Creating a PC Builder with AI support: need detailed hardware datasets or APIs?
Creating a PC Builder with AI support: need detailed hardware datasets or APIs?
Hello everyone,
I’m a final-year student in Computer Engineering working on my senior capstone project. My aim is to design a PC Configuration Recommendation System. The objective is to suggest the best PC builds tailored to a user’s budget and intended use—such as gaming or rendering/workstation—while balancing price, performance, and compatibility. To achieve this, I need to assign a numerical "performance score" to components like CPUs, GPUs, and others, ensuring accurate ranking. I also require a solid dataset.
I’m not content with just generic metrics; I want precise technical specifications and performance data. My concern is that using a generic synthetic score (like PassMark or basic clock speeds) could lead to misleading rankings. For example, an AMD Ryzen X3 chip might score lower than a high-end Intel processor due to architectural features like 3D V-Cache, which significantly impacts gaming performance. A simple lookup won’t reflect these nuances.
I’m seeking structured data formats—CSV, JSON, SQL dumps, or API access—and databases that capture these architectural details better than standard tests. I’m also questioning the reliability of many benchmark websites and asking which sources are considered the "Gold Standard" for raw performance metrics.
Beyond FPS or Cinebench numbers, what other technical indicators should I consider to build a realistic performance formula?
I’d appreciate guidance on finding trustworthy, structured data sources to develop an effective scoring algorithm.
Thank you in advance for your assistance!
Benchmark tools tailored to specific operations are used, though even these can only provide rough estimates due to the difficulty of replicating identical environments. Published outcomes are often approximate and influenced by varying tester expertise. There is no single standard metric for comparison since each system has unique software needs.
I'm aware of what you mentioned. Thank you for feedback. Still I need a least bad source to work on it.
You are absolutely right. A spec sheet is not performance and what I mean by data actually is not about just spec sheet.
I consider to use spec sheets to define components in the database and I also consider bunch of other metrics like user benchmarks, tests etc... to assign a performance score to those components. This is one of the hardest tasks for a gaming pc builder. It has to be a great algorithm for performing components.
Hardware selections are arbitrary, and there is no such thing as a perfect dataset. However, when exact data isn't available, we frequently have to use heuristics and approximations in engineering.
A dataset to "do the work for me" is not what I'm searching for. In order for my algorithm (the engine) to filter out the worst 90% of combinations, I am searching for input data (fuel).
I'm here to ask where to find reliable data, tests, benchmarks.
Updating data is about scrapping the newest widgets' spec sheets and tests/benchmarks data (or whatever I could find by then). After the process of getting right data it is creating a performance score of the new widget.