3 ms·

The mind boggles at what is being accomplished in semiconductor production at this scale. The smallest features in the (fin width, e.g.) in the 3nm process ar

The mind boggles at what is being accomplished in semiconductor production at this scale. The smallest features in the (fin width, e.g.) in the 3nm process are roughly 5nm across. That's the diameter of a hemoglobin molecule. Or, to put it another way, it's about 40 silicon atoms. All this enables single chip processors with over 100 billion transistors, so, more transistors on laptop processor than neurons in your brain. Wholesale, that may run close to 1 ten millionth of a cent per transistor. All this, and yet the process for producing transistors in a planar layout on silicon (that is, an integrated circuit) was invented after I was born.

bpye 4y ago
> All this enables single chip processors with over 100 billion transistors, so, more transistors on laptop processor than neurons in your brain. I was sceptical, but this is true! The human brain supposedly has ~86 billion neurons [0]. [0] - Table 2 - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2776484/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2776484/
- mgsouth 4y ago
  True, but neurons are way more complicated than a transistor. Even a synapse is more complicated and powerful. Neurons are more like microcontrollers.
  - stevenhuang 4y ago
    It gets even more complicated when it is hypothesized neurons use spike timings to "compute". > A potentially more powerful way to construct symbols with spikes is to use spike timing, i.e., to consider exactly when spikes occur within the time window. The maximum number of symbols a neuron can construct with spike timing is limited only by the temporal resolution of the code. In formal terms, the information capacity of spike timing is much higher than the information capacity of spike count. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6665236/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6665236/
  - walnutclosefarm 4y ago
    Way more complex to be sure. The "more powerful" part requires a bit of interpretation. One transistor certainly can't do the work of one neuron, even given forever to do it. But a not-too-large-to-count collection of transistors can do the work of a single neuron, and do it much faster. I don't know that 100 billion transistors can out compute 80 billion neurons with a 100 trillion synapses, but it's not implausible to me. Neurons have a firing rate on the order of 10Hz; an entire CPU can be clocked at 100 million times that rate. It's certainly not obvious that those 11 orders of magnitude can't make up for the functional limitations of transistors. Obviously standard microprocessor architectures are not comparable to the architecture of the brain, so a true head to head comparison is impossible, but it's not absurd to imagine we've achieved or surpassed parity.
    - tim333 4y ago
      These things are approximate but Hans Moravec made some reasonable estimates back in 1997 that human brain equivalence was about 100 teraflops https://jetpress.org/volume1/moravec.htm https://jetpress.org/volume1/moravec.htm We are around there now, see "NVIDIA GeForce RTX 4090 Is The First Gaming Graphics Card To Deliver 100 TFLOPs of Compute Performance" https://wccftech.com/nvidia-geforce-rtx-4090-is-the-first-gaming-graphics-card-to-deliver-100-tflops-of-compute-performance https://wccftech.com/nvidia-geforce-rtx-4090-is-the-first-ga...
      - walnutclosefarm 4y ago
        Thanks for the links. Confirming, I guess, my instinct that we're near parity in computing power between chips and brains.
Sephr 4y ago
You probably want to qualify "single chip processors" with "small". Single chip processors with over 100 billion transistors have been possible for years. We've had trillion-transistor single chip processors since 2019[1]. 1. https://www.cerebras.net/blog/introducing-the-cerebras-cs-1-the-industrys-fastest-artificial-intelligence-computer/ https://www.cerebras.net/blog/introducing-the-cerebras-cs-1-...
- walnutclosefarm 4y ago
  Yeah. I've never been able to think of wafer scale integration as "single chip." But your comment is fair. Won't fit in a laptop though ;-)

bpye 4y ago

> All this enables single chip processors with over 100 billion transistors, so, more transistors on laptop processor than neurons in your brain. I was sceptical, but this is true! The human brain supposedly has ~86 billion neurons [0]. [0] - Table 2 - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2776484/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2776484/

mgsouth 4y ago
True, but neurons are way more complicated than a transistor. Even a synapse is more complicated and powerful. Neurons are more like microcontrollers.
- stevenhuang 4y ago
  It gets even more complicated when it is hypothesized neurons use spike timings to "compute". > A potentially more powerful way to construct symbols with spikes is to use spike timing, i.e., to consider exactly when spikes occur within the time window. The maximum number of symbols a neuron can construct with spike timing is limited only by the temporal resolution of the code. In formal terms, the information capacity of spike timing is much higher than the information capacity of spike count. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6665236/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6665236/
- walnutclosefarm 4y ago
  Way more complex to be sure. The "more powerful" part requires a bit of interpretation. One transistor certainly can't do the work of one neuron, even given forever to do it. But a not-too-large-to-count collection of transistors can do the work of a single neuron, and do it much faster. I don't know that 100 billion transistors can out compute 80 billion neurons with a 100 trillion synapses, but it's not implausible to me. Neurons have a firing rate on the order of 10Hz; an entire CPU can be clocked at 100 million times that rate. It's certainly not obvious that those 11 orders of magnitude can't make up for the functional limitations of transistors. Obviously standard microprocessor architectures are not comparable to the architecture of the brain, so a true head to head comparison is impossible, but it's not absurd to imagine we've achieved or surpassed parity.
  - tim333 4y ago
    These things are approximate but Hans Moravec made some reasonable estimates back in 1997 that human brain equivalence was about 100 teraflops https://jetpress.org/volume1/moravec.htm https://jetpress.org/volume1/moravec.htm We are around there now, see "NVIDIA GeForce RTX 4090 Is The First Gaming Graphics Card To Deliver 100 TFLOPs of Compute Performance" https://wccftech.com/nvidia-geforce-rtx-4090-is-the-first-gaming-graphics-card-to-deliver-100-tflops-of-compute-performance https://wccftech.com/nvidia-geforce-rtx-4090-is-the-first-ga...
    - walnutclosefarm 4y ago
      Thanks for the links. Confirming, I guess, my instinct that we're near parity in computing power between chips and brains.

Sephr 4y ago

You probably want to qualify "single chip processors" with "small". Single chip processors with over 100 billion transistors have been possible for years. We've had trillion-transistor single chip processors since 2019[1]. 1. https://www.cerebras.net/blog/introducing-the-cerebras-cs-1-the-industrys-fastest-artificial-intelligence-computer/ https://www.cerebras.net/blog/introducing-the-cerebras-cs-1-...

walnutclosefarm 4y ago
Yeah. I've never been able to think of wafer scale integration as "single chip." But your comment is fair. Won't fit in a laptop though ;-)