AIC-100 · GPU & AI Compute · Foundation
Foundations for AI Compute — full syllabus
The architecture, operating system and networking groundwork every GPU systems engineer is assumed to have and often does not.
Who this course is for
Engineers moving into GPU/AI infrastructure who keep hitting gaps in the fundamentals — how the CPU, OS and network actually behave underneath the frameworks.
Prerequisites
- Basic command-line Linux
- Some exposure to a compiled language (C/C++ helpful)
- No GPU experience required
Course outline
Day 1 — Computer architecture for AI engineers
- Fetch/decode/execute and where cycles actually go
- Memory hierarchy: L1/L2/L3/RAM and why it dominates performance
- Cache organisation, miss patterns and coherency basics
- SIMD/vector processing (AVX-512, NEON, SVE)
- NUMA topology on multi-socket platforms
Day 2 — Operating systems for AI engineers
- Processes, threads and the 1:1 NPTL model
- CPU scheduling: FIFO, Round Robin, CFS
- Virtual memory: page tables, TLB, page faults, copy-on-write
- Swapping, the OOM killer and huge pages (2 MB/1 GB)
- I/O subsystem: VFS, page cache, block scheduler
- Concurrency primitives and IPC
Day 3 — Networking for AI clusters
- TCP/IP stack end to end
- Flow control and congestion control (CUBIC, BBR)
- Ethernet switching, VLANs, IP routing
- Data-centre topologies: spine-leaf and fat-tree
- QoS (802.1p, DSCP)
- First look at kernel bypass, RDMA and DPDK
Hands-on labs
- Lab: map your machine's topology with lscpu, numactl and /proc — cores, caches, NUMA nodes, PCIe paths
- Lab: trace a real process with strace and /proc/pid/maps; observe scheduling decisions with schedtool and chrt
- Lab: capture and analyse live traffic with tcpdump/Wireshark; measure latency and bandwidth with iperf3
- Lab: measure cache behaviour with hardware performance counters on a memory-bound microbenchmark
Capstone project
A guided audit of a real (or VM) AI server: you produce a one-page topology and configuration report — CPU/NUMA/PCIe layout, OS memory and scheduler settings, network path — identifying the three configuration issues most likely to hurt GPU workload performance, with evidence for each.
What you leave with
- A hardware/OS/network topology report you can reproduce on any server
- Working fluency with lscpu, numactl, strace, perf, tcpdump and iperf3
- A mental model of the stack underneath every GPU framework
- The vocabulary to read platform specs and whitepapers critically
Upcoming dates
| Dates | Where | Seats | Early bird | Regular | |
|---|---|---|---|---|---|
| 25 Oct – 27 Oct 20263 full days | RiyadhIn person · KAFD Conference Centre | 10 of 14 | — | SAR 6,750 | |
| 1 Nov – 3 Nov 20263 full days | Kuwait CityIn person · Al Hamra Tower | 5 of 14 | — | KWD 560 | |
| 1 Nov – 3 Nov 20263 full days | MuscatIn person · Knowledge Oasis Muscat | 10 of 14 | — | OMR 690 | |
| 8 Nov – 15 Nov 20266 half-days | Gulf bandLive online · 09:00–13:00 GMT+3 | 18 of 20 | US$1,170until 9 Oct | ||
| 9 Nov – 11 Nov 20263 full days | OttawaIn person · Kanata North Tech Park | 5 of 14 | CAD 2,200until 10 Oct | ||
| 16 Nov – 18 Nov 20263 full days | TorontoIn person · MaRS Discovery District | 10 of 14 | CAD 2,200until 17 Oct | ||
| 16 Nov – 18 Nov 20263 full days | LondonIn person · Shoreditch Works | 5 of 14 | GBP 1,260until 17 Oct | ||
| 16 Nov – 23 Nov 20266 half-days | Europe bandLive online · 09:00–13:00 CET | 7 of 20 | US$1,170until 17 Oct | ||
| 16 Nov – 23 Nov 20266 half-days | Americas bandLive online · 13:00–17:00 ET | 12 of 20 | US$1,170until 17 Oct | ||
| 23 Nov – 25 Nov 20263 full days | BerlinIn person · Factory Görlitzer Park | 10 of 14 | EUR 1,490until 24 Oct |
Book a seat, or bring this course to your team
Seats can be reserved online; private delivery runs on-site or live online, adapted to your stack.
Questions about fit or prerequisites? Email hello@kernelsystems.academy. To save this syllabus, print this page to PDF from your browser.