binome · HackerTrans

el mes pasadodiscuss

Considering how good LLMs are getting at building DSP code now, I expect to see a tidal wave of new digital modes start to pop up now. It's easy to conduct over the air tests on HF using the global kiwisdr receiver network and build agentic experiment, test, evaluate loops on simulated and real RF channels.

binome

hace 5 mesesdiscuss

I vibe-trained moonshine-tiny on amateur radio morse code last weekend, and was surprised at the ~2% CER I was seeing in evals and over the air performance was pretty acceptable for a couple hour run on a 4090.

binome

hace 6 mesesdiscuss

I'm curious as to what actually is the CPU <-> NPU bandwidth in these whitebox OLTs? Traditionally that has been sized for small amounts of punted control plane packets, then programming a fast path into the NPU for revenue traffic.

binome

hace 6 mesesdiscuss

This doesn't look like anything malicious, 8048 is just prepending these announcements to 52320.. If anything, it looks like 269832(MDS) had a couple hits to their tier 1 peers which caused these prepended announcements to become more visible to collectors.

binome

hace 8 mesesdiscuss

Interesting take, but generally large, incumbent eyeball networks have refrained from open peering at IX's for decades at this point. They maintained presences, but usually just to grab a few specific peers they wanted, not to peer broadly with everyone across the exchange, and the bulk of traffic from large providers into eyeball networks comes across PNIs or on-net CDN nodes, not IX.

If anything, this move to centralized PNIaaS platforms makes interconnecting with the eyeball networks even easier for smaller providers. The portals allow for straightforward visibility on what they want to charge for paid peering, and instant automated EVCs and turnup, shortcutting the long and windy process of negotiating terms and establishing individual XCs in DCs that you agree to peer in.

comments