Definition
TOPS, or Tera Operations Per Second, is the headline throughput rating stamped on neural processing units (NPUs), edge AI accelerators, and the AI engines built into modern systems-on-chip. It counts how many trillions of integer operations the silicon can perform every second, where one "operation" is typically a single multiply-accumulate (MAC) step inside a matrix multiplication. Almost every published TOPS figure is measured at INT8 precision, because 8-bit integer math packs the most operations through a fixed amount of hardware.
How TOPS Is Calculated
Peak TOPS comes straight from the hardware design: roughly 2 x (number of MAC units) x (clock frequency in Hz), divided by a trillion. Because a multiply-accumulate is counted as two operations, the factor of two appears in the formula. This is a theoretical ceiling under ideal conditions, not a number you will reach running a real model with memory stalls and imperfect utilization.
Why Precision Matters
A TOPS rating is meaningless without its precision. The same chip rated at 40 TOPS for INT8 might deliver only about 20 for FP16, because lower-bit formats let more operations flow through the same datapath. When sizing hardware for a local model, always compare TOPS at the precision you actually intend to run, and remember that memory bandwidth often limits real inference far more than raw TOPS.
For sovereign Bitcoiners running models on their own hardware, TOPS helps you judge whether an NPU or edge board can handle a given workload before you buy. Pair this with our entries on FP16/INT8 precision and Local LLM to size your air-gapped inference rig correctly.
In Simple Terms
TOPS, or Tera Operations Per Second, is the headline throughput rating stamped on neural processing units (NPUs), edge AI accelerators, and the AI engines built…