{ "abstract": "Modern computers contain many computational systems that can already operate concurrently, yet their cooperation is usually organized through pairwise APIs, queues, function calls, serialized messages, or private shared buffers. A physical display-to-camera experiment suggested a different primitive: one system may publish state into an observable field, while another acquires only the region, resolution, channels, and time span it needs. This paper specifies the Virtual Optical Compute Fabric (VOCF): a GPU-resident and transport-independent spatial runtime in which computational systems publish, observe, transform, and recursively refine shared state through programmable multiscale views. The word optical describes an observation-based programming model; it does not require photonic hardware, visible RGB pixels, monitors, or camera capture inside the host. Internally, a surface may be a bitplane, storage texture, sparse tensor, graph field, token lattice, event plane, or structured buffer. The architecture consists of Fabrics, Surfaces, Panels, Lenses, Operators, Instances, Epochs, Commit Gates, Bridges, and Receipts. Many Lenses may sample one canonical Surface without materializing many complete copies. A single host can therefore instantiate global, dual, quad, tiled, or thousands of sparse views, limited by memory traffic, transform cost, dispatch overhead, synchronization, contention, locality, and compute rather than monitor refresh rate. VOCF is a specified composition, not a performance result. The first required experiment compares multiscale field coordination with full-field and queue-based baselines under controlled workloads.", "author": "Hayden Lindley", "category": "Systems & Architecture", "claims": { "claims": [ { "claim_id": "P30-C01", "evidence_refs": [ "P30-S04", "P30-S05", "P30-S06" ], "limitations": [ "API availability does not establish that the proposed abstraction is efficient.", "Feature support, limits, synchronization behavior, and performance vary by backend and device." ], "public": true, "scope": "Implementation substrate available in Vulkan, WebGPU, and CUDA-class systems.", "statement": "Current GPU APIs expose storage buffers, storage images or textures, image or texture views, compute pipelines, explicit resource bindings, and synchronization mechanisms sufficient to implement a bounded single-host prototype of Surfaces, Lenses, and Operators.", "status": "SOURCE_REVIEWED" }, { "claim_id": "P30-C02", "evidence_refs": [ "P30-S01", "P30-S11", "P30-S12", "P30-S13", "P30-S14", "P30-S15" ], "limitations": [ "No reference runtime or conformance suite is included in version 0.1.0.", "Several policies, wire formats, and type-system details remain open." ], "public": true, "scope": "The Virtual Optical Compute Fabric architecture defined in this paper.", "statement": "A runtime composed of Fabrics, Surfaces, Panels, Lenses, Operators, Instances, Epochs, Commit Gates, Bridges, and Receipts forms a coherent specification for observation-based multiscale coordination among computational systems.", "status": "SPECIFIED" }, { "claim_id": "P30-C03", "evidence_refs": [ "P30-S01", "P30-S04", "P30-S05", "P30-S06", "P30-S16" ], "limitations": [ "Lens scheduling, indirection, dispatch, and synchronization may cost more than the work avoided.", "The hypothesis may not apply to dense or non-spatial workloads." ], "public": true, "scope": "Sparse and spatially structured workloads under the VOCF-1 experiment.", "statement": "For workloads with genuine spatial locality, sparsity, or multiscale structure, dynamic Lenses can reduce read amplification relative to repeated full-field observation while preserving accepted output correctness.", "status": "HYPOTHESIS" }, { "claim_id": "P30-C04", "evidence_refs": [ "P30-S01", "P30-S07", "P30-S09", "P30-S10", "P30-S13" ], "limitations": [ "A common field does not solve semantic incompatibility.", "Adapters, schemas, validation, and authority policies remain necessary.", "The abstraction may add more complexity than it removes for simple pipelines." ], "public": true, "scope": "Integration among models, agents, simulations, verifiers, and deterministic operators.", "statement": "A common Surface and Lens boundary can reduce some pairwise adapter and context-assembly burden in heterogeneous compound systems.", "status": "HYPOTHESIS" }, { "claim_id": "P30-C05", "evidence_refs": [ "P30-S01", "P30-S07", "P30-S08", "P30-S09" ], "limitations": [ "The fidelity of a virtual Instance to later physical deployment is unverified.", "Nested scheduling and resource isolation may distort comparisons." ], "public": true, "scope": "Architecture-laboratory use of nested VOCF Instances.", "statement": "Recursive Instances can represent and compare multiple virtual computational topologies within one host under a shared measurement harness.", "status": "SPECIFIED" }, { "claim_id": "P30-C06", "evidence_refs": [ "P30-S01", "P30-S02", "P30-S03", "P30-S04", "P30-S05" ], "limitations": [ "Different transports have radically different bandwidth, latency, reliability, and directionality.", "Some workloads may require transport-specific application redesign.", "No cross-transport conformance experiment has been run." ], "public": true, "scope": "Transport-independent Bridge architecture.", "statement": "The higher-level observation contract can be mapped across GPU-local, process-local, network, and physical optical transports while retaining Surface, Lens, Epoch, and receipt semantics.", "status": "HYPOTHESIS" }, { "claim_id": "P30-C07", "evidence_refs": [ "P30-S01", "P30-S09", "P30-S10", "P30-S12" ], "limitations": [ "Visualizations can omit or distort machine state.", "No human-subject or expert-operator study has been run.", "Improved legibility may not improve decisions or outcomes." ], "public": true, "scope": "Human inspection and debugging of VOCF runtime state.", "statement": "A human-visible projection of the same state used by machine Operators can improve inspection of compound computational work without making rasterized pixels the canonical representation.", "status": "HYPOTHESIS" } ], "disclosure_note": "Claims marked non-public in the authored ledger are omitted from this public artifact.", "paper": "p30-virtual-optical-compute-fabric", "version": "0.1.0", "withheld_claim_count": 0 }, "date": "2026-08-01", "diagrams": { "diagrams": [ { "brief": "One host maintains GPU-native Surfaces; Operators receive bounded Lenses and return proposed writes through an Epoch commit gate.", "format": "svg", "id": "P30-F01", "path": "/diagrams/p30-virtual-optical-compute-fabric/P30-F01.svg", "status": "source-generated conceptual architecture", "title": "From virtual screens to computational surfaces" }, { "brief": "A canonical Surface supports global, dual, quad, tiled, and thousands of sparse Lenses without requiring full copied screens.", "format": "svg", "id": "P30-F02", "path": "/diagrams/p30-virtual-optical-compute-fabric/P30-F02.svg", "status": "source-generated conceptual architecture", "title": "One surface, many views" }, { "brief": "Epoch state is observed through Lenses, transformed by Operators, resolved as proposals, and advanced through a receipted commit.", "format": "svg", "id": "P30-F03", "path": "/diagrams/p30-virtual-optical-compute-fabric/P30-F03.svg", "status": "source-generated conceptual architecture", "title": "Observation and commit loop" }, { "brief": "Several virtual compute topologies coexist within one host and may later bridge across process, network, or physical optical boundaries.", "format": "svg", "id": "P30-F04", "path": "/diagrams/p30-virtual-optical-compute-fabric/P30-F04.svg", "status": "source-generated conceptual architecture", "title": "Recursive instances and transport bridges" } ], "paper": "p30-virtual-optical-compute-fabric" }, "evidence_state": "SPECIFIED", "id": "p30-virtual-optical-compute-fabric", "institution": "Glyphd Labs", "integrity": { "claim_ledger_sha256": "4f3459557b2a6daa54ce7a646a671bc8089337313f4e6cc2e0397cd44b4f8465", "paper_mdx_sha256": "eb4787f637ba4e154e29a3d19f61d031485159b8e86c9f59108fbfd17be0a452" }, "public_disclosure_boundary": "Publicly discloses the Virtual Optical Compute Fabric concept, terminology, formal observation model, runtime objects, authority boundaries, resource model, relationship to prior architectures, first experiment, falsification criteria, and public-safe integration with existing Glyphd research programs.\n\nDoes not disclose private implementation code, unpublished optimization heuristics, restricted source materials, credentials, private commercial plans, founder-controlled legal analysis, or patent claims. No working runtime, benchmark result, efficiency gain, compression result, photonic acceleration, safety certification, production deployment, or market adoption is claimed. “Optical” names an observation-based computational topology and optional transport bridge; it does not assert that the single-host architecture performs computation with physical light.", "publication_class": "public-safe working paper", "publication_url": "https://www.glyphd.com/papers/p30-virtual-optical-compute-fabric", "revisions": { "history": [ { "author": "Hayden Lindley / Glyphd Labs; edited by GPT-5.6 Pro", "date": "2026-08-01", "summary": "Initial public-safe working paper preserving the founder concept as a formal architecture: GPU-native Surfaces, programmable multiscale Lenses, Operators, recursive Instances, Epoch commits, transport Bridges, authority boundaries, resource model, and VOCF-1 experiment.", "version": "0.1.0" } ], "paper": "p30-virtual-optical-compute-fabric" }, "schema_version": "1.0.0", "slug": "p30-virtual-optical-compute-fabric", "sources": { "notes": "Primary API specifications and established dataflow and blackboard literature support the implementation substrate and historical comparison. The Virtual Optical Compute Fabric composition, performance expectations, transport continuity, and human-inspection value remain specified or hypothetical as marked in the claim ledger.", "paper": "p30-virtual-optical-compute-fabric", "sources": [ { "date": "2026-08-01", "id": "P30-S01", "path": "Glyphd Labs authoring conversation, public-safe synthesis", "title": "Founder concept record: virtualize the optical environment inside one computer" }, { "id": "P30-S02", "title": "Decimen Optical Transfer", "url": "https://github.com/bashalarmistalt/decimen-optical-transfer/" }, { "date": "2026-08-01", "id": "P30-S03", "title": "Streaming QR codes at 60 FPS achieves nearly 190 KB/s data rate in phone-to-phone tests", "url": "https://www.tomshardware.com/networking/streaming-qr-codes-at-60-fps-achieves-nearly-190-kb-s-data-rate-in-phone-to-phone-tests-browser-based-method-requires-no-app-no-networking-no-pairing-and-no-permissions-beyond-camera-access" }, { "date": "2026-07-17", "id": "P30-S04", "title": "Vulkan 1.4 Specification", "url": "https://registry.khronos.org/vulkan/specs/latest/html/vkspec.html" }, { "id": "P30-S05", "title": "WebGPU Specification", "url": "https://gpuweb.github.io/gpuweb/" }, { "id": "P30-S06", "title": "CUDA Programming Guide", "url": "https://docs.nvidia.com/cuda/cuda-programming-guide/" }, { "date": "2016-05-27", "id": "P30-S07", "title": "TensorFlow: A System for Large-Scale Machine Learning", "url": "https://arxiv.org/abs/1605.08695" }, { "date": "1974-01-01", "id": "P30-S08", "title": "The Semantics of a Simple Language for Parallel Programming", "url": "https://dblp.org/rec/conf/ifip/Kahn74" }, { "date": "1986-01-01", "id": "P30-S09", "title": "The Blackboard Model of Problem Solving and the Evolution of Blackboard Architectures", "url": "https://doi.org/10.1609/aimag.v7i2.537" }, { "date": "1986-07-30", "id": "P30-S10", "title": "Blackboard Application Systems, Blackboard Systems and a Knowledge Engineering Perspective", "url": "https://doi.org/10.1609/aimag.v7i3.550" }, { "date": "2026-07-27", "id": "P30-S11", "path": "content/papers/p02-fovea-address-fabric/paper.mdx", "title": "FOVEA Address Fabric" }, { "date": "2026-07-27", "id": "P30-S12", "path": "content/papers/p04-surfacemind/paper.mdx", "title": "SurfaceMind" }, { "date": "2026-07-27", "id": "P30-S13", "path": "content/papers/p09-cas-native-agent-coordination/paper.mdx", "title": "Agents That Work on Addressed State" }, { "date": "2026-07-27", "id": "P30-S14", "path": "content/papers/p06-actor-before-agent/paper.mdx", "title": "Actor Before Agent" }, { "date": "2026-07-27", "id": "P30-S15", "path": "content/papers/p21-mce1-human-in-master-control/paper.mdx", "title": "MCE-1: Human in Master Control" }, { "date": "2021-05-05", "id": "P30-S16", "title": "TENET: A Framework for Modeling Tensor Dataflow Based on Relation-centric Notation", "url": "https://arxiv.org/abs/2105.01892" } ], "version": "0.1.0" }, "subtitle": "The Virtual Optical Compute Fabric: multiscale field communication among computational systems", "title": "Computation by Observation", "version": "0.1.0" }