101: Matt Godbolt

Date: 2024-04-10

Duration: 02:29:23

Matt Godbolt joins to talk about early microprocessors, working in the games industry, performance optimization on modern x86 CPUs, and the compute infrastructure that powers the financial trading industry. We also discuss Matt’s work on bringing YouTube to early mobile phones, and the origin story of Compiler Explorer, Matt’s well-known open source project and website.

Matt’s Site: https://xania.org/

Matt on LinkedIn: https://www.linkedin.com/in/godbolt/

Matt on X: https://twitter.com/mattgodbolt

Matt on Mastodon: https://hachyderm.io/@mattgodbolt

Matt on Bluesky: https://bsky.app/profile/mattgodbolt.bsky.social

Show Notes

Welcome Matt (00:01:51)
Jonathan Yu Mastodon Post (00:02:30)
- https://mastodon.social/@jawnsy/111933302725552765
Matt’s Introduction to Computing (00:03:42)
Sinclair ZX Spectrum (00:04:04)
- https://en.wikipedia.org/wiki/ZX_Spectrum
BASIC (00:04:42)
- https://en.wikipedia.org/wiki/BASIC
Z80 Processor (00:05:33)
- https://en.wikipedia.org/wiki/Zilog_Z80
Playing Computer Games (00:06:08)
Writing Computer Games (00:06:44)
Writing Assembly with a Pencil and Paper (00:07:51)
BBC Micro (00:09:53)
- https://en.wikipedia.org/wiki/BBC_Micro
Microcomputers (00:10:02)
- https://en.wikipedia.org/wiki/Microcomputer
Acorn Computers (00:11:02)
- https://en.wikipedia.org/wiki/Acorn_Computers
BBC Master (00:11:56)
- https://en.wikipedia.org/wiki/BBC_Master
MOS 6502 (00:12:02)
- https://en.wikipedia.org/wiki/MOS_Technology_6502
RISC (00:12:11)
- https://en.wikipedia.org/wiki/Reduced_instruction_set_computer
Type-In Programs (00:13:09)
- https://en.wikipedia.org/wiki/Type-in_program
BBC BASIC Assembler (00:14:24)
- https://www.bbcbasic.co.uk/bbcwin/manual/bbcwina.html
Acorn Archimedes (00:14:39)
- https://en.wikipedia.org/wiki/Acorn_Archimedes
6502 Unspecified Opcodes (00:15:11)
- https://www.masswerk.at/nowgobang/2021/6502-illegal-opcodes
Apple I & II (00:18:24)
- https://en.wikipedia.org/wiki/Apple_I
- https://en.wikipedia.org/wiki/Apple_II
Nintendo Entertainment System (NES) (00:18:26)
- https://en.wikipedia.org/wiki/Nintendo_Entertainment_System
Bender (Futurama) (00:18:32)
- https://en.wikipedia.org/wiki/Bender_(Futurama)
The Terminator (00:18:35)
- https://en.wikipedia.org/wiki/The_Terminator
- https://www.apl2bits.net/2016/07/18/terminator-6502/
6502 vs. Z80 Price Comparison (00:18:50)
6502 vs. Z80 Microarchitecture (00:21:13)
Z80 Rotate Instruction (00:24:08)
- https://jgmalcolm.com/z80/advanced/shif
Read-Modify-Write Instructions (00:24:24)
- https://en.wikipedia.org/wiki/Read%E2%80%93modify%E2%80%93write
Programmable Logic Array (PLA) (00:25:29)
- https://en.wikipedia.org/wiki/Programmable_logic_array
Memory Mapped Hardware (00:26:04)
- https://en.wikipedia.org/wiki/Memory-mapped_I/O_and_port-mapped_I/O
Encrypting Games (00:26:18)
Obfuscating Decryption using Unspecified Instructions (00:28:29)
Writing a 6502 Simulator in 6502 Assembly (00:30:08)
Writing a 6502 Emulator in Javascript (00:30:27)
- https://github.com/mattgodbolt/jsbeeb
Zero Page Addressing (00:31:13)
- https://en.wikipedia.org/wiki/Zero_page
Non-Maskable Interrupts (NMI) (00:32:55)
- https://en.wikipedia.org/wiki/Non-maskable_interrupt
6502 Addressing Modes (00:33:43)
- https://www.nesdev.org/obelisk-6502-guide/addressing.html
Register Windows (00:34:29)
- https://en.wikipedia.org/wiki/Register_window
Microarch Club Episode with Robert Garner (00:34:45)
- https://microarch.club/episodes/11/
The Rising Complexity of Computing (00:35:29)
Tom Lyon (00:36:33)
- https://akapugs.blog/
Sun Microsystems (00:36:37)
- https://en.wikipedia.org/wiki/Sun_Microsystems
Abstractions as a Tool (00:37:06)
Going to University (00:38:17)
Studying Physics (00:38:45)
Distance Learning University Lectures (00:39:37)
Discovering the Internet (00:40:28)
Atari (00:41:08)
- https://en.wikipedia.org/wiki/Atari
Amiga (00:41:09)
- https://en.wikipedia.org/wiki/Amiga
The Acorn RISC Machine (The First ARM Chip) (00:41:20)
- https://arstechnica.com/features/2020/12/how-an-obscure-british-pc-maker-invented-arm-and-changed-the-world/
Bill Mensch (00:41:57)
- https://en.wikipedia.org/wiki/Bill_Mensch
Internet Relay Chat (IRC) (00:44:21)
- https://en.wikipedia.org/wiki/IRC
Writing an IRC Client in ARM Assembly (00:44:40)
- https://github.com/mattgodbolt/IRClient
Learning C (00:45:52)
GCC (00:46:36)
- https://gcc.gnu.org/
Multi-User Dungeon (MUD) (00:46:47)
- https://en.wikipedia.org/wiki/Multi-user_dungeon
Applying to a Games Company (00:48:02)
Argonaut Games (00:50:07)
- https://en.wikipedia.org/wiki/Argonaut_Games
Super FX Chip (00:50:30)
- https://en.wikipedia.org/wiki/Super_FX
Super Nintendo Entertainment System (SNES) (00:50:34)
- https://en.wikipedia.org/wiki/Super_Nintendo_Entertainment_System
Jez San (00:50:52)
- https://en.wikipedia.org/wiki/Jez_San
Direct Memory Access (DMA) (00:52:02)
- https://en.wikipedia.org/wiki/Direct_memory_access
Argonaut RISC Core (ARC) (00:52:14)
- https://en.wikipedia.org/wiki/ARC_(processor)
Beginning of Games Consoles (00:52:39)
Hitachi SH-4 (00:53:00)
- https://en.wikipedia.org/wiki/SuperH
Dreamcast (00:53:04)
- https://en.wikipedia.org/wiki/Dreamcast
Multiple Issue (00:53:22)
- https://www.cs.umd.edu/~meesh/411/CA-online/chapter/multiple-issue-processors-i/index.html
Pipelining (00:53:25)
- https://en.wikipedia.org/wiki/Instruction_pipelining
Argonaut Games Red Dog: Superior Firepower (00:54:06)
- https://github.com/mattgodbolt/reddog
Instruction Level Parallelism (ILP) (00:55:13)
- https://en.wikipedia.org/wiki/Instruction-level_parallelism
Out-of-Order Processing (00:55:35)
- https://en.wikipedia.org/wiki/Out-of-order_execution
Intel U-Pipe & V-Pipe (00:55:38)
- http://ps-2.kev009.com/eprmhtml/eprmx/h12204.htm
Argonaut Blazing Renderer (BRender) (00:55:51)
- https://blazingrender.net/
Intel Pentium Pro (00:57:03)
- https://en.wikipedia.org/wiki/Pentium
Intel Pentium Klamath (00:57:14)
- https://en.wikipedia.org/wiki/Pentium_II#Klamath_(80522)
Intel VTune (00:58:02)
- https://en.wikipedia.org/wiki/VTune
Dual-Issue (00:58:38)
Very Long Instruction Word (VLIW) (00:58:40)
- https://en.wikipedia.org/wiki/Very_long_instruction_word
Data Hazards (00:58:48)
- https://en.wikipedia.org/wiki/Hazard_(computer_architecture)
Carl Graham (00:59:01)
Writing Games for Multiple Platforms (01:00:01)
PlayStation (01:00:58)
- https://en.wikipedia.org/wiki/PlayStation
Xbox (01:01:20)
- https://en.wikipedia.org/wiki/Xbox
PlayStation 2 (01:01:59)
- https://en.wikipedia.org/wiki/PlayStation_2
DirectX (01:02:43)
- https://en.wikipedia.org/wiki/DirectX
Component Object Model (COM) (01:02:55)
- https://en.wikipedia.org/wiki/Component_Object_Model
Porting from Xbox to PlayStation 2 (01:04:55)
Blending Modes (01:06:45)
- https://www.khronos.org/opengl/wiki/Blending
Current State of Game Engine Development (01:07:51)
Unreal Engine (01:07:58)
- https://en.wikipedia.org/wiki/Unreal_Engine
Deciding to Start a Company (01:08:49)
Optimizing C++ Build Time (01:09:48)
ProFactor (01:10:56)
Rockstar Games (01:11:45)
- https://en.wikipedia.org/wiki/Rockstar_Games
Developing a C++ Code Formatter (01:13:03)
Applying to Google (01:13:48)
DeepMind (01:14:17)
- https://en.wikipedia.org/wiki/Google_DeepMind
Making YouTube Work on Early Mobile Phones (01:14:27)
Hardware MPEG Decoders (01:15:07)
- https://en.wikipedia.org/wiki/H.264/MPEG-4_AVC_products_and_implementations#Decoding
YouTube Infrastructure (01:15:33)
J2ME (01:16:34)
- https://en.wikipedia.org/wiki/Java_Platform,_Micro_Edition
Android (01:16:44)
- https://en.wikipedia.org/wiki/Android_(operating_system)
Working at Large vs. Small Companies (01:17:02)
Learning at Google Scale (01:18:00)
Moving to Finance Industry (01:20:23)
Moving to Chicago (01:21:50)
Performance Optimization in Finance (01:21:59)
Reverse Engineering How Intel Processors Work (01:24:08)
Resources for Performance Analysis (01:25:28)
Agner Fog (01:25:42)
- https://www.agner.org/optimize/
Bjarne Stroustrup: Tour of C++ (01:26:39)
- https://www.stroustrup.com/tour2.html
Charles Petzold: The Annotated Turing (01:27:13)
- https://en.wikipedia.org/wiki/The_Annotated_Turing
Register Rename Buffer (01:28:31)
- https://en.wikipedia.org/wiki/Register_renaming
Agner Fog’s Open Source Tooling (01:28:46)
- https://github.com/mattgodbolt/agner
uops.info (01:29:02)
- https://uops.info/
Branch Predictor Reverse Engineering (01:30:19)
- https://en.wikipedia.org/wiki/Branch_predictor
Branch Target Buffer (01:31:54)
- https://one2bla.me/cs6290/lesson4/branch-target-buffer.html
Meltdown & Spectre Paper Attribution (01:32:57)
- https://meltdownattack.com/
- https://xania.org/201602/bpu-part-two
Compute Infrastructure in Finance (01:33:29)
Low Latency Trading Infrastructure (01:35:29)
Executing Orders Over TCP (01:35:56)
Market Data Over Multicast (01:38:07)
Lead-Lag Trades (01:40:18)
FPGAs for Accelerating Trading (01:41:49)
- https://en.wikipedia.org/wiki/Field-programmable_gate_array
Engineering Organization Structure in Trading Firms (01:43:45)
cocotb (01:46:13)
- https://www.cocotb.org/
Updating Trading Hardware (01:47:05)
Tilera RISC CPU (01:50:45)
- https://en.wikipedia.org/wiki/Tilera
- https://en.wikipedia.org/wiki/TILE64
Propagation Delay (01:54:46)
- https://en.wikipedia.org/wiki/Signal_propagation_delay
Number of Cycles in an Integer Division (01:55:26)
- https://en.algorithmica.org/hpc/arithmetic/division/
Hashmap Performance Optimization (01:56:40)
Rising Complexity in Modern Processors (01:58:11)
Microarch Club Episode with Thomas Sohmers (02:01:06)
- https://microarch.club/episodes/10/
CPU Cache Levels (02:02:34)
- https://en.wikipedia.org/wiki/CPU_cache
Impact of New Workloads on Microarchitecture (02:03:34)
Groq & LPUs (02:04:00)
- https://wow.groq.com/lpu-inference-engine/
- https://codeconfessions.substack.com/p/groq-lpu-design
Branch Predictor Hints (02:05:11)
- https://arcb.csc.ncsu.edu/~mueller/cluster/ps3/SDK3.0/docs/accessibility/sdkpt/cbet_3reducbrmis.html
Just in Time (JIT) Compilation (02:05:56)
- https://en.wikipedia.org/wiki/Just-in-time_compilation
Profile Guided Optimization (PGO) (02:06:16)
- https://en.wikipedia.org/wiki/Profile-guided_optimization
Tail Latency (02:08:42)
- https://brooker.co.za/blog/2021/04/19/latency.html
Compiler Explorer Origin Story (02:10:22)
- https://godbolt.org/
C++ Ranged For Loops (02:11:15)
- https://en.cppreference.com/w/cpp/language/range-for
tmux (02:14:30)
- https://github.com/tmux/tmux/wiki
node.js (02:15:13)
- https://nodejs.org
Read Evaluate Print Loop (REPL) (02:17:43)
- https://en.wikipedia.org/wiki/Read%E2%80%93eval%E2%80%93print_loop
Remote Code Execution (RCE) (02:19:21)
- https://en.wikipedia.org/wiki/Arbitrary_code_execution
Hosting on Personal Domain (02:20:30)
Godbolt as a Verb (02:21:01)
Two’s Complement Podcast (02:22:10)
- https://www.twoscomplement.org/
Ben Rady (02:23:12)
- https://www.benrady.com/
Getting Feedback on Podcasts (02:25:45)
Podcasts as a Hiring Tool (02:26:52)
Getting a Job via Open Source (02:27:15)

Transcript

Coming soon.