Taalas uses a design that requires masked ROM layers specific to a model, but Etched uses a newer design with a specialized RAM that supports any transformer model.
2 hours and no comments makes me wonder if the Etched chip exists or not.
> Sohu figures are per chip, derived from Etched's published 8-chip server claim of 500,000 tok/s on Llama 70B; not independently verified. (claimed by Etched)
So these are not verified benchmarks and they are all claims. Again is this chip real?
Can’t wait for single chip asic qwen3.8 27b
It’s small so should be cheap per chip. And it so much smarter than it ought to be for its size.
Problem is asic take forever to cut and are always months behind the latest open weight sota
Taalas uses a design that requires masked ROM layers specific to a model, but Etched uses a newer design with a specialized RAM that supports any transformer model.
2 hours and no comments makes me wonder if the Etched chip exists or not.
> Sohu figures are per chip, derived from Etched's published 8-chip server claim of 500,000 tok/s on Llama 70B; not independently verified. (claimed by Etched)
So these are not verified benchmarks and they are all claims. Again is this chip real?
Keep on eye on TechTechPotato (https://www.youtube.com/channel/UC1r0DG-KEPyqOeW6o79PByw) for updates. Ian has put enough pressure on other companies to get them to allow third-party verification.