FLOPS Calculator
The theoretical peak FLOPS of a CPU from its cores, clock and vector width — and how long a workload of a given size takes at a realistic efficiency.
Peak FLOPS is cores × clock × floating-point operations per cycle per core.
How the flops calculator works
Peak FLOPS is cores × clock × floating-point operations per cycle per core. A modern core with two 256-bit AVX2 fused-multiply-add units does 16 double-precision (32 single-precision) operations a cycle; AVX-512 doubles that; a scalar core does two. Real code reaches 10–70% of peak depending on how well it vectorises and how memory-bound it is, so the time row divides by an efficiency you choose.
Formula: peak FLOPS = sockets × cores × clock (Hz) × FLOPs per cycle; time = work ÷ (peak × efficiency)
Worked examples
| Inputs | Peak performance | Note |
|---|---|---|
| 16 cores at 3 GHz, AVX2 with two FMA units | 768 GFLOPS | 768 GFLOPS; 100 TFLOP in 7.2 min |
| Dual-socket 64-core AVX-512 server | 10.24 TFLOPS | 10.24 TFLOPS |
| A scalar microcontroller at 200 MHz | 0.4 GFLOPS | 0.4 GFLOPS |
FAQFrequently asked questions
Why is my measured performance so far below peak?
Peak assumes every core issues a full-width fused multiply-add every cycle with data already in registers. Real code waits on memory, branches, and uses scalar or partially filled vectors; 30–50% is good for dense linear algebra, under 10% for pointer-chasing code.
How many FLOPs per cycle does my CPU do?
Two per FMA unit per lane: AVX2 has 4 double lanes × 2 = 8 per unit, and most cores have two units (16); AVX-512 has 8 lanes × 2 × 2 units = 32 double or 64 single. Some chips run AVX-512 at a lower clock.
What about GPUs?
The same arithmetic with thousands of lanes: a data-centre GPU peaks at tens of TFLOPS double precision and hundreds in reduced precision. Enter the GPU’s core count and clock with the right width, or use the vendor’s stated peak.
Is a GFLOP a billion operations?
Yes — 10⁹ floating-point operations, counting a fused multiply-add as two. The LINPACK benchmark behind the TOP500 counts them the same way.
Where these figures come from
- IEC 80000-13 — Information science and technology (quantities and units) — the decimal (kB, MB) versus binary (KiB, MiB) prefixes used throughout
- RFC 4632 — Classless Inter-domain Routing (CIDR) — the address-plan arithmetic behind the subnet calculator
- RFC 1918 — Address Allocation for Private Internets — the private ranges the subnet calculator recognises
- NIST SP 800-63B — Digital Identity Guidelines, Authentication — length over composition rules; the basis of the password guidance here
- NIST SP 800-57 Part 1 — Recommendation for Key Management — key-strength comparisons used by the key-space calculator
- Australian Cyber Security Centre — Essential Eight — national guidance on passphrases and system hardening
Last checked: September 2026. Units follow the SI decimal convention (IEC 80000-13 defines the binary alternatives); network and security figures cite the defining standard.