James Mitchell

My talk at @GCAPConf on Expert Level Debugging that shows how to manually unwind corrupted call stacks using stack memory and registers is now online. youtube.com/watch?v=UQNdkd…

Take the C++ performance Quiz! quiz.cpp-perf.com #cpp #cplusplus

All feedback is welcome, there is bound to be some bugs in the questions/answers

Sharing my fun side project: A C++ Performance Quiz Covers algorithms to low-level gotchas, with Compiler Explorer links to see the generated machine code. Built for fun & intuition, not to judge skill! quiz.cpp-perf.com (Personal project, independent of employer)

@notreaIlyjohn @ChShersh That would probably be worse for any text that is heavily ASCII as you introduce a loop carried data dependency, as each loop iteration depends on the data of all previous iterations, where the branch does not so if it predicts correctly it can be significantly faster.

@ChShersh It would be better to make a lookup table and then ```c++ const uint8_t* data = (const uint8_t*)s.data(); while (i < s.size()) { uint8_t len = utf8_len_table[data[i]]; i += len; width++; } ``` no branch mispredictions

I needed to implement this function in C++ to count graphemes (aka visible characters) in a string because string .size() returns the number of bytes Invaluable my TUI in C++ implementation

@abdimoalim_ If it was doubly linked you could go from the head and tail simultaneously and get even better throughput, with less reliance on the nodes being consecutive and potential undefined behaviour.

Beating the L1 cache with value speculation (by Francesco Mazzoli): mazzo.li/posts/value-sp…

@ID_AA_Carmack My Twitter for you page is full of American right wing politics, I am an Australian game developer, the algorithm is broken. Blue checkmarks ensure visibility and why would I pay for a platform like that? Just look at the responses need to scroll heaps to see one non-blue check

It would be nice if some of the Twitter diaspora returned. So many creatives, but also many developers, that generally enriched the experience are no longer active. Those that performatively left and those with a seething hatred of Elon probably won’t be back soon, but a lot of people just disengaged on vague cultural grounds that can be reevaluated. There are probably some technical tweaks to the algorithm that could make them more comfortable. I don’t mind the existence of independent echo chambers that people are happy within. There is only a problem when some echo chambers are allowed and others aren’t. Reach out to lapsed friends!

@Cali_burbia @MickWest Is it possible that the pilot suffered from Sensory Illusion? An easy way to test is see if you can reproduce the same thing with glare, if you can it seems like a plasuable explanation over something extraordinary. en.wikipedia.org/wiki/Sensory_i…

@MickWest Your analysis goes to great length to suggest a prosaic explanation w/o considering why a trained pilot isn’t accustomed to seeing glare rotation. That’s a problem. You think pilots haven’t looked at flir while banking? Do better.

Marik is wrong, and perhaps a little confused. I stand by my Gimbal "glare" hypothesis, as it's backed by the facts. What I actually said: metabunk.org/threads/some-r… I'm quite happy with the position that the totality of the evidence I outlined in my original video shows that it's a rotating glare, even if there are some unanswered questions about some aspects of the rotation. I don't think they are significant to the overall argument. Frankly, I think you can tell it's a glare just by looking at it, with the motion, the background light pattern movement, and the diffraction spikes. The observables are just more measurable. I'm open to being proven wrong. I see Marik is tweeting about me panicking as my life's work collapses around me, which is a bit odd. I largely lost interest in this years ago, as I think it went about as far as it could. It's fun to revisit from time to time, and I may at some point shift it all over to the more general Sitrec framework for some more experiments as the available tools improve. But right now, I think if you want to declare it to not be a glare, you've got a significant challenge in explaining your determination to anyone who has not spent years looking at it. Zaine, I admire your persistence, but I think you keep getting things wrong in very complicated ways. I just don't think it's worth my spending many hours trying to unravel and confirm or rebut that when nobody else understands what it is I'd be confirming or rebutting. I anticipate having more free time around April, maybe I'll revisit it then. But, as I said three years ago: No rush though. Here's my original analysis. Aside from some uncertainty about a small part of the initial rotation, everything else is still valid. youtube.com/watch?v=qsEjV8…

Playing with AI slop for a non-existant CPU performance tool, now I wish it was real.

@Bias_Zebra @lemire What a crazy assumption, I dont want everyone poor, I want incentive for people to do things, I just dont want extreme wealth inequality. I am for capitalism with reasonable taxes and suppprt systems for everyone. A rich country shouldnt have starving homeless kids.

@reductor @lemire Yes leftists would rather everyone be equally poor, we already know this

Compare these two outcomes: 1. Everyone becomes twice as a rich, except for the richest 1% who become 10x richer. 2. Everyone remains more or less the same, no progress. The story we have been told, repeatedly, is that scenario (1) is bad as it increases relative differences between the richest and poorest people. And that these differences affect people's health, sense of well-being and so forth. It is definitively true that some people are, by nature, envious. We all rare, to a point. I am envious of Linus Torvalds for having invented Linus. What was I doing ? I could have done that! But my envy does not make me unhappy. If it did, it would be a mental disease. Yes, if Joe has enough to eat and you do not, then it is a natural instinct to go after Joe. But in scenario (1), there is plenty to eat for everyone. When a friend of mine is doing well... earning a lot of money... I am happy. It is maybe time to retire this meme that inequality of outcomes is inherently bad.

@britcatala @lemire Inflation is caused by many things, doubling everyones income and 10x for the rich is essentially printing money. However in this case the inflation would be part of the wage-price spiral which is a macroeconomics theory where rising prices lead to higher prices.

@reductor @lemire Inflation is caused by printing money, not by producing more. Your point is against central banks, not company owners.

@iyoushetwt And probably are OOP, time to learn OOP?

>Linux is C >Git is C >Python interpreter (CPython) is C >Ruby interpreter (MRI) is C >PostgreSQL is C >SQLite is C >Redis is C >MySQL is C++ >MongoDB is C++ >Unreal Engine is C++ >Chrome is C++ >Firefox is C++ >Windows kernel is C >macOS kernel (XNU) is C >Photoshop is C++ >VMware is C++ >TensorFlow (Core) is C++ -still, you are not convinced to learn C and C++

@rajaboys @iyoushetwt I used to always use EditPlus, just checked its still around and last updated today.

@iyoushetwt EditPlus – my primary code editor during the early stages of my career for HTML template conversion and PHP development.

What was the first code editor you ever used? Mine was Sublime Text

@benjoffe What is your use case? It seems strange that this would be a bottleneck

Counting dates backwards for a 30-40% speed gain: benjoffe.com/fast-date-64

My @GCAPConf talk on Lessons Learned from two decades of writing bad code has gone up on YouTube. I introduce a new way to iterate Linked Lists that is twice as fast as the traditional approach without changing the data structure. youtu.be/1YrenWdSeO0

@geofflangdale I am conflicted if its good/bad to recommend his content, he is knowledgable about many things he discusses and has some good content for people to learn from in an area that there isnt much, then some is strawman arguments to his followers that dont see the strawmaning like OOP

@reductor Yup - 5 years, and imo aside from Granny3D, not the lead on the big name projects (Bink/Oodle) that he mentions constantly in his bios. I just constantly see people quoting his weird stuff on OO as if it's gospel and not the opinion of some random dude.

Idle curiosity: how many people worked at RAD Game Tools during the peak Bink/Oodle period, and what was Casey Muratori's contribution to their flagship products?

@feliperybas @valigo Its really isnt a must watch, its a strawman argument. With a contrived example and benchmark which will intentionally make the "clean code" slow. Combined with saying clean code is something that it is not, just like OOP often gets misrepresented as something it is not.

>commits most of OOP sins >getters and setters >classes with inheritance >loves managers >loves singletons >loves small functions >uses STL >still sold 6-7 million copies of his games. Is OOP good after all, or it doesn't matter what you do if end product is good?

@unclebobmartin People who lookup components by entity id frequently on ECS and think thats how you use it are often using it incorrectly and can end up just as slow as naieve OOP.

@unclebobmartin I dont hate OOP you can write fast OOP just like you can write slow ECS code, neither are magic bullets. SIMD use cases are small, arrays are the powerhouse you aim to do bulk operations on one or more arrays, helping hardware prefetching, less load data dependencies, etc.

Morning bathrobe rant: ECS