Travis Downs

uarch-bench. A benchmark for low-level CPU micro-architectural features

avx-turbo. Test the non-AVX, AVX2 and AVX-512 speeds across various active core counts

robsize. ROB size testing utility

page-info. Programatically obtain information about the pages backing a given memory region

sort-bench. A benchmark for sorting algorithms

travisdowns.github.io. Performance Matters blog content

concurrency-hierarchy-bench. Supporting code for the concurrency hierarchy described in this blog post: https://travisdowns.github.io/blog/2020/07/06/concurrency-costs.html.

freq-bench. Fine-grained frequency and voltage transition tests

zero-fill-bench. Benchmark for memory store throughput

x86-loop-test. ASM methods to test small loop performance on x86

toupper-bench. Benchmark supporting baseless libel against clang-format

store-bench. Benchmark for various store patterns on x86

fill-bench. Sometimes picking the right zero makes all the difference

polyregex. Regex matching in P with backreference (see fine print)

nasm-utils. Miscellaneous macros useful for writing x86 asm in nasm or yasm

bimodal-performance. Reproduction code for weird bimodal performance on Intel Skylake CPU

dump-vdso. Dumps the VDSO page

hardware-effects. Demonstration of various hardware effects.

interrupt-test. Assembly

clang-format-find. C++

non-silent. Testing silent vs non-silent L2 evictions

uop-test. Tests uop-fusion (micro-fusion) behavior on x86

libpfc. A small library and kernel module for easy access to x86 performance monitor counters under Linux.

divq-test. Tests for div throughput on Intel hardware

pf-test. Prefetch tests

staticman. 💪 Static sites with superpowers

fastscancount. Fast implementations of the scancount algorithm: C++ header-only library

microarchitecturometer. Measures microarchitectural details such as ROB size. Like https://github.com/travisdowns/robsize but without runtime code generation, which doesn't work on some platforms. Public domain except where otherwise noted.

x86-info-term. A terminal viewer for x86 instruction/intrinsic information using Python 3 + curses

pmu-tools. Intel PMU profiling tools

Flops. How many FLOPS can you achieve?

backward-cpp. A beautiful stack trace pretty printer for C++

binary-search. Binary search experiments

fio-snippets. Copy & paste read snippets for fio

turbo-cycles-mystery. Trying to figure out what's up with TurboBoost and REF_TSC cycles reported by the x86 perf counters

ceiling_div. Benchmarks for ceiling divide algorithms

virtual-dispatch. Code for https://stackoverflow.com/questions/46579750

perf-test. Some very simple binaries to test Linux perf overcounting behavior