The sad state of debug performance in C++
-
In this article, we’ll explore how C++’s abstraction model heavily relies on compiler optimizations, unveiling some unfortunate examples of unexpected performance loss.
So don't bother debugging
-
In this article, we’ll explore how C++’s abstraction model heavily relies on compiler optimizations, unveiling some unfortunate examples of unexpected performance loss.
So don't bother debugging
An interesting article written from the perspective of game development, where debugging is a pain when dealing with a slow debug build. My complaint is the opposite, namely that optimized release builds are basically impossible to debug. If you need to debug them, you need to disable many of the optimizations. I would also claim that many systems are designed with hardly any consideration of CPU, memory, and disk usage. The hardware crowd makes it faster and bigger, and then the software crowd pisses it all away. I hardly ever play computer games, but my guess is that these guys are also wasting resources trying to outdo the competition when it comes to how real a game looks. Maybe they need to back off a little instead of whining about the performance of their debug builds.
Robust Services Core | Software Techniques for Lemmings | Articles
The fox knows many things, but the hedgehog knows one big thing. -
In this article, we’ll explore how C++’s abstraction model heavily relies on compiler optimizations, unveiling some unfortunate examples of unexpected performance loss.
So don't bother debugging
Vittorio Romeo wrote:
I also don’t have any proof of this, but I suspect that writing low-level code with the desire of optimizing the debugging experience ironically ends up increasing the frequency of debugging.
Very interesting observation. Well written article talking about the tradeoffs between clean code and debugging.
