Scientists create ultra fast memory using light
by giuliomagnifico on 12/4/2025, 6:11:21 PM
https://www.isi.edu/news/81186/scientists-create-ultra-fast-memory-using-light/
Comments
by: cycomanic
People have done these sort of "optical computing" based demonstrations for decades, despite David Miller showing that fundamentally digital computing with optical photons will be immensely power hungry (I say digital here, because there are some applications where analog computing can make sense, but it almost never relies memory for bits).<p>Specifically this paper is based on simulations, and I've only skimmed the paper, but the power efficiency numbers sound great because they say 40 GHz read/write speeds, but these consume comparatively large powers even if not reading or writing (the lasers have to be running constantly). I also think they did not include the contributions of the modulation and the required drivers (typically you need quite large voltages)? Somebody already pointed out that the size of these is massive, and that's again fundamental.<p>As someone working in the broad field, I really wish people would stop these type of publications. While these numbers might sound impressive at a first glance, they really are completely unrealistic. There are lots of legitimate applications of optics and photonics, we don't need to resort to this sort of stuff.
12/10/2025, 9:50:25 PM
by: adrian_b
Free version of the research paper:<p><a href="https://arxiv.org/abs/2503.19544v1" rel="nofollow">https://arxiv.org/abs/2503.19544v1</a><p>The memory cell is huge in comparison with semiconductor memories, but it is very fast, with a 40 GHz read/write speed.<p>There are important applications for a very high speed small memory, e.g. for digital signal processing in radars and other such devices, but this will never replace a general-purpose computer memory, where much higher bit densities are needed.
12/10/2025, 9:58:51 PM
by: ilaksh
MRAM and MRAM-CIM is like 10 years ahead of this and going to make a huge impact on efficiency and performance in the next few years, right? Or so I thought I heard.<p>Memristors are also probably coming after MRAM-CIM and before photonic computing.
12/10/2025, 9:36:57 PM
by: cs702
Cool. Memory bandwidth is a major bottleneck for many important applications today, including AI. Maybe this kind of memory "at the speed of light" can help alleviate the bottleneck?<p>For a second, I thought the headline was copied & pasted from the hallucinated 10-years-from-now HN frontpage that recently made the HN front page:<p><a href="https://news.ycombinator.com/item?id=46205632">https://news.ycombinator.com/item?id=46205632</a>
12/10/2025, 9:18:41 PM
by: lebuffon
Wow 300mm chips. They must be huge!<p>(I am sure they meant nm, but nobody is checking the AI output)
12/10/2025, 8:58:42 PM
by: xienze
This just in, OpenAI has already committed to buying the entire world’s supply once it becomes available.
12/10/2025, 9:17:52 PM
by: moi2388
“ This represents more than a laboratory proof-of-concept; it’s a functional component manufactured using industry-standard processes.”<p>Nice AI text again
12/11/2025, 7:11:45 AM
by: dahngue
[dead]
12/11/2025, 7:33:14 AM