Using light as a neural network, as this viral video depicts, is...
Deedy@deedydas
18 views
Nov 29, 2024
1
Using light as a neural network, as this viral video depicts, is actually closer than you think. In 5-10yrs, we could have matrix multiplications in constant time O(1) with 95% less energy. This is the next era of Moore's Law.
Let's talk about Silicon Photonics...
1/9
Let's talk about Silicon Photonics...
1/9
2
The core concept: Replace electrical signals with photons.
While current processors push electrons through metal pathways, photonic systems use light beams, operating at fundamentally higher speeds (electronic signals in copper are 3x slower) with minimal heat generation.
2/9
While current processors push electrons through metal pathways, photonic systems use light beams, operating at fundamentally higher speeds (electronic signals in copper are 3x slower) with minimal heat generation.
2/9
3
It's way faster.
While traditional chips operate at 3-5 GHz, photonic devices can achieve >100 GHz switching speeds. Current interconnects max out at ~100 Gb/s. Photonic links have demonstrated 2+ Tb/s on a single channel. A single optical path can carry 64+ signals.
3/9
While traditional chips operate at 3-5 GHz, photonic devices can achieve >100 GHz switching speeds. Current interconnects max out at ~100 Gb/s. Photonic links have demonstrated 2+ Tb/s on a single channel. A single optical path can carry 64+ signals.
3/9
4
It's way more energy efficient.
Current chip-to-chip communication costs ~1-10pJ/bit. Photonic interconnects demonstrate 0.01-0.1pJ/bit. For data centers processing exabytes, this 200x improvement means the difference between megawatt and kilowatt power requirements.
4/9
Current chip-to-chip communication costs ~1-10pJ/bit. Photonic interconnects demonstrate 0.01-0.1pJ/bit. For data centers processing exabytes, this 200x improvement means the difference between megawatt and kilowatt power requirements.
4/9
5
The AI acceleration potential is revolutionary.
Matrix operations, fundamental to deep learning, become near-instantaneous:
Traditional chips: O(n²) operations.
Photonic chips: O(1) - parallel processing through optical interference.
1000×1000 matmuls in picoseconds.
5/9
Matrix operations, fundamental to deep learning, become near-instantaneous:
Traditional chips: O(n²) operations.
Photonic chips: O(1) - parallel processing through optical interference.
1000×1000 matmuls in picoseconds.
5/9
6
Where are we today?
Real products are shipping:
— Intel's 400G transceivers use silicon photonics.
— Ayar Labs demonstrates 2Tb/s chip-to-chip links with AMD EPYC processors. Performance scales with wavelength count, not just frequency like traditional electronics.
6/9
Real products are shipping:
— Intel's 400G transceivers use silicon photonics.
— Ayar Labs demonstrates 2Tb/s chip-to-chip links with AMD EPYC processors. Performance scales with wavelength count, not just frequency like traditional electronics.
6/9
7
The manufacturing challenges are immense.
— Current yield is ~30%. Silicon's terrible at emitting light and bonding III-V materials to it lowers yield
— Temp control is a barrier. A 1°C change shifts frequencies by ~10GHz.
— Cost/device is $1000s
7/9
— Current yield is ~30%. Silicon's terrible at emitting light and bonding III-V materials to it lowers yield
— Temp control is a barrier. A 1°C change shifts frequencies by ~10GHz.
— Cost/device is $1000s
7/9
8
To reach mass production we need: 90%+ yield rates, sub-$100 per device costs, automated testing solutions, and reliable packaging techniques. Current packaging alone can cost more than the chip itself.
We're 5+ years from hitting these targets.
8/9
We're 5+ years from hitting these targets.
8/9
9
Companies to watch: ASML (manufacturing), Intel (data center), Lightmatter (AI), Ayar Labs (chip interconnects).
The technology requires major investment, but the potential returns are enormous as we hit traditional electronics' physical limits.
9/9
The technology requires major investment, but the potential returns are enormous as we hit traditional electronics' physical limits.
9/9