Terabyte per second to Kilobit per second

Large language models have billions of parameters that must be read from memory for every inference pass. An LLM with 70 billion parameters at 16-bit precision needs 140 GB of data read per forward pass. At 3 TB/s, the H100 can perform roughly 20 inference passes per second — bandwidth directly determines tokens-per-second output.

During LLM inference each token requires reading all model weights from memory. A 70-billion-parameter model at 16-bit precision means 140 GB read per forward pass. At 30 tokens per second, that is 4.2 TB/s of memory reads — right at the limit of an H100's HBM3. This is why AI inference is "memory-bound": the GPU's compute cores sit idle waiting for data. Quantising weights to 8-bit or 4-bit halves or quarters the bandwidth demand, directly increasing tokens per second.

The NVIDIA B200 GPU with HBM3e achieves approximately 8 TB/s of memory bandwidth as of 2025. Each generation roughly doubles bandwidth — from 2 TB/s (A100) to 3.35 TB/s (H100) to 4.8 TB/s (H200) to 8 TB/s (B200). The trajectory suggests 16+ TB/s within a few years.

About 16.7 minutes. A petabyte is 1,000 terabytes, so at 1 TB/s, the math is simple division. For context, the Library of Congress contains roughly 10–20 petabytes of data. Transferring it all at 1 TB/s would take about 3–6 hours.

Yes — petabytes per second (PB/s). Experimental optical interconnects and photonic computing architectures are pushing toward PB/s-class bandwidth. Some supercomputer storage systems already aggregate into the PB/s range when all nodes operate simultaneously. It is the next frontier for AI training clusters.

Audio codecs compress sound into a stream of bits played back in real time, so the natural unit is bits per second. At 128 kbps, an MP3 encoder allocates 128,000 bits to represent each second of audio. Higher kbps means more data per second, better quality, and larger files.

Technically yes — dial-up ISPs like NetZero still exist in the US, and some rural areas with no broadband rely on them. But at 56 kbps, loading a modern webpage (average 2.5 MB) would take over 5 minutes. It is functionally unusable for anything beyond basic email.

At 128 kbps, the encoder discards more audio detail — cymbals sound washy, stereo imaging narrows, and quiet passages lose nuance. At 320 kbps, most listeners cannot distinguish the MP3 from the original CD in blind tests. The file is 2.5× larger but audibly transparent to most ears.

A standard VoIP call uses 8–64 kbps depending on the codec. The widely used Opus codec delivers excellent voice quality at 16–32 kbps. Traditional landline phone calls used 64 kbps (G.711 codec). HD Voice on modern smartphones uses about 32 kbps with the AMR-WB codec.

The screeching was the modem handshake — two modems negotiating their connection speed by exchanging test tones over the phone line. Each phase of the screech tested different frequencies and protocols. The modems were literally talking to each other in audio, finding the fastest kbps rate the line could support.