See Also:

  • compilers
  • computer architecture
  • operating systems
  • assembly
  • parallel
  • concurrency
  • ebpf
  • fuzzing

Profiling

I mean. This is the whole name of the game really.

https://thume.ca/2023/12/02/tracing-methods/ All my favorite tracing tools: eBPF, QEMU, Perfetto, new ones I built and more - Tristan Hume

Intel PT ebpf

Estimating Maximum Possible Perf

If compute bound: Single core freq ~3Ghz * 8 byte words -> 24Gb/s RAM SSD speeds - look it up. sequential vs no nsequential v different. Maybe ~1 gb/s as a order of magnitude

Network

latency numbers every programmer should know napkin math

https://twitter.com/lemire/status/1746276695030600182 estimating memory bandwidth https://lemire.me/blog/2024/01/13/estimating-your-memory-bandwidth/

What kind of stuff is in my cpu. How to estimate various parameters.

lscpu -C # cache
#NAME ONE-SIZE ALL-SIZE WAYS TYPE        LEVEL SETS PHY-LINE COHERENCY-SIZE
#L1d       48K     192K   12 Data            1   64        1             64
#L1i       32K     128K    8 Instruction     1   64        1             64
#L2       512K       2M    8 Unified         2 1024        1             64
#L3         8M       8M   16 Unified         3 8192        1             64

lscpu -e
#CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ      MHZ
#  0    0      0    0 0:0:0:0          yes 3900.0000 400.0000 1500.000
#  1    0      0    1 1:1:1:0          yes 3900.0000 400.0000 1500.000
#  2    0      0    2 2:2:2:0          yes 3900.0000 400.0000 1500.000
#  3    0      0    3 3:3:3:0          yes 3900.0000 400.0000 1500.000
#  4    0      0    0 0:0:0:0          yes 3900.0000 400.0000 1500.000
#  5    0      0    1 1:1:1:0          yes 3900.0000 400.0000 1500.000
#  6    0      0    2 2:2:2:0          yes 3900.0000 400.0000 2719.516
#  7    0      0    3 3:3:3:0          yes 3900.0000 400.0000 2873.892

#  Address sizes:         39 bits physical, 48 bits virtual
# Vulnerabilities. That's ineresting

https://en.wikichip.org/wiki/intel/core_i7/i7-1065g7 my cpu Max 55.63 GiB/s Bandwidth Single 13.91 GiB/sDouble 27.82 GiB/sQuad 55.63 GiB/s

Instruction Level Parallelism (ILP)

SIMD

https://news.ycombinator.com/item?id=38443253 https://mcyoung.xyz/2023/11/27/simd-base64/ Designing a SIMD Algorithm from Scratch

Filtering a vector with simd - rust

art of simd - slotkin

https://arxiv.org/abs/2306.00229 Minotaur: A SIMD-Oriented Synthesizing Superoptimizer

https://dl.acm.org/doi/10.1145/3372297.3423352 HACLxN: Verified Generic SIMD Crypto (for all your favourite platforms)

https://simdjson.org/ https://roaringbitmap.org/ compressed bitmaps

http://0x80.pl/notesen/2023-04-09-faster-parse-ipv4.html

Simd for C++ developers

What this? https://branchfree.org/2019/02/25/paper-parsing-gigabytes-of-json-per-second/ https://news.ycombinator.com/item?id=24069530 roaring bitmaps simdjson judy arrays People are mentioned warming up the branch predictors on purpose somehow Branchless programming

"
#include <chrono>
#include <thread>
#include <vector>
#include <cstdint>
#include <iostream>

volatile size_t g_sum = 0;
__attribute__ ((noinline))
uint64_t sum(const uint8_t *data, size_t start, size_t len, size_t skip = 1) {
    uint64_t sum = 0;
    for (size_t i = start; i < len; i+= skip) {
        sum += data[i];
    }
    g_sum += sum;
    return sum;
}
" 


https://github.com/cmuratori/blandwidth https://github.com/MattPD/cpplinks/blob/master/performance.tools.md#memory---benchmarking

https://www.intel.com/content/www/us/en/developer/articles/tool/intelr-memory-latency-checker.html intel memory latency checker

Memory

See also note on memory-management

dhat check for memory allocation sites that are worst https://en.wikipedia.org/wiki/Memory_pool https://en.wikipedia.org/wiki/Object_pool_pattern Can use vector to store fixed size chunks. Your own private malloc specialized for one size https://en.wikipedia.org/wiki/Slab_allocation

memory overhead. Probably Powers of two. But allocator may have “slop” can lose a lot of memory that way

What every programmer should know about memory https://people.freebsd.org/~lstewart/articles/cpumemory.pdf

RAM - DDR https://en.wikipedia.org/wiki/Synchronous_dynamic_random-access_memory double data rate. read cycle time - time between reads tosame row CAS latency - time from column to recieving data

RAS, CAS, WE bits select command type

Cache lines - 64 bytes. Even if you read/write 1 byte you’re writing 64

Power

Power profiling is determining what is using electricity / battery life up. Very relevant to my common problem with my laptop dying

https://firefox-source-docs.mozilla.org/performance/#power-profiling

powertop https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/8/html/monitoring_and_managing_system_status_and_performance/managing-power-consumption-with-powertop_monitoring-and-managing-system-status-and-performance

https://wiki.archlinux.org/title/Laptop_Mode_Tools

Hmm. wifi is using a lot of power

intel_atomic_commit - huh. is this actually correlated with touching the touchpad? https://community.frame.work/t/tracking-touchpad-interrupts-battery-usage-issues-idma64-2/13630

nic:docker0 is using a lot. sudo systemctl disable docker.service sudo ifconfig docker0 down sudo ifconfig br-557a6ccfc9fc down

A mysterious br-something device is usng like 5W of power. https://askubuntu.com/questions/814316/how-to-determine-identity-of-obscure-nicbr-devices-in-powertop?rq=1 INternet still works if I turn that off

hrtimer_wakeup

Cache

Page

https://www.computerenhance.com/p/powerful-page-mapping-techniques how to allocate huge tables

Resources

Easyperf

Performance matters, it unlocks new applications, important for business python -> avx extensions: x60,000 in one example Measurement is really important and hard. CPU can overclock for a little bit. Try to control the environment

Use statistical tests to determine if real change. student t for example Plot your benchmark data. Bimodal? Two different behaviors are happening microbenchmarks: be careful. Is it inlining a bunch of stuff? Anything except your exact final application and environment is a proxy. That the proxy at all represents the real behavior is fishy. Never forget that. System clock and system counters.

Agner Fog

manual 1

Reduce data dependencies a[i++] may be faster than a[++i] because of a data dependency reduction bool in C++ outputs 0/1 but may have come from a source that didn’t. This means it needs branching code for simple satuff short circuiting && ||, try to short circuit early

Stuff

https://github.com/google/highway Performance-portable, length-agnostic SIMD with runtime dispatch

https://x.com/ashvardanian/status/1759819906160480651?s=20 https://github.com/ashvardanian/StringZilla https://github.com/unum-cloud/ucall https://github.com/unum-cloud/usearch

rust perf book https://nnethercote.github.io/perf-book/title-page.html

Intel PT LBR

1 billion row challenge 1brc https://twitter.com/search?q=1brc&src=typed_query https://github.com/gunnarmorling/1brc https://www.morling.dev/blog/one-billion-row-challenge/ https://github.com/gunnarmorling/1brc/discussions/categories/show-and-tell https://news.ycombinator.com/item?id=39020906 remoe java unsafe

criterion https://github.com/bheisler/criterion.rs https://github.com/bheisler/iai use cachegrind for stable measurements in CI https://github.com/sharkdp/hyperfine benchamrking commandline programs https://github.com/dandavison/chronologer timeline over git commit

casey muratori series

cp algorithms competitive programming algorithsm

geoff langdale blog

iterating over set bits

Computing Adler32 Checksums at 41 GB/s https://wooo.sh/articles/adler32.html

how fast are linux pipes anyway pv - pipe viewer. pipe thorughput http://www.ivarch.com/programs/pv.shtml

really cool blog posts https://mazzo.li/archive.html

Software pipelining https://twitter.com/geofflangdale/status/1531858991336550400?s=20&t=geOHu86_aPOLlcz-y8fE0Q

The Art of Writing Efficient Programs: An advanced programmer’s guide to efficient hardware utilization and compiler optimizations using C++ examples - Pikus

Given the potential for straightline speculation w/ deleterious performance impact, does it makes sense to align functions with speculation blocking instructions like INT3 instead of nops? microbenchmarks of return address prediction (ras)

programming parallel computers course asm("# foo"); nice trick. Inject comment into assembly

fread vs mmap Rough advice: fread is simple and often comparable to mmap (system dependent). mmap can sometimes be up to 4x faster, use madvise, weird exceptions/signals need to be handled.

performance tuning on linux - cromwell

dan luu new cpu features

Parsing series of integers with SIMD This Wojciech Muła guy is a wizard

unaligned vector load + length-driven PSHUFB. What’s everyone’s favourite way to handle page crossings? overreading for short variables possibly into out of bounds memory? pshufb

umash very fast hash Algorithms for Modern Hardware - book on algorithms on modern hardware

CLMUL fast instruction set for galois field calculations. carryless

x86 instrinsic cheatsheet

OSACA an analyzer of assembly code. It is on godbolt

This gruop has a number of interesting tools. https://github.com/RRZE-HPC It scrapes info from

  • Likwid
  • kerncraft loop kernel analysis and performance modelling

List of interesting optimizers - These are compiler optimizations, so hopefully your compiler does them for you, but maybe it doesn’t and maybe

https://twitter.com/lemire/status/1461181871841320962?s=20 Lemire converting integerrs to fix digit representations By considering data dependencies and using lookup tables take from 25ns to 2ns.

MIT optimization course https://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-172-performance-engineering-of-software-systems-fall-2018/

https://news.ycombinator.com/item?id=29107147 https://randomascii.wordpress.com/2012/12/29/the-surprising-subtleties-of-zeroing-a-register/ surprising subtleites of zeroing a register.

https://www.agner.org/optimize/ agner fog optimization manuals

https://twitter.com/nadavrot/status/1464364562409422852?s=20 memset and memcpy ooptimizations

https://twitter.com/PieCalculus/status/1464252793225678850?s=20 Go does not need a garbage collector. Compares and contrasts java GC with others. Claims Java poorly designed make high pressure on GC

https://www.intel.com/content/www/us/en/developer/articles/technical/intel-sdm.html https://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-optimization-manual.pdf intel opimization manual

https://twitter.com/pervognsen/status/1455409607426207744?s=20 mimalloc- de moura, daan leijen, ben zorn https://lobste.rs/s/4awecj/mimalloc_free_list_sharding_action https://github.com/microsoft/snmalloc/blob/c5b65d07b8079b22eec9f78bec197ea7a0fd15f2/difference.md

I feel like most algorithms and data structures are os ordinary they are kind of boring?

Sparse Sets - knuth - bitvectors + Bitvectors http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.681.8766&rep=rep1&type=pdf ullmann bitvector algos for binary constraint and subgraph iso.

Books: CLRS

Sorting algorithms Hash tables Dynamic programming Tries Graph algorithsm - shortest path, spanning tree

https://news.ycombinator.com/item?id=26590234#26592091 hash table in C. some interesting commments too linear search - an assoc list but he kept it in an array http://burtleburtle.net/bob/hash/doobs.html - hashing from z3 source code https://craftinginterpreters.com/hash-tables.html

lkinear probing vs linked list in hash table.

concurrent hash map

https://algorithmica.org/en/eytzinger https://news.ycombinator.com/item?id=26695694 Interesting. Cache-oblivious binary search. Uses the “Heap” ordering or what have you Plus a branchless comparator? I think also a big point is How do you even know when cache something is a problem. How do you use feedback and self correct? How do you organize tight loops? “smart” ways of keeping structure.

microbenchmarking performance counters - cache misses, TLB ht/miss, mispredicted branches nanobench https://arxiv.org/pdf/1911.03282.pdf VTune, perf, PAPI, libpfc,

What every programmer should know abouyt memory https://people.freebsd.org/~lstewart/articles/cpumemory.pdf

modern microprcessor 90 minute guide http://www.lighterra.com/papers/modernmicroprocessors/

https://en.wikipedia.org/wiki/Program_optimization Bentley Writing Efficient Program

https://news.ycombinator.com/item?id=28955461 - a rust optimization story https://pvk.ca/Blog/2012/07/03/binary-search-star-eliminates-star-branch-mispredictions/ https://dirtyhandscoding.wordpress.com/2017/08/25/performance-comparison-linear-search-vs-binary-search/ https://www.youtube.com/watch?v=1tEqsQ55-8I&ab_channel=MollyRocket - handmade hero guy talkin about optimizations https://www.youtube.com/watch?v=pgoetgxecw8&ab_channel=MollyRocket - refterm optimization talk. this is fascinating

  1. optimization - measuring.
  2. non-pessimization - don’t do unnecessary work
  3. fake optimziation - people just repeatin shit

https://www.uops.info/ https://uica.uops.info/ uica online demo gives info on what’s hurtin ya. Cycle counts and stuff microp_ops. Ports? Queue? DaY 112 of hnadmade hero. perf counter. simd. converting to simd. measuring port usage with iaca

perf seems balla. Works on ocaml btw https://ocaml.org/learn/tutorials/performance_and_profiling.html https://www.brendangregg.com/perf.html https://www.youtube.com/watch?v=fhBHvsi0Ql0&ab_channel=USENIX - linux systems performance

https://www.gem5.org/ The gem5 simulator is a modular platform for computer-system architecture research, encompassing system-level architecture as well as processor microarchitecture https://ieeexplore.ieee.org/document/8718630?denied= gem5, MARSS×86 , Multi2Sim, PTLsim, Sniper, and ZSim. gem5 as an alternaitve to qemu? http://www.diva-portal.org/smash/get/diva2:1058030/FULLTEXT01.pdf

https://www.infoq.com/presentations/microarchitecture-modern-cpu/

NUMA - non uniform memory access register file? l1 cache. instruction and data. instruction is one way lstopo --no-io tells you how your computer looks large /huge pages. faster for TLB. Hugetablefs is linux suppotrt? /proc/cpuinfo Transparent Huge Pages- madvise is a call yes I’d like huge tables. defer cache lines - 64 bytes. even if you read/write 1 byte your’re writing 64 M exculsively own and dirty, E exlucsive and clean, S shared, , I invalid _builtin_prefecth. linear access is good splitting into revcord of arrays tends to be better for cache if only using one field. compressed memory is worth it. compuitayion is fast. memory is slow. Array of structs vs struct of arrays. Compressed pointers? https://en.wikipedia.org/wiki/Tagged_pointer https://v8.dev/blog/pointer-compression pinning isolcpus boot time option. pinning of thread or memory to cpu taskset. linux admin styuff. isolate cpus to certain tasks numactl and libnuma loop stream decoder branch predictor, pipelikne stall or bubble. branch target predcitro ports, execution units. some logic, some airthmetic. perf - interrogate counters. record report annotate stat skid - bad - precision knobs :p :pp :ppp perf record -b perf record –call-graph lbr -j any_call,any_ret program -e intel_pt//u LBR - last branch record - linux weekly https://lwn.net/Articles/680985/ intel processors record control flow Intel processor trace IPC - intrcutions per cycle. 4 is maximum ish. less than 1 is perf stat performance ocunters - perf -list TMAM top down microarctecture analsyis method perf -dtopdown toplev go throgyh process. and kleen. fancy frontend to perf/ -l1 l2 __builtin_expect profile guided optimization may do builtin expect for you loop alginment - 32 bit boundaries. straight from uop cache. llvm flag. align-all-nofallthru-blocks align-all-function https://easyperf.net/blog/2018/01/18/Code_alignment_issues code alignment can changed your perfoamnce. BOLT - vinary optimization layout tranformer. defrag your code. Puts hot code in same memory location at runtime Daniel Lemire - simd parser. mechnisms for avodiing branching. masking operations. Summary - cache aligned / cache aware data structures. B-trees. Compress data. Avoid random memory access. Huge pages can help. 10% speedup by enabling maybe. libnuma source memory. branch0free and lock-free. perf /toplev. Use vectorization where you can. https://alblue.bandlem.com/ his blog https://speakerdeck.com/alblue/understanding-cpu-microarchitecture-for-performance?slide=62 reference https://speakerdeck.com/alblue/understanding-cpu-microarchitecture-for-performance?slide=63 links

People

Blog links neato: https://easyperf.net/notes/ https://epickrram.blogspot.com/ https://lemire.me/blog/ http://psy-lob-saw.blogspot.com/ https://richardstartin.github.io/ https://travisdowns.github.io/ https://www.agner.org/optimize/ https://www.real-logic.co.uk/ https://groups.google.com/g/mechanical-sympathy nethrcote blog https://nnethercote.github.io/