A local MCP codebase-memory server can keep both retrieval and answer synthesis on your own machine, but a useful result depends on two things beyond choosing a language model: the quality of the indexed context and how many inference requests your GPU can handle at once. In his July 15, 2026 post-mortem, product engineer Enrique Bruzual describes a project built around Ollama and ChromaDB, compares two models on one Windows PC, and explains how a concurrency limit stopped local synthesis from overwhelming an 8 GB graphics card. His measurements are a small, hardware-specific case study—not a general performance ranking.
How the project divides retrieval and synthesis
Bruzual describes zerikai_memory as supporting cloud, local and hybrid modes. In local mode, ChromaDB retrieves structured code entities and Ollama performs synthesis. The model is given a project brief and retrieved material such as function signatures, file paths, line ranges and docstrings; its response can include inline #file:line citations. These are implementation details of this project, not a universal MCP specification.
As an Amazon Associate I earn from qualifying purchases.
The separation matters: retrieval supplies evidence, while the model turns that evidence into an answer. Bruzual characterizes the project’s ChromaDB retrieval—with L2 distance and lexical reranking—as model-agnostic. That lets the synthesis model be changed without changing the retrieved context, making a controlled model comparison possible.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →The broader combination of local code search and MCP is not unique to this implementation. For example, the independent codebase-semantics-mcp project documents stdio transport and Ollama embeddings. It is a separate repository with its own architecture, not another name for zerikai_memory.
#1 Best Overall
- [Local AI Inference & 70B Model Ready] Equipped with the AMD Ryzen 7 PRO 8845HS processor, NEXUS is engineered for heavy local AI workloads. With a full-size GPU bay, it runs 70B LLMs natively without an internet connection. Ideal for AI developers and tech enthusiasts who need private environment for coding and model testing.
- [132TB Mass Storage with ZFS Integrity] Features a hybrid storage architecture (3×NVMe + 4×3.5" HDD) supporting up to 132TB. Utilizing the enterprise-grade ZFS file system and ECC memory, it prevents data corruption and bit rot—a must-have for professional photographers and video editors safeguarding 4K/8K RAW footage.
- [OpenClaw-Driven Automation Workflow] The built-in OpenClaw execution layer allows complex automated tasks to be processed locally. Even when offline, your backup schedules and AI file organization continue seamlessly. Say goodbye to monthly cloud subscriptions and high latency.
- [Dual 10GbE & USB4 Ultra-Connectivity] Experience server-class speeds with dual 10GbE ports and a 40Gbps USB4 interface. It enables multi-user real-time collaboration on large project files directly from the NAS, ensuring zero-lag editing for creative studios and production teams.
- [Open-Source ZimaOS for Total Privacy] Running on the fully open-source ZimaOS, NEXUS ensures your data stays physically on-premise with no backdoors. It acts as a "Digital Fortress" for privacy-conscious families and small businesses who demand absolute data sovereignty.
What the model comparison does—and does not—show
For a latency test, Bruzual used static ChromaDB payload samples built from real workspace entities. The script made three queries with three samples per model and measured raw latency at the HTTP layer. The reported results were:
| Model | Mean latency | Standard deviation | Minimum | Maximum |
|---|---|---|---|---|
mistral:7b |
6.14 seconds | 3.58 seconds | 2.92 seconds | 14.57 seconds |
ornith:9b |
13.39 seconds | 5.76 seconds | 8.77 seconds | 25.67 seconds |
These are measurements reported by Bruzual in 2026 from a small test on his machine, not results from Ollama, ChromaDB or an independent benchmark organization. They measure request latency for sampled payloads, not end-to-end synthesis quality on a live codebase. The first ornith:9b request took 25.67 seconds as memory spilled into shared system memory before Ollama pinned the model; warmed samples were reported at 9–17 seconds. Bruzual reported that mistral:7b fit within the test machine’s 8 GB of dedicated VRAM and took 3–7 seconds when warm. Those observations are specific to his hardware and test conditions.
Rank #2
- Next-Gen Processing Power: Powered by the AMD Ryzen 7 8845HS processor (8 Cores, 16 Threads, Zen 4 architecture) and Radeon 780M graphics. Effortlessly handles fluid 4K/8K real-time media transcoding, multiple operating system virtualizations (PVE/ESXi), and simultaneous background tasks without a stutter.
- Secure Local AI & Privacy: Features an integrated Ryzen AI NPU delivering up to 38 TOPS of total processing power. Deploy 8B/14B Large Language Models (LLM) locally, run automated programming assistants, and enjoy lightning-fast AI photo recognition—all completely offline, keeping your sensitive data 100% secure.
- Pro-Studio Collaboration: Engineered with dual 2.5GbE network ports and optimized high-speed architecture. Eliminate transmission bottlenecks so multiple video editors, photographers, or 3D designers can collaborate, render, and share heavy assets directly from the NAS in real time.
- Massive Docker Ecosystem: Seamlessly deploy and run over 20+ Docker containers simultaneously. Perfect for hosting your home assistant, private web servers, automated downloaders, and personal databases with enterprise-level stability.
- Futuristic Heat Dissipation: Designed with an advanced cooling system tailored for continuous, high-load hardware operation. Enjoy high-speed read and write speeds across multiple drive bays while maintaining whisper-quiet operation in your home or studio.
The article also describes a live MCP-layer test with five queries. Both models received the same retrieved context and system prompt. One important failure occurred when retrieval did not supply enough evidence: one model responded confidently with unsupported details, while the other acknowledged it could not determine the answer. File citations alone therefore are not proof of grounding. Check whether cited lines actually support the claims, and whether the model admits when retrieved context falls short.
Recommended Free Tools
A separate comparison of brief generation using cloud DeepSeek and local ornith was uncontrolled because docstring density differed between runs. It cannot establish that one model is generally better. Bruzual’s interpretation was that richer indexed context improved the generated brief: “The takeaway is not that ornith beats DeepSeek for brief generation. It is that embedding-docstring enrichment is visible and measurable in the output.”
Rank #3
- 【Local AI & LLM Powerhouse】 Fueled by the Ryzen 8845HS NPU and RTX 5070 GPU, this NAS is your private AI workstation. Effortlessly deploy local LLMs and run Stable Diffusion without costly cloud subscriptions. Enjoy 100% data privacy and absolute protection for your proprietary code and sensitive data.
- 【Studio-Grade Media Workflow】 Engineered for 4K/8K video editors and creative studios. Leveraging the RTX 5070's dual AV1 encoders, your team can edit RAW footage and render graphics directly on the NAS over 10Gbe. Eliminate transfer bottlenecks and streamline collaborative post-production.
- 【Advanced Virtualization Hub】 Power through heavy workloads with the 8-core, 16-thread Ryzen 8845HS and RTX 5070’s hardware virtualization capabilities. Smoothly run dozens of Docker containers, Windows/Linux VMs, or network services simultaneously. The ultimate all-in-one sandbox for full-stack developers and IT pros.
- 【Automated Smart Backup Workflow】 Streamline your data management with automated multi-device syncing across phones, cameras, and PCs. The built-in AI NPU automatically executes facial recognition, scene categorization, and smart tagging for media asset management, ensuring lightning-fast archiving via 10GbE.
- 【Secure Enterprise Private Cloud】 Build your company’s ultra-fast, encrypted private cloud for seamless remote collaboration. Team members worldwide can access projects, co-edit files, or preview heavy 3D assets in real-time. Fortified with financial-grade encryption to protect your corporate intellectual property.
Why the GPU saturated, and how concurrency was limited
On the tested Windows 11 computer, local deep-brief synthesis launched all nine brief-section tasks at once with asyncio.gather. Because there was no concurrency gate, the tasks could issue simultaneous Ollama requests. Bruzual reports that this saturated the GPU in the 8 GB configuration.
The reported fix was a global ollama_semaphore and a safe wrapper: local-mode calls acquire the semaphore, while cloud and hybrid calls bypass it. The setting OLLAMA_MAX_CONCURRENCY controls the limit; Bruzual reports a default of 1 for 8 GB hardware. In effect, the application queues local synthesis requests instead of allowing all nine to compete for constrained GPU memory at once. The exact code and configuration belong to his implementation; adapt the pattern to your own request paths and deployment modes.
What hardware this case study supports
Bruzual’s test system had Windows 11, an NVIDIA RTX 3050 with 8 GB dedicated GDDR6 VRAM, an Intel i7-12700 CPU and 32 GB RAM. He describes 8 GB of dedicated VRAM as the limit for that setup; when memory spilled into shared system memory over PCIe, inference slowed. This is a report about one configuration, not a minimum specification for every MCP server using Ollama and ChromaDB.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsFor this particular workflow, Bruzual recommends 10–12 GB of dedicated VRAM for more headroom and names the RTX 3060 12 GB as an example. He recommends mistral:7b for systems below 8 GB VRAM or when latency matters more than citation precision. These are his setup-specific recommendations, not a universal requirement or current graphics-card buying guide.
Best Value
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 64GB pool, which is perfect for running LLMs such as Deepseek 32B, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 4% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
How to evaluate the setup on your own codebase
A small latency comparison can help choose what to test next, but the decision should account for evidence quality and contention as well as speed. Keep the retrieval side constant when comparing synthesis models, and use the same prompts and queries so differences are interpretable.
- Measure cold and warm latency. Record first-request behavior separately from warmed runs; model loading and memory pressure can make them substantially different.
- Hold retrieved context constant. Compare models with the same entities and prompt, and keep index inputs—especially docstring coverage—consistent.
- Check citation accuracy. Confirm that each cited file and line exists and supports the associated claim; the presence of a citation is not enough.
- Test missing-context behavior. Include questions the index cannot answer and observe whether the model signals uncertainty or invents details.
- Stress concurrent synthesis deliberately. Watch VRAM use and latency under the number of simultaneous requests your application can launch, then set a local concurrency cap appropriate to the available memory.
The central engineering lesson in Bruzual’s account is that better synthesis cannot reliably compensate for missing source context, while unbounded local parallelism can turn a workable model into a slow or saturated workload. Retrieval enrichment and VRAM-aware request control address different failure modes; both need to be evaluated in the actual project.
Quick Recap
Sources
- Enrique Bruzual, “Post-Mortem: Building a Local MCP Server for Codebase Memory using Ollama and ChromaDB,” DEV Community, July 15, 2026.
- ChromaDB Go Client documentation on embedding models, which lists Ollama as an available embedding integration; this is technical context, not confirmation of the project’s exact implementation.
- The independent codebase-semantics-mcp repository, documenting another local MCP and Ollama code-search example.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
