“Welcome to the Jungle” (CPU design and Moore’s law; cf. “How Many Computers In Your Computer?”)

“There’s plenty of room at the Top: What will drive computer performance after Moore’s law?”, Leiserson et al 2020 (gains from End-To-End Principle-based systems design¹)

“Scalability! But at what COST?”, McSherry et al 2015 (when laptops > clusters: the importance of good baselines; “Big Data” doesn’t fit in a laptop)

“What it takes to run Stack Overflow”⁠, Nick Craver

“Anatomy of a hack: How crackers ransack passwords like qeadzcwrsfxv1331; For Ars, 3 crackers have at 16,000+ hashed passcodes—with 90% success”

“The Science of Brute Force”

“Bracket pair colorization 10,000× faster”

“Parsing Gigabytes of JSON per Second”, Langdale & Lemire2019

55 GiB /  ​s FizzBuzz⁠; “Modern storage is plenty fast. It is the APIs that are bad.”⁠; “10M IOPS, one physical core. #io_uring #linux”⁠; “Achieving 11M IOPS & 66 GB /  ​s IO on a Single ThreadRipper Workstation”⁠; “Serving Netflix Video at 50 gigabyte /  ​s on FreeBSD”, Gallatin2021/​“Serving Netflix Video Traffic at 100 gigabyte /  ​s and Beyond”, Gallatin2022

“True Stories of Algorithmic Improvement”

Ultra-low bit-rate audio for tiny podcasts with Codec2 and WaveNet decoders (1 hour = 1MB)

“The computers are fast, but you don’t know it” (replacing slow pandas with 6,000× faster parallelized C++)

“Scaling SQLite to 4M QPS on a Single Server (EC2 vs Bare Metal)”⁠, Expensify

Demoscene:

• “8088 Domination”: “how I created 8088 Domination, which is a program that displays fairly decent full-motion video on a 1981 IBM PC”⁠/​part 2
• “A Mind Is Born” (256 bytes; for comparison, this line of Markdown linking the demo is 194 bytes long, and the video several million bytes)

Casey Muratori: refterm v2⁠, Witness pathing (summary)

Algorithmic Progress:

• “Algorithmic Progress in 6 Domains”, Grace2013
• “A Time Leap Challenge for SAT Solving”, Fichte et al 2020 (hardware overhang: ~2.5× performance increase in SAT solving since 2000, about equally due to software & hardware gains, although slightly more software—new software on old hardware beats old on new.)
• 2019 Unlocking of the LCS35 Challenge
• “Measuring hardware overhang”⁠, hippke (“with today’s algorithms, computers would have beat the world chess champion already in 1994 on a contemporary desk computer”)

Memory Locality:

• “What Every Programmer Should Know About Memory”, Drepper2007
• “You’re Doing It Wrong: Think you’ve mastered the art of server performance? Think again.”⁠, Poul-Henning Kamp 2010 (on using cache-oblivious B-heaps to optimize Varnish performance 10×)
• “The LMAX Architecture”, Fowler2011
• “5000× faster CRDTs: An Adventure in Optimization”
• “In-place Superscalar Samplesort (IPS⁴o): Engineering In-place (Shared-memory) Sorting Algorithms”, Axtmann et al 2020
• “Fast approximate string matching with large edit distances in Big Data (2015)”
• “ParPaRaw: Massively Parallel Parsing of [CSV] Delimiter-Separated Raw Data”, Stehl & Jacobsen2019
• “Scaling our Spreadsheet Engine from Thousands to Billions of Cells: From Maps To Arrays”, Lukas Köbis2022
• “FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness”, Dao et al 2022

DL⁠/​Deep Reinforcement Learning:

• Training A3C to solve Atari Pong in <4 minutes on the ARCHER supercomputer through brute parallelism
• “megastep helps you build 1-million FPS reinforcement learning environments on a single GPU”⁠, Jones
• “The Mathematics of 2048: Optimal Play with Markov Decision Processes”
• Bruteforcing Breakout (optimal play by depth-first search of an approximate MDP in an optimized C++ simulator for 6 CPU-years; crazy gameplay—like chess endgame tables⁠, a glimpse of superintelligence)
• “Sample Factory: Egocentric 3D Control from Pixels at 100,000 FPS with Asynchronous Reinforcement Learning”, Petrenko et al 2020; “Megaverse: Simulating Embodied Agents at One Million Experiences per Second”, Petrenko et al 2021; “Brax: A Differentiable Physics Engine for Large Scale Rigid Body Simulation”, Freeman et al 2021; “Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning”, Makoviychuk et al 2021 (Rudin et al 2021); WarpDrive
• “Large Batch Simulation for Deep Reinforcement Learning”, Shacklett et al 2021
• “Scaling Scaling Laws with Board Games”, Jones2021
• “Mastering Real-Time Strategy Games with Deep Reinforcement Learning: Mere Mortal Edition”⁠, Winter
• “Scaling down Deep Learning”, Greydanus2020
• “Podracer architectures for scalable Reinforcement Learning”, Hessel et al 2021 (highly-efficient TPU pod use: eg. solving Pong in <1min at 43 million FPS on a TPU-2048); “Q-DAX: Accelerated Quality-Diversity for Robotics through Massive Parallelism”, Lim et al 2022; “EvoJAX: Hardware-Accelerated Neuroevolution”, Tang et al 2022
• “Training the Ant simulating at 1 million steps per second on CUDA (NVIDIA RTX 2080), using Tiny Differentiable Simulator and C++ Augmented Random Search.” (Erwin Coumans2021)
• “TextWorldExpress: Simulating Text Games at One Million Steps Per Second”, Jansen & Côté2022