-
Stay up-to-date with everything C++! Content directly fetched from the subreddit just for you. Join our group for discussions : @programminginc Powered by : @r_channels
Is this a false positive from Helgrind?
Is the following code thread-safe according to the C++ standard?
#include <stdexcept>
#include <thread>
void f() {
try { throw std::exception(); }
catch (std::exception& e) { }
}
int main() {
std::thread t1(f);
std::thread t2(f);
t1.join();
t2.join();
return 0;
}
Compiling it as g++ -O3 main.cpp and running in Helgrind (valgrind --tool=helgrind ./a.out) warns about possible data races in __cxa_throw:
Possible data race during read of size 1 at 0x4C33087 by thread #3
Locks held: none
at 0x4C283AC: UnwindSetSpColumn (unwind-dw2.c:1130)
by 0x4C283AC: uwinitcontext1 (unwind-dw2.c:1351)
by 0x4C28958: UnwindRaiseException (unwind.inc:93)
by 0x493F2A6: cxathrow (ehthrow.cc:93)
by 0x10911A: f() (in /tmp/a.out)
by 0x4972CF3: executenativethreadroutine (thread.cc:104)
by 0x485414B: mythreadwrapper (hgintercepts.c:406)
by 0x4CC839C: startthread (pthreadcreate.c:447)
by 0x4D4D2A3: clone (clone.S:100)
This conflicts with a previous write of size 1 by thread #2
Locks held: none
at 0x4C255D5: initdwarfregsizetable (unwind-dw2.c:1318)
by 0x4CCD8FA: pthreadonceslow.isra.0 (pthreadonce.c:116)
by 0x4CCD978: pthreadonce@@GLIBC2.34 (pthreadonce.c:143)
by 0x4C2839E: gthreadonce (gthr-default.h:713)
by 0x4C2839E: uwinitcontext1 (unwind-dw2.c:1341)
by 0x4C28958: UnwindRaiseException (unwind.inc:93)
by 0x493F2A6: cxathrow (ehthrow.cc:93)
by 0x10911A: f() (in /tmp/a.out)
by 0x4972CF3: executenativethreadroutine (thread.cc:104)
Address 0x4c33087 is 7 bytes inside data symbol "dwarfregsizetable"
Am I safe to assume that these are false positives?
https://redd.it/1ebgmv1
@r_cpp
What's so hard about constexpr allocation?
Barry Revzin wrote up a nice overview of the challenges of extending dynamic allocation during constant evaluation beyond the capabilities introduced in C++20:
https://brevzin.github.io/c++/2024/07/24/constexpr-alloc/
https://redd.it/1ebcnad
@r_cpp
Different types of classes in one list.
Is it possible to have a list that contains say a class called Human and another one called Animal
https://redd.it/1eb8fb0
@r_cpp
Had anyone used Clang's -fexperimental-modules-reduced-bmi option with CMake?
LLVM 19 had been branched yesterday, and it initially supports the one-phase module compilation model and reduced bmi.
I'm wondering if it can relax the current CMake's same compilation rule (mentioned in https://crascit.com/2024/04/04/cxx-modules-cmake-shared-libraries/: module and its importing target have the same compile options in Clang. This constraint does not hold on MSVC AFAIK)
Additionally, can it be turned on by just call add_options(-fexperimental-modules-reduced-bmi)?
https://redd.it/1eb32zv
@r_cpp
Possible idea for friends?
Friendship in C++ is a useful tool to have, but what if there was a way to promote "friends of friends"? E.g. Class A allows (trusts) Class B to see internals. Would it be useful to have a way for Class B to then "vouch for" another class as a friend to Class A?
I know it's most definitely an assinine idea with heaps of issues I'm too lazy to explore right now, bit what are your thoughts?
https://redd.it/1eavbgm
@r_cpp
Is it true that C++ is becoming more and more like JavaScript?
I came across a quote that says:
>
The more I study c++ the more I think that it is becoming more and more like javascipt.
I want to know if there is any truth to the above quoted statement.
https://redd.it/1eaw5iz
@r_cpp
Anyone having success with C++23 modules and CMake? in 2024
I am having a hard time getting modules working on my Arch Linux system. Right now, either import std; fails, or it cannot fins a number of header files in /usr/include because the header files are located in a different place in the directory tree.
For example, it looks for stddef.h at:/usr/include/stddef.h
But it is not located there. Instead:find /usr/include -name stddef.h /usr/include/linux/stddef.h
And this is going on with a number of header files.
It was tricky to configure CMakeLists.txt for module support.
If anyone has this working on Linux without a hitch, I am all ears. If not, I will have to go back to those pesky header files.
https://redd.it/1eafzjs
@r_cpp
Do you consider LLVM codebase to be Modern C++?
View Poll
https://redd.it/1ead8xr
@r_cpp
What is the best/are the best option(s) for multiple precision arithmetic libraries in C++ that support trigonometric functions/complex numbers?
Note: In the case that this is not a C++ specific enough question, I would ask that you refer me to a more appropriate subreddit (I know that the rules state that questions are off-topic, but this question is that much about C++ as it is about existing libraries for it for a specific problem).
I am performing arithmetic computations with complex numbers that are pushing the numeric limits of standard C++. The computations themselves are not cryptographic oriented, so security aspect of the computations is not important. I am looking for the fastest high/arbitrary precision integer/floating point arithmetic library for C++ that provides support for trigonometric functions and complex numbers in the sense that you can use the said functions/structures with the extended integer/floating point structures that the library provides, and not have a situation where an extended arithmetic type is narrowed down to some similar C++ data type during the call to some trigonometric function.
I was led to believe that the GMP library: https://gmplib.org/ is an extremely fast choice for arbitrary precision arithmetic, but I have some trouble determining whether is can provide what I want. To be specific, as I am reading through the manual it is unclear to me whether I can use trigonometric functions/complex numbers with the extended arithmetic structures of the library.
Some other sources mentioned the GNU MPC library: https://www.multiprecision.org/, but I have no idea how that fares to GMP or any other high/arbitrary precision arithmetic library. Wikipedia lists multiple C++ compatible libraries: https://en.wikipedia.org/wiki/List\_of\_arbitrary-precision\_arithmetic\_software, but again I have no idea how they compare to each other or w.r.t. GMP.
In any case, what would you recommend for my situation? Or can you tell me whether GMP (or MPC) is exactly what I am looking for?
Thanks!
https://redd.it/1ea4upx
@r_cpp
Read from a pipe in windows
Hello All,
I am a very new C++ developer. I aim to read data from a pipe.
Let me give an overview, in my application I have 2 processes, parent and child. The child sends data to parent and parents waits for the data. This read call is blocking in nature due to which my main application gets hanged, and I can’t gracefully close it. I tried to create another thread and have the read block call in this background thread. But this also hangs my application, and thread.join() gets hanged.
In this situation also, I am unable to close my application gracefully.
Any help is appreciated, thanks!
https://redd.it/1e9w1ts
@r_cpp
I want code format on clion new version
I used reformatting code on clion old version with Ctrl+Alt+L.
but after I updated clion to newest version with clion nova, reformatting code does not work.
I can use it with google code style when I turn off nova function, but I don't want to turn off the new function.
Is there a way to reformat code with nova on?
https://redd.it/1e9gfkv
@r_cpp
New C++ Conference Videos Released This Month - July 2024 (Updated To Include Videos Released 07/15/2024 - 07/21/2024)
This month the following C++ videos have been published to YouTube. A new post will be made each week as more videos are released
**ACCU Conference**
07/15/2024 - 07/21/2024
* NASA Apollo 11: To the Moon and Back With 4KB of RAM - Adam Smart - [https://youtu.be/xxxB4sW8RYg](https://youtu.be/xxxB4sW8RYg)
* The Fight for Right to Repair: From Seed to Sabotage - Joshua Webster - [https://youtu.be/Znttrm2QQOA](https://youtu.be/Znttrm2QQOA)
* Work-Work Balance - Organisational Culture Matters - Will Woods - [https://youtu.be/eJv0touFMgg](https://youtu.be/eJv0touFMgg)
* Keynote: Reflection in C++26 - Welcome to the meta::\[\[verse\]\]! - Inbal Levi - [https://youtu.be/G4i45R7sX8I](https://youtu.be/G4i45R7sX8I)
* Safer C++ with MISRA-C++-2023 - Peter Sommerlad - [https://youtu.be/oCZ1Rn-4AQE](https://youtu.be/oCZ1Rn-4AQE)
07/08/2024 - 07/14/2024
* Amplifying Black Women's Voices - Building Inclusive Tech Spaces Beyond Tokenism - Yemi Adenekan - [https://youtu.be/0makY11qYs8](https://youtu.be/0makY11qYs8)
* Shaping Perception - The Evolution of 3D Graphics and Cultural Transformation - Ben Kyd - [https://youtu.be/0T7r6zWWUBw](https://youtu.be/0T7r6zWWUBw)
* Gamification - Falling Down the Leader Board? - Char Ryder - [https://youtu.be/qO5sxo\_liR8](https://youtu.be/qO5sxo_liR8)
* Understanding the Filter View to Use It Right - Nicolai M. Josuttis - [https://youtu.be/jR3WE-hAhCc](https://youtu.be/jR3WE-hAhCc)
* When Less Is More: Designing & Decoding Complex Distributed Systems - Jessica Winer & Jacqueline Pan - [https://youtu.be/hz1CrNSz4OE](https://youtu.be/hz1CrNSz4OE)
07/01/2024 - 07/07/2024
* How to Maximize the Benefits of Neurodiversity in Software Development Teams and Support Neurodivergent Developers - Dom Davis - [https://youtu.be/bt25w\_O05KI](https://youtu.be/bt25w_O05KI)
* Future of Software Programming with AI Bots at Hand - Amir Kirsh - [https://youtu.be/bz7M6rOKZLo](https://youtu.be/bz7M6rOKZLo)
* Contracts for C++ - Timur Doumler - [https://youtu.be/lHRyuTFig0U](https://youtu.be/lHRyuTFig0U)
**C++Now**
07/15/2024 - 07/21/2024
* An Adventure in Modern C++ Library Design - Robert Leahy - [https://youtu.be/W29fY7Ml4-w](https://youtu.be/W29fY7Ml4-w)
* Fast Conversion From Cpp Floating Point Numbers - Cassio Neri - [https://youtu.be/w0WrRdW7eqg](https://youtu.be/w0WrRdW7eqg)
* Reflection Is Good for C++ Code Health - Saksham Sharma - [https://youtu.be/GQ5HKL0WRGQ](https://youtu.be/GQ5HKL0WRGQ)
07/08/2024 - 07/14/2024
* C++ Reflection - Back on Track - David Olsen - [https://youtu.be/nBUgjFPkoto](https://youtu.be/nBUgjFPkoto)
* C++ Coroutines at Scale - Implementation Choices at Google - Aaron Jacobs - [https://youtu.be/k-A12dpMYHo](https://youtu.be/k-A12dpMYHo)
* Keynote: Employing Senders & Receivers to Tame Concurrency in C++ Embedded Systems - Michael Caisse - [https://youtu.be/wHmvszK8WCE](https://youtu.be/wHmvszK8WCE)
**C++Online**
07/08/2024 - 07/14/2024
* Lightning Talk: How I Became a C++ Mentor the "Hard" Way - Paulo Chiliguano - [https://youtu.be/dLobaazrvYg](https://youtu.be/dLobaazrvYg)
* Lightning Talk: Coccinelle for C++ - Julia Lawall - [https://youtu.be/eLyyfXMtR9o](https://youtu.be/eLyyfXMtR9o)
* Lightning Talk: Disentangling The Mikado Method When Refactoring C++ Legacy Code - Phil Nash - [https://youtu.be/W\_v0pvr9pT4](https://youtu.be/W_v0pvr9pT4)
07/01/2024 - 07/07/2024
* Lightning Talk: Keeping Up to LLVM IR: An Introduction to LLVM IR - Isaac David Bermudez Lara - [https://youtu.be/5FbhwOxoJsg](https://youtu.be/5FbhwOxoJsg)
* Lightning Talk: Use SIMD Pairwise Reduction in C++! - Andrew Drakeford - [https://youtu.be/QYkU0u9M9Mg](https://youtu.be/QYkU0u9M9Mg)
* Lightning Talk: Why False Is Less Than True in Boolean Algebra - Ben Deane - [https://youtu.be/vAwYP0Ox96I](https://youtu.be/vAwYP0Ox96I)
**Audio Developer Conference**
07/15/2024 - 07/21/2024
* Digital Modelling of the Roland RE-201 - Jordan Evans -
https://x.com/krisjusiak/status/1815395887247471019
https://redd.it/1e9gdm5
@r_cpp
Any known issues using Microsoft c++ with fopen/FILE/fseek etc?
I am compiling a C++ code that runs fine on Linux and MacOS but fails when manipulating binary files of about 10GB.
I do not have a handle on Windows programming and can barely run the command line compile, cl version 19.16.27026.1 on Windows 10.
Would there be any known reason why such a code might hang or crash when the code seems to work fine on other systems (and on smaller files)?
https://redd.it/1e9dgjp
@r_cpp
Why doesn't std::piecewise_construct work on std::tuple anymore? Didn't it used to?std::piecewise_construct works with std::pair and some others to construct in place with given parameters as tuples, but it doesn't work with std::tuple itself. Is this an oversight or should this be addressed before the next release? I distinctly recall it working before but have no idea how it would have gone away. Is this obvious or am I making this sound too weird?
https://redd.it/1ebdtg3
@r_cpp
Is there a standard for structuring a library?
I've been building complex linear algebra library with vectors, points and all that jazz. So far it has been very difficult to test and check issues within a lot of operator overloading, cases, methods.... so I've defined a tracer and divided each type in this structure:obj (dir) -> obj.hpp (main header file that defines a type with consts, statics, virtuals ...) -> obj_constructors.cpp ( all copy, move , casting back and forth... )-> obj_methods.cpp ( internal functions required for maintainanace of object, if required)-> obj_operators.cpp ( ALL operators and different type operator overloading)-> obj_functions.cpp ( functions that are tied to setup and change of state of objects)
So when error/warning is triggered, tracer shows me file, path, function and line number where it happened and it which was function that called it. My project so far has 55 files , and this has helped me a lot in navigating and fixing ish i've done but I am worrying if this can bite me in the ass.
I've tried to find similar libraries on github to find at least some standard that I can rely on. Most of other project have one `.hpp` file with all of the code with endless lines and Ive tried that but i've found it is very difficult to keep track of all dependencies and jumps in the code. Only way I was able to read it was with debugger, which meant I had to compile it and go function by function.
So I ask you, kind people of Reddit, is there any large project standard that allows readability as well as troubleshooting?
https://redd.it/1eb7bs8
@r_cpp
It's kinda odd that C++ doesn't have computed goto
As a hobby, I maintain a programming language implementation. Today I refactored the bytecode interpreter to use jump threading instead of visiting the instructions with a switch.
I used GCC's computed goto extension, which luckily is also available in Clang.
The speedups are consistently between 150 and 200% across my benchmarks, so I really don't want to go back to the switch.
But I also like my code to be standard conformant C++. And I have to keep the switch based implementation around to continue supporting MSVC, which as far as I know has no equivalent to computed gotos.
I don't really have a question or anything, and I know this is a niche application, but still I find it odd, that in this case, C++ doesn't seem to care about its promise "to leave no room for a faster, lower level language" (I think Bjarne once said that)
https://redd.it/1eb569m
@r_cpp
Do you consider LLVM codebase to be Modern C++? (C++11 and above)
View Poll
https://redd.it/1eazuu6
@r_cpp
Communication with my RPI robot
So I've been messing around with TCP async streams for communication but I'm very new to networking and especially in C++ (this entire project is C++). And I've been struggling very hard with all this I've been working on it for about 2 weeks now.
Right now the ideal thing for me would be to stream/transmit a Xbox controller object from my client to my server to read all of the axis and buttons and have my server stream encoders, voltages, and current back to the client for monitoring
I really haven't been able to get anything substantial working with my current code just a simple async TCP server and client so I'm open to setting up a different communication method
(I've been looking into "LoRa transceivers" that would work well I just don't know where to get started if I were to do that)
like I said being able to stream/transmit the Xbox controller object or a serial port from the client to the sever would be ideal. Then I assume the best way to get the data back to the client would be to have a table or JSON file streamed/updated constantly then read on the client side
any info or point in the right direction is appreciated
https://redd.it/1eaw9el
@r_cpp
Toy Compiler With C(++)
nevo.krien/toy-compiler-with-c-3d13279c5113" rel="nofollow">https://medium.com/@nevo.krien/toy-compiler-with-c-3d13279c5113
https://redd.it/1eaomgc
@r_cpp
Best way to Load and Save DDS Texture files?
With functions like "save_dds(&pixels)" and "load_dds(&pixels)".
I need dtx1 and dtx5 compression, and fortunately no mip maps yet but I'd like to have the option anyway. I want to do some hard-core pixel manipulation on textures for games I'm trying to mod.
I noticed the majority of plausible possibilities were all in c++ (not c). That being said, what's the best place to start? DirectX? DirectDraw? OpenGL? Some third party thing? I noticed most online solutions involve me populating a giant file format structure specifying the width, height, pixel format, and a million other settings for the texture file. And they're all written in c++. All well and good for me but I have no idea where to start!
So yeah, any suggestions?
NOTE: When I dump the textures, they look like they've been cut into small pieces with scissors and tapped back together into a mangled quilt of the original image. Is it possible these are textures for a 3d skin maybe? It's a 2d game, but it's very flash-like in that it heavily uses image transformations for movement, so IDK. Maybe there's a better tool for what I'm trying to do?
I've been using Special K to inject textures into the game. I have no idea what else it's capable of.
https://redd.it/1eah0j6
@r_cpp
Seeking Guidance on Transitioning to Core C++ and Changing Jobs: Tips and Timeline?
Hi everyone,
I'm a C++ developer with two years of experience in both native C++ and managed C++/CLI. My current role involves creating desktop applications using a combination of C++ (with MFC for the UI) and managed C++. I understand that this isn't core C++, but it's been my focus at my current company.
I'm eager to transition into a role that involves more low-level programming or other fields that utilize core C++. Additionally, I'm looking to switch companies and would greatly appreciate any advice on how to go about this.
Specifically, I have a few questions:
1. Transitioning to Core C++:
What steps should I take to move from managed C++ to core C++?
Are there any particular areas of core C++ that I should focus on to make myself more marketable?
2. Job Preparation:
How should I prepare for job interviews focused on core C++ roles?
What resources (books, online courses, projects) would you recommend to build my skills in core C++?
3. Job Switching Timeline:
Based on your experiences, how long might it take to switch from a managed C++ role to a core C++ role?
Any tips on navigating the job market and finding opportunities that match my new focus?
Thank you all in advance for your suggestions and insights!
https://redd.it/1eaclll
@r_cpp
@Microsoft: When will C++23 Compiler implementation start?
Is there an approximate plan when the implementation of C++23 in MSVC will start? The STL is already well advanced, but nothing has happened with the compiler for months. If I remember correctly, it was mentioned somewhere that they have been focussing on bug fixing lately. That's great and is also noticeable in daily work. Nevertheless, I am eagerly awaiting the DR C++20 implementation of P2564R0. (consteval needs to propagate up)
https://redd.it/1ea6gho
@r_cpp
Interview with Dave Rowland: Architect and Lead C++ Developer of Tracktion's Waveform (digital audio workstation software for mixing and mastering audio and music production) and a Software Engineering Manager at Prism Sound
https://thewolfsound.com/talk023/
https://redd.it/1ea1162
@r_cpp
Counting Bytes Faster Than You'd Think Possible
https://blog.mattstuchlik.com/2024/07/21/fastest-memory-read.html
https://redd.it/1e9lgvd
@r_cpp
[https://youtu.be/gqdTI0QYmEQ](https://youtu.be/gqdTI0QYmEQ)
* Native Instruments NKS2: Building a Physical Experience for Virtual Instruments - Moritz Heppner & Tim Adnitt - [https://youtu.be/xdBVMqM9a\_s](https://youtu.be/xdBVMqM9a_s)
07/08/2024 - 07/14/2024
* Why You Shouldn’t Write a DAW - David Rowland - [https://youtu.be/GMlnh6\_9aTc](https://youtu.be/GMlnh6_9aTc)
* Inference Engines and Audio - Harriet Drury - [https://youtu.be/CcSi\_Lgxz-4](https://youtu.be/CcSi_Lgxz-4)
* Building an Accessible Juce App - Harry Morley - [https://youtu.be/wrZyRMxLlx0](https://youtu.be/wrZyRMxLlx0)
07/01/2024 - 07/07/2024
* Creating Ubiquitous, Composable, Performant DSP Modules - Stefano D’Angelo - [https://youtu.be/lKhVtbnILDk](https://youtu.be/lKhVtbnILDk)
* Running High Channel Count Audio Applications on Linux RT - Olivier Petit - [https://youtu.be/1TgzjOoiqIs](https://youtu.be/1TgzjOoiqIs)
* Odd Challenges of Using Deep Learning in Designing a Feedback Delay Network Reverb - ADC23 - [https://youtu.be/5URLvwFmlb0](https://youtu.be/5URLvwFmlb0)
https://redd.it/1e9he84
@r_cpp
New C++ Conference Videos Released This Month - July 2024 (Updated To Include Videos Released 07/15/2024 - 07/21/2024)
This month the following C++ videos have been published to YouTube. A new post will be made each week as more videos are released
ACCU Conference
07/15/2024 - 07/21/2024
NASA Apollo 11: To the Moon and Back With 4KB of RAM - Adam Smart - [https://youtu.be/xxxB4sW8RYg](https://youtu.be/xxxB4sW8RYg)
The Fight for Right to Repair: From Seed to Sabotage - Joshua Webster - https://youtu.be/Znttrm2QQOA
Work-Work Balance - Organisational Culture Matters - Will Woods - [https://youtu.be/eJv0touFMgg](https://youtu.be/eJv0touFMgg)
Keynote: Reflection in C++26 - Welcome to the meta::[[verse\]\]! - Inbal Levi - https://youtu.be/G4i45R7sX8I
Safer C++ with MISRA-C++-2023 - Peter Sommerlad - [https://youtu.be/oCZ1Rn-4AQE](https://youtu.be/oCZ1Rn-4AQE)
07/08/2024 - 07/14/2024
Amplifying Black Women's Voices - Building Inclusive Tech Spaces Beyond Tokenism - Yemi Adenekan - https://youtu.be/0makY11qYs8
Shaping Perception - The Evolution of 3D Graphics and Cultural Transformation - Ben Kyd - [https://youtu.be/0T7r6zWWUBw](https://youtu.be/0T7r6zWWUBw)
Gamification - Falling Down the Leader Board? - Char Ryder - https://youtu.be/qO5sxo\_liR8
Understanding the Filter View to Use It Right - Nicolai M. Josuttis - [https://youtu.be/jR3WE-hAhCc](https://youtu.be/jR3WE-hAhCc)
When Less Is More: Designing & Decoding Complex Distributed Systems - Jessica Winer & Jacqueline Pan - https://youtu.be/hz1CrNSz4OE
07/01/2024 - 07/07/2024
How to Maximize the Benefits of Neurodiversity in Software Development Teams and Support Neurodivergent Developers - Dom Davis - [https://youtu.be/bt25w\_O05KI](https://youtu.be/bt25w_O05KI)
Future of Software Programming with AI Bots at Hand - Amir Kirsh - https://youtu.be/bz7M6rOKZLo
Contracts for C++ - Timur Doumler - [https://youtu.be/lHRyuTFig0U](https://youtu.be/lHRyuTFig0U)
C++Now
07/15/2024 - 07/21/2024
An Adventure in Modern C++ Library Design - Robert Leahy - https://youtu.be/W29fY7Ml4-w
Fast Conversion From Cpp Floating Point Numbers - Cassio Neri - [https://youtu.be/w0WrRdW7eqg](https://youtu.be/w0WrRdW7eqg)
Reflection Is Good for C++ Code Health - Saksham Sharma - https://youtu.be/GQ5HKL0WRGQ
07/08/2024 - 07/14/2024
C++ Reflection - Back on Track - David Olsen - [https://youtu.be/nBUgjFPkoto](https://youtu.be/nBUgjFPkoto)
C++ Coroutines at Scale - Implementation Choices at Google - Aaron Jacobs - https://youtu.be/k-A12dpMYHo
Keynote: Employing Senders & Receivers to Tame Concurrency in C++ Embedded Systems - Michael Caisse - [https://youtu.be/wHmvszK8WCE](https://youtu.be/wHmvszK8WCE)
C++Online
07/08/2024 - 07/14/2024
Lightning Talk: How I Became a C++ Mentor the "Hard" Way - Paulo Chiliguano - https://youtu.be/dLobaazrvYg
Lightning Talk: Coccinelle for C++ - Julia Lawall - [https://youtu.be/eLyyfXMtR9o](https://youtu.be/eLyyfXMtR9o)
Lightning Talk: Disentangling The Mikado Method When Refactoring C++ Legacy Code - Phil Nash - https://youtu.be/W\_v0pvr9pT4
07/01/2024 - 07/07/2024
Lightning Talk: Keeping Up to LLVM IR: An Introduction to LLVM IR - Isaac David Bermudez Lara - [https://youtu.be/5FbhwOxoJsg](https://youtu.be/5FbhwOxoJsg)
Lightning Talk: Use SIMD Pairwise Reduction in C++! - Andrew Drakeford - https://youtu.be/QYkU0u9M9Mg
Lightning Talk: Why False Is Less Than True in Boolean Algebra - Ben Deane - [https://youtu.be/vAwYP0Ox96I](https://youtu.be/vAwYP0Ox96I)
Audio Developer Conference
07/15/2024 - 07/21/2024
Digital Modelling of the Roland RE-201 - Jordan Evans -
[C++20] Zero/Minimal overhead static/const branching
Hardware branch prediction is very powerful (https://www.intel.com/content/www/us/en/developer/articles/technical/intel-sdm.html) but it's also not perfect and costly misdirections may happen (http://ithare.com/infographics-operation-costs-in-cpu-clock-cycles/).
Branchless computing (https://www.youtube.com/watch?v=g-WPhYREFjk) is a technique which may help a lot with mitigating the overhead of branching.
Alternative way, coming from linux kernel (https://docs.kernel.org/staging/static-keys.html) is to leverage knowledge about branches direction and literally change the instructions via run-time code patching to avoid the potential branching overhead all together.
When to go for such extreme solution? When performance really matters:
- `and` branches are known at compile-time
- `or` branches are not changing often at run-time
- `and/or` branches are expensive to compute/require memory access
- `and/or` branches are hard to learn by the hardware branch predictor due to their random nature
Some examples include logging, tracing, configuration, hot-path, algo, etc.
> https://github.com/boost-ext/sb - x86-64/linux/gcc/clang moves the solution to user space (for run-time) as well as to compile-time with C++20.
The following is a walkthrough of the run-time solution via code patching and the compile-time via stateful meta-programming.
> static_branch - Minimal overhead run-time branch (https://godbolt.org/z/797WeYYGK)
void fun() { // can be inlined/constexpr or not
if (sb::static_branch<"semi run-time branch">::get()) {
std::puts("taken");
} else {
std::puts("not taken");
}
}
int main() {
fun(); // not taken
sb::static_branch<"semi run-time branch">::set(true);
fun(); // taken
sb::static_branch<"semi run-time branch">::set(false);
fun(); // not taken
}
main: // $CXX -O3
lea rdi, [rip + .L.str.1]
nop # code patching (nop->nop)
lea rdi, [rip + .L.str.2]
.Ltmp1:
call puts@PLT # not taken
call static_branch<"semi run-time branch">::set(true) # relatively slow
lea rdi, [rip + .L.str.1]
jmp .Ltmp2 # code patching (nop->jmp)
lea rdi, [rip + .L.str.2]
.Ltmp2:
call puts@PLT # taken
call static_branch<"semi run-time branch">::set(false) # relatively slow
lea rdi, [rip + .L.str.1]
nop # code patching (nop->nop)
lea rdi, [rip + .L.str.2]
.Ltmp3:
call puts@PLT # not taken
.L.str.1: .asciz "taken"
.L.str.2: .asciz "not taken"
> const_branch - Zero overhead compile-time branch (https://godbolt.org/z/xEboP8WYe)
template<auto tag = []{}> // Note: `tag` is required to delay `fun` instantiation
auto fun() {
if constexpr (sb::const_branch<"compile-time branch">::get<tag>()) {
std::puts("taken");
} else {
std::puts("not taken");
}
}
int main() {
fun(); // not taken
static_assert(sb::const_branch<"compile-time branch">::set<true>());
fun(); // taken
static_assert(sb::const_branch<"compile-time branch">::set<false>());
fun(); // not taken
}
main: // $CXX -O3
lea rbx, [rip + .L.str.1]
mov rdi, rbx
call puts@PLT # not taken
lea rdi, [rip + .L.str.2]
call puts@PLT # taken
mov rdi, rbx
call puts@PLT # not taken
xor eax, eax # return 0
ret
.L.str.1: .asciz "not taken"
.L.str.2: .asciz "taken"
More info about how does it work under the hood - https://github.com/boost-ext/sb?tab=readme-ov-file#faq
Acknowledgments
- https://docs.kernel.org/staging/static-keys.html
- https://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html
- https://www.intel.com/content/www/us/en/developer/articles/technical/intel-sdm.html
- https://www.agner.org/optimize/instruction_tables.pdf
- https://www.felixcloutier.com/x86
- https://uops.info/table.html
- https://arxiv.org/abs/2308.14185
Updates -
I, Equality Operator | consteval
https://consteval.ca/2024/07/22/equality/
https://redd.it/1e98wu3
@r_cpp