AI metasurface startup @neurophos has test silicon - showing the final AI inference accelerator will reach close to 300 TOPS/W (4-bit). CEO Patrick Bowen gave me an update on which software startup the company has acquired and its new data center partner:
https://t.co/7q5ubfqrKG
We’re attending Hot Chips 2025 from Aug 24–27 with @PatrickTBowen · Hod Finkelstein · @DonovanPopps · @preston_woo · Carl Ramey · Michael Shebanow
DM us if you’re also attending.
@neurophos OPU: laser-powered million-TFLOPS inference chips. 100× the GPU token rate with zero extra power.
#HotChips2025 #OpticalCompute #AIHardware
At Embedded World 2025, we caught up with Coby Hanoch, CEO of Weebit Nano, in this exclusive video where he discusses how Weebit Nano’s advanced ReRAM memory technology is driving the next gen of intelligent devices. 🚀
Video here: https://t.co/sgXRlEv5hq
#EmbeddedWorld2025
Hey @ChipotleUK since when can you not have veggies instead of rice? Just been told they are only for vegetarians on Charing Cross Rd. New rule or have I just been lucky in the past?
So when looking at the NVIDIA provided benchmark results, pay close attention to the settings they are pushing. A lot of the big boosts are DLSS 4 vs 3, and the multi-frame generation (3 generated frames vs. older 1 frame). 3rd party reviews can't come fast enough.
This could be an incredible revolution in Cosmology.
The Dark Energy model of the universe, which won a Nobel Prize in 2011, may be completely wrong.
The accelerating expansion instead is simply because time runs faster in the voids between galaxies.
Let me explain: