Skip to content

Assignment 1 · CPU parallelism

Threads, SIMD, task parallelism, and performance analysis.

Program Contents
P1 · CPU threads Worker code, row assignment, load balance, and scaling
P2 · SIMD intrinsics Vector code, masks, and lane utilization
P3 · ISPC SIMD execution, task count, and speedup
P4 · Iterative sqrt Input-dependent convergence and divergence
P5 · SAXPY Memory bandwidth and cache traffic
P6 · K-means Profiling, worker code, debugging, and verification

Required work completed. Extra credit: P2's array sum and P4's AVX2 implementation are complete; P5's optimization was skipped.

Measurement method · Handout