Resources
Synoros’s research, technical writing, and engineering references. Published material, ongoing drafts, and reference documents.
Papers
Concise publication version
31 minAfter the universe
A research direction for substrate-faithful representations of meaning.
The paper as it appears for technical readers. Three moves: empirical convergence across LLM embeddings, quantum computation, and neural population geometry; containment over analogy; a single Lorentzian substrate viewed through four geometric tiers, with falsification criteria.
Read the publication versionFull version
~28k wordsAfter the universe
A structural argument for substrate-faithful representation of meaning.
Eight sections, ~28,000 words. Full physics setup from special relativity through hyperbolic spacetime, expanded empirical synthesis across four programs, the architecture sketch, and the closing argument that the building is worth doing.
Read the full paperIn development
Language model textbook
In developmentLong-form pedagogical text on how language models work, where the abstractions break, and how to deploy them against production workloads.
Substrate prototype
In developmentEngineering work on the four-tier architecture proposed in the paper. First milestone replicates the Nickel and Kiela hyperbolic WordNet result.
Engineering notes
All notesQuantization Explained: VRAM, Quality, and the RAM Performance Cliff
What quantization does to model quality and speed, how Q4/Q5/Q8 compare on real hardware, and why running a model split between GPU memory and system RAM can be 5-30x slower. Third-party benchmarks, practical recommendations.
Qwen 3.5 Benchmarks: 0.8B to 14B on RDNA4 Vulkan
Qwen 3.5 benchmarks across 80 prompts on AMD RDNA4 16GB with Vulkan. Unconstrained (2048-token) results show qwen3:14b leading at 0.594, with 9B pulling ahead of 4B when not artificially truncated. Includes analysis of how token budget constraints distort benchmark rankings.
LLM Benchmark Comparison: 12-Source Rankings
Cross-referenced LLM benchmark data from 12+ sources. Consensus rankings for coding, math, reasoning, tool calling. Interactive charts, cost vs. quality scatter plots, and full methodology.
How We Price: Aggregate Component Costs by Scrape
We publish the exact component costs, eBay search queries, retail sources, and markup formula for every Foundry GPU server. Prices are scraped from live market data and fed into an auto-pricing engine. This is the full breakdown.
ROCm HIP Idle GPU Bug on RDNA4: Vulkan Workaround
When loading Qwen 3/3.5 models via Ollama using ROCm/HIP on AMD RDNA4 GPUs, the GPU becomes permanently stuck at 100% utilization. We trace the root cause to HSA runtime teardown and benchmark the Vulkan backend as a drop-in fix with comparable performance.