Free PC benchmark — full system sweep

idle OBS overlay Report card
RIG SCORE
Rig Score
CPU 30% · GPU 40% · storage 20% · memory 10% 1000 = the anchor rig (the bright tick) · log scale 250–4000

Run the full sweep to see your PC rated in plain English.

Processor
Run the sweep to rate your CPU.
Graphics
Run the sweep to rate your GPU.
Internet
Run the sweep to rate your connection.
Storage
Run the sweep to rate your drive.

Want every number and what it means?

System profile

reading hardware…

Video decode

checking…
The browser's own prediction of whether each codec plays smoothly and on dedicated hardware (power-efficient) at each resolution. Then optionally play a real clip to measure dropped frames.
plays locally · nothing is uploaded
CPU
MOps/s
Single-thread
Multi-thread
Scaling
Peak
SHA-256 MB/s gzip MB/s SubtleCrypto · Compression Streams
SIMD GFLOPS RAM GB/s WASM f32x4 · 64 MB DRAM read
What this means. Single-core speed drives most games; multi-core helps editing, compiling and heavy multitasking. These are JavaScript scores — fair to compare across browsers, but not against native benchmark tools.
GPU
FPS
Avg FPS
Min (1% low)
Frames
Render scale
GFLOPS WebGPU · modern compute path
FPS WebCodecs · 1080p · hardware where available
What this means. Average FPS is the headline; the 1% low is the honest one — how rough the worst moments feel. TFLOPS is raw compute. Browsers can't read GPU temperature or true utilisation.
Download
Mbps
Peak
Avg
Data moved
Latency
What this means. Speed (Mbps) wins downloads, but latency decides how games and calls feel. This is browser-measured throughput to a nearby edge server, not your raw ISP line.
Upload
Mbps
Peak
Avg
Data sent
Endpointcloudflare
What this means. Upload matters for streaming, video calls and cloud backups. Measured to your own server, with a Cloudflare fallback.
Storage · SSD sequential read / write
File size
MB/s write (peak)
MB/s read (peak)
Sustained write
Sustained read
Test file1 GB
BackendOPFS
What this means. Sequential speed looks big on a spec sheet; real load times track random speed. Browsers can't bypass the OS cache, so treat read numbers as a floor — a larger test file gives a truer figure.