Covers computer memory, processors, storage, and hardware architecture — building toward GPU compute and CUDA kernels for ML workloads.


Memory (DRAM)

CPU

Foundations

  • cpu-architecture-overview — the 10-layer abstraction stack from transistor to application; map of the CPU pages
  • logic-gates-and-transistors — MOSFET, CMOS NAND, standard cell library, flip-flops, combinational vs sequential
  • cpu-clocks — quartz oscillator, PLL (VCO + divider + PFD + loop filter), fractional-N, H-tree distribution, DVFS, critical path
  • rtl-to-silicon — RTL → synthesis → place-and-route → STA → DRC/LVS → GDSII → tape-out

Architecture & implementation

  • cpu-isa — RISC vs CISC, ARM64 vs x86-64, RISC-V, the ISA contract; why fixed-width matters
  • cpu-microarchitecture — pipeline, superscalar, OoO, register rename, branch predictor, speculation
  • cpu-pipeline — 5-stage classic pipeline, hazards (RAW/control/structural), forwarding, depth trade-offs, OoO
  • cpu-decoder — RISC combinational decode, CISC pre-decode + PLA + microcode, 8086 group decode ROM

Specific implementations

  • cpu-m1-soc — Apple M1 system-on-chip die layout (Firestorm + Icestorm clusters, GPU, NPU, unified memory)
  • cpu-m1-firestorm-core — M1 high-performance core: front end (8-wide decode), back end (~630-entry ROB), execution units
  • cpu-6502 — MOS 6502 die analysis; visual6502 transistor-level simulator; pedagogical decoder topology

Variants

  • cpu-types — CPU vs MCU vs SoC vs GPU vs NPU/TPU vs ASIC vs FPGA — the flexibility ↔ efficiency spectrum

GPU

Storage

System Integration