Coffee Compiler Club meeting on Feb 28, 2025
Typical rambling conversation loosely related to compilers, language runtimes, GC, performance and optimization.
Google docs: https://docs.google.com/document/d/1atS5k5gAbwTi6ZyWnYRoKcc4dK40LIV1uQ6VATEqiLs/edit?tab=t.0
[Started recording time] - please subtract the time in order to match the timing offset in youtube]
Start - 00:04:20
[Timestamps]:
00:00:00 - What was the hunch for Sea of nodes, C1 , C2
00:12:00 - Sea of nodes nodes repr by long, how much mileage restricting nodes to 65k
00:18:00 - Method handles, functions , inline optimization
00:24:00 - Debug info in Sea of nodes - safe points, V8
00:34:00 - Convert C2 frame back to interpreter frame, stackoverflow , guard pages
00:42:00 - A compiler world without a stack
00:51:00 - Chapter 20 Register allocator status, RA theory(and its NP complete)
00:56:40 - Tracing JIT tracking register
00:58:00 - Method calls in simple (future chapter)
01:01:00 - Assembly vs high level code
01:07:00 - How RA allocate loops, interference graphs, graph coloring
01:10:00 - Register allocation in action
01:29:20 - Roadmap for simple
01:37:00 - better byte code
01:41:00 - verification if loaded program - bytecode,
01:42:00 - JIT in a constrained env
01:48:00 - Simple working in the browser, spilling, register allocation, allocation