GPU (Graphical Processing Unit)
Specialized computer processors originally designed for graphics rendering but now widely used for parallel computation including machine learning; Nvidia's GPUs dominate AI training
As defined in Forget Everything You Believed About Computing w/ Gill Verdon | EP #102
Where it comes up · 1
Most important claims · 4 of 4Search all claims for “GPU (Graphical Processing Unit)” →
- GPUs were originally created for video graphics but accidentally proved excellent for matrix multiplication, leading to their dominance in AI despite not being designed from first principles for that purpose; this co-evolution between algorithms and hardware is suboptimal.Forget Everything You Believed About Computing w/ Gill Verdon | EP #102
- The market is a powerful aligning force for AI development: products without positive utility fail to gain users and companies fail, creating selective pressure toward aligned, interpretable models that perform well, which is gentler and more effective than top-down regulation of compute.Forget Everything You Believed About Computing w/ Gill Verdon | EP #102
- Modern deep learning algorithms co-evolved with GPUs—algorithms that ran well on GPUs were selected for, creating a bidirectional feedback loop where hardware and algorithms reinforce each other; thermodynamic computing aims to create a similar co-evolution fork in the space of both hardware and algorithms.Forget Everything You Believed About Computing w/ Gill Verdon | EP #102
- Monte Carlo algorithms running on traditional CPUs or GPUs achieve roughly 1,000 samples per second, whereas Extropic's thermodynamic prototype achieves samples at picosecond timescales (10^-12 seconds), yielding orders-of-magnitude speedups for algorithms that map well onto the hardware physics.Forget Everything You Believed About Computing w/ Gill Verdon | EP #102