4 ms·

I remember this was a holy war amongst programmers (even myself!) - readability OR performance. With tools like ChatGPT - we can have both readability AND perf

by acl777 3y ago

I remember this was a holy war amongst programmers (even myself!) - readability OR performance. With tools like ChatGPT - we can have both readability AND performance, right?? Or am I missing something?

circuit10 3y ago
If you mean automatically optimising code, the reliability isn’t quite there (it can’t check its own code) and the limited context length means it might struggle to reason about the codebase as a whole
- mlajtos 3y ago
  Could you be the human judge please? https://mlajtos.mu/posts/no-code-clean-performance https://mlajtos.mu/posts/no-code-clean-performance
pjc50 3y ago
Correctness? I've not seen anyone seriously attempting to benchmark chatgpt output, without heavily cherry picking it first.
jerf 3y ago
Dunno about "readibility" but I fully expect ChatGPT to produce vast swathes of code that nobody understands. While I'll cop to being more skeptical than the average HN denizen about ChatGPT, this isn't actually cynicism about ChatGPT, but about human cognition. If we do have something that produced large swathes of mostly correct code, it won't be long before the humans using it aren't even checking it anymore. Same reason that cars can do a little bit of assisting, and if they could do 100% of the driving that might be safe, but 98% of the driving is not a very good idea at all. In 20 years, someone will ask CodeGPT 8.3 to explain what some code does, and it'll give a perfectly understandable explanation back, except the human still won't understand it because the human doesn't actually understand how computers work anymore. They'll think they understand it, though.
- dragonelite 3y ago
  That's why i think the copilot and chat gpt hype will end up as one big balloon and in ten years time we will have a pile of generated code no one really understands anymore that needs to be fixed. Kind of like after the out sourcing adventure people had in tech in the early 2010s or so.
mlajtos 3y ago
I explored this question a bit: https://mlajtos.mu/posts/no-code-clean-performance https://mlajtos.mu/posts/no-code-clean-performance
- CyberDildonics 3y ago
  You have spammed your site three times in this thread
hinkley 3y ago
ChatGPT has nothing to do with it. You can have code that is both easier to read and higher performance, it's just than nobody has made it a goal in life to show the rest of us how to do it. If I ever get fed up with writing code for a living I might sit down and figure out how to communicate it. There's dribs and drabs in my comment history, usually under topics of performance. The problem is that they've chosen a micro benchmark for this argument. The toy-est of toy problems. Yes, you have to make tradeoffs in these cases, but toy problems aren't really a good communication medium. If you want to process a thousand records of data in a decent amount of time, there are lots of strategies that move you toward achieving both.

by acl777 3y ago

circuit10 3y ago

If you mean automatically optimising code, the reliability isn’t quite there (it can’t check its own code) and the limited context length means it might struggle to reason about the codebase as a whole

mlajtos 3y ago
Could you be the human judge please? https://mlajtos.mu/posts/no-code-clean-performance https://mlajtos.mu/posts/no-code-clean-performance

pjc50 3y ago

Correctness? I've not seen anyone seriously attempting to benchmark chatgpt output, without heavily cherry picking it first.

jerf 3y ago

Dunno about "readibility" but I fully expect ChatGPT to produce vast swathes of code that nobody understands. While I'll cop to being more skeptical than the average HN denizen about ChatGPT, this isn't actually cynicism about ChatGPT, but about human cognition. If we do have something that produced large swathes of mostly correct code, it won't be long before the humans using it aren't even checking it anymore. Same reason that cars can do a little bit of assisting, and if they could do 100% of the driving that might be safe, but 98% of the driving is not a very good idea at all. In 20 years, someone will ask CodeGPT 8.3 to explain what some code does, and it'll give a perfectly understandable explanation back, except the human still won't understand it because the human doesn't actually understand how computers work anymore. They'll think they understand it, though.

dragonelite 3y ago
That's why i think the copilot and chat gpt hype will end up as one big balloon and in ten years time we will have a pile of generated code no one really understands anymore that needs to be fixed. Kind of like after the out sourcing adventure people had in tech in the early 2010s or so.

mlajtos 3y ago

I explored this question a bit: https://mlajtos.mu/posts/no-code-clean-performance https://mlajtos.mu/posts/no-code-clean-performance

CyberDildonics 3y ago
You have spammed your site three times in this thread

hinkley 3y ago

ChatGPT has nothing to do with it. You can have code that is both easier to read and higher performance, it's just than nobody has made it a goal in life to show the rest of us how to do it. If I ever get fed up with writing code for a living I might sit down and figure out how to communicate it. There's dribs and drabs in my comment history, usually under topics of performance. The problem is that they've chosen a micro benchmark for this argument. The toy-est of toy problems. Yes, you have to make tradeoffs in these cases, but toy problems aren't really a good communication medium. If you want to process a thousand records of data in a decent amount of time, there are lots of strategies that move you toward achieving both.