffers does not accept or custody treasury deposits, so there is nothing to | |
| withdraw from this repository. | |
| If you are asking about RTC mining, bounty payouts, wallet transfers, or any | |
| other token movement, use the official RustChain wallet and bounty instructions. | |
| Do not share private keys, seed phrases, keystores, passwords, or verification | |
| codes in issues or pull requests. | |
| ## How does antiquity affect treasury rewards? | |
| Proof of Antiquity is the RustChain mechanism that rewards real physical | |
| hardware, with vintage or distinctive hardware contributing to the proof model. | |
| RAM Coffers is relevant because it shows that the same physical machine can run | |
| useful AI inference, not only attest to its existence. | |
| In practical terms: | |
| - RAM Coffers improves inference performance on verified physical hardware. | |
| - Proof of Antiquity belongs to RustChain's reward and attestation layer. | |
| - Reward amounts, eligibility, and payout status should be checked against the | |
| current RustChain rules and public bounty records. | |
| ## Where should new readers start? | |
| Start with: | |
| 1. `README.md` for the research context and performance summary. | |
| 2. `QUICK_START.md` for the shortest build and run path. | |
| 3. `CONTRIBUTING.md` for contribution expectations. | |
| 4. `BCOS.md` for certification context. | |
| 5. `docs/PPA_INTEGRATION.md` for how RAM Coffers fits into Proof of Physical AI. | |