● live2025CASE STUDY 01 / 07

FPS Calculator

Predicts real-world gaming FPS for any CPU + GPU combo across 50+ titles.

FastAPIPythonPostgreSQLReact 19Tailwind CSSDockerNginx

50

AAA titles

5

Languages

3

Resolutions

3

Upscalers modeled

fps.tech-latest.com
FPS Calculator screenshot

01 / The problem

A binary "will my PC run this" check tells you nothing useful — CPU/GPU balance shifts per game, per resolution, and per upscaler, not just per hardware tier.

02 / The approach

Built a calibrated Amdahl frame-time model with a per-game cpu_fraction (manually benchmarked or genre-defaulted, not guessed) that splits frame time between CPU and GPU work, then layers on resolution/preset multipliers and DLSS/FSR/XeSS-aware gating.

03 / The catch

The hardware scoring has to stay in lockstep with its sibling site, Bottleneck Calculator — both read the same benchmark API, so a scoring change on one can silently break predictions on the other.

Predicts gaming frame rates for any CPU + GPU combination across 50+ AAA titles using a calibrated Amdahl frame-time model at 1080p, 1440p, and 4K. Built with a FastAPI backend backed by PostgreSQL and Alembic migrations, a React 19 + Tailwind frontend with i18next localisation, and deployed via Coolify blue-green on Docker. Automated with GitHub Actions (pytest).

Highlights

Per-game cpu_fraction, not a generic genre default

VRAM penalty modeling for VRAM-limited scenarios (e.g. 4GB GPU at 4K Ultra)

AEO-optimized — FAQ schema and result prose written for AI search citation, not just Google

Admin dashboard for game CRUD, usage analytics, and CSV export

Gallery

Like what you see?

Let's build something together.