Goals

Honeycomb is trying to be a general, checkable processor architecture for machine-learning workloads whose bottleneck is data movement. It is not a claim that every workload gets faster. It is a narrower claim: if weights and streams can live near the datapath, and if movement is explicit enough to schedule, then latency-bound inference should improve while the machine remains small enough to verify.

There is no Honeycomb silicon. The current evidence is a mix of Lean theorems, generated RTL, golden simulations, analytical models, and recorded open-source physical-design runs.

What must be true

Honeycomb succeeds only if these statements hold together:

The approach page explains the architecture choice behind those goals: one repeated cell, explicit movement, and distance treated as cost.

Evidence standard

Public claims use the book’s evidence categories:

The book is deliberately conservative about boundaries. It proves the model and several RTL-generation paths; it does not yet prove parser-level SystemVerilog equivalence, signoff-clean GDS, power, or fabricated silicon.

Current state

The repository already contains:

Binding the cell memories as SRAM macros is a central result. The inferred-memory version mapped to 40,994 flip-flops; the macro-backed version maps to 802 flip-flops plus five SRAM macros. That is the difference between a proof/simulation boundary and a plausible physical cell.

Performance envelope

The public performance stance is same-node and workload-specific. Honeycomb targets latency- and memory-bound regimes first, not peak FLOP leadership.

The prefill and batch-1 / batch-32 decode table on the home page is generated from the book’s recorded constants. The longer-horizon rows are:

WorkloadGPU baselineHoneycomb targetTag
Long-context decodeKV movement dominates above long contextswin target: resident KV and SRAM bandwidth reduce off-chip traffic[modeled]
Training, per devicedense GEMM remains strong on GPUsbehind target: Honeycomb trades compute density for memory locality[modeled]
Training, at scalecluster communication limits scalingparity-to-ahead target if in-fabric reduce replaces off-chip all-reduce[modeled / target]
Any checked workloadtesting onlyexact semantics and proof-carrying artifacts where the model is closed[proved]

The honest summary: Honeycomb is expected to win where memory traffic dominates latency, be comparable on some compute-bound inference at the same node, trail GPUs on raw floating-point density, and offer stronger correctness evidence than conventional accelerator flows.

Training

Training is a harder claim. The forward pass benefits from resident weights, but backward passes have less reuse and more state. The current model treats training capacity as roughly 8x the inference footprint [modeled]. That is a sizing rule: more trainable capacity means more cells, not a magic change in per-cell SRAM ratio.

The strongest training argument is communication. A mesh can reduce gradients inside the fabric instead of round-tripping everything through off-chip links. That still leaves open work: low-precision gradient numerics need their own convergence study and their own arithmetic proofs.

Variants

Honeycomb is a parameterized cell generator, not one fixed point. Different chips can choose different memory sizes, MAC counts, widths, and numeric options while keeping the same ISA and proof structure.

Within a chip, the goal remains one uniform cell. Across chips, parameters may change only if the proofs re-close under those parameters. A hand-written RTL fork or separate semantics would break the core claim.

Near-term milestones

Non-goals

Honeycomb is not pursuing cache coherence, virtual memory, implicit remote loads, in-place JIT, a distinct conductor core, service tiles, analog compute-in-memory, FPGA-reconfigurable cells, a full OS, POSIX, or measured-silicon claims before silicon exists. Parser-level SystemVerilog refinement is a documented boundary, not a current proof.