A structured documentation hub for GPU teardowns, component identification, and salvage workflows — building the informational layer that repair culture runs on.
The gap is informational, not technical.
The skills exist. The documentation does not.
Chinese compute repair shops operate with a hardware fluency that is almost entirely absent in the US maker and repair community — doing component-level GPU teardown and salvage at scale. The knowledge base that makes that possible was built during ordinary times. Decap is the attempt to build the documentation layer that makes that fluency transferable.
Each GPU teardown produces a structured record: board layout, component maps, part identifications, reuse notes. Over time these records become a lookup library indexed by model.
What a component is, where it lives, what it does, and what the common alternates are. Cross-referenced by part type across GPU families — building a shared vocabulary for the field.
Desoldering, rework, component handling, and salvage workflows documented with enough specificity to be actually useful to someone working with the hardware.
How to acquire e-waste GPU hardware, what to look for, and how to assess a board before investing time in a salvage workflow — the front end of the pipeline.
How salvaged components find second-life applications. The bridge between salvage and fabrication culture — turning e-waste into a usable materials library.
Scope stated as a design decision, not a legal disclaimer. The knowledge being documented is the pre-modification layer: what things are and how they work, not how to exploit or extend them commercially.
GPU hardware represents some of the most complex and densely specified silicon in consumer electronics — and almost none of that documentation exists at the component salvage level. Board schematics are proprietary. Repair documentation is nearly nonexistent. The knowledge lives in specialist communities and repair shops, largely inaccessible.
E-waste volumes are accelerating. GPUs have a short useful life in compute contexts but contain components with much longer useful lives if extracted and reused. The bottleneck is not the hardware — it is the documentation needed to work with it confidently.
A repair culture that lives in YouTube videos and Discord servers is fragile. The knowledge exists but it is hard to find, inconsistently structured, and often tied to individual expertise rather than being genuinely open. Structured documentation — indexed, cross-referenced, stable — is a different kind of resource. It persists. It can be found when needed. It does not require knowing the right person.
Electronics interest but limited GPU-specific knowledge. A structured starting point that currently doesn't exist.
Looking for component-level reference material. The documentation that schematic NDAs currently prevent from existing openly.
Wanting to understand what they're handling. Better decisions about what to salvage, recycle, and what has reuse potential.
Curious about internals. Interested in the frankenstein culture of building with salvaged components. Documentation as entry point.
Building curricula around hardware literacy and sustainability. Structured, citable reference instead of scattered forum posts.
Decap is in early planning. No teardowns exist yet. No wiki. The documentation infrastructure is the first thing to build — the format, the indexing structure, the vocabulary — before the hardware work begins.
Decap sits alongside Chamfer (open household mechanical knowledge) and Entisol (compression research), sharing the same documentation-first, open-knowledge ethos. The common thread: reclaiming specialist knowledge that exists but isn't publicly accessible in structured form.
Phase 1 looks like an e-waste education site. Phase 2 reveals it was always infrastructure.