←back to thread

Sustainable memristors from shiitake mycelium for high-frequency bioelectronics

(journals.plos.org)
144 points PaulHoule | 8 comments | 31 Oct 25 13:32 UTC | HN request time: 0s | source | bottom
Show context
embedding-shape ◴[31 Oct 25 14:43 UTC] No.45772497[source]▶
> We demonstrate fungal computing via mycelial networks interfaced with electrodes, showing that fungal memristors can be grown, trained, and preserved through dehydration, retaining functionality at frequencies up to 5.85 kHz, with an accuracy of 90 ± 1%. Notably, shiitake has exhibited radiation resistance, suggesting its viability for aerospace applications

Soon we'll have shiitake replacing transistors in our airplane and spacecraft computers, while sitting and eating ramen on the vehicles themselves. The future is shaping up to be interesting.

replies(8): >>45772688 #>>45772891 #>>45773077 #>>45773703 #>>45774592 #>>45774669 #>>45777416 #>>45778848 #
1. zdragnar ◴[31 Oct 25 15:11 UTC] No.45772891[source]▶
Having only dabbled the slightest in hardware... are functional frequencies topping out at 6 kHz useful for memristors in modern computing? I feel like having separate components each magnitudes faster would be better than combining them into a memristor that sounds so slow.
replies(4): >>45774439 #>>45777500 #>>45777673 #>>45777713 #
2. xeonmc ◴[31 Oct 25 17:22 UTC] No.45774439[source]▶
If it enables massively concurrent in-memory compute then the frequency disadvantage could just be scaled away.
replies(1): >>45780841 #
3. m4rtink ◴[31 Oct 25 22:40 UTC] No.45777500[source]▶
Might be enough for microcontrollers and overall simple control applications?
4. bee_rider ◴[31 Oct 25 23:02 UTC] No.45777673[source]▶
A bunch of mushrooms as a giant 6kHz memory array with in-memory computing seems pretty pointless IRL. But it adds a nice air of plausibility to some sci-fi stories!
5. estimator7292 ◴[31 Oct 25 23:08 UTC] No.45777713[source]▶
If you mean x86 class performance, no, obviously not.

But 6kHz is not nothing. For application-specific computers, you can do a lot with very little. You aren't going to be building high performance general purpose computers, but for an atonomous circuit quietly ticking away computing orbital trajectories or stellar navigation, you don't need modern x86 class performance.

replies(1): >>45778911 #
6. ◴[01 Nov 25 03:01 UTC] No.45778911[source]▶
7. makapuf ◴[01 Nov 25 11:25 UTC] No.45780841[source]▶
IF you don't care about the issue of moving information, and provided your problem is embarassingly parallel.
replies(1): >>45791707 #
8. avmich ◴[02 Nov 25 17:01 UTC] No.45791707{3}[source]▶
Fortunately, a lot of problems are embarassingly parallel.