This ain’t the '90s anymore; CPUs have been fast for a while now. If your mid-development partial compilation time isn’t basically negligible with the CPUs we already have, your build script is probably fucked up or the module you’re working on is way too large. You should rarely be working on something with such cross-cutting concerns that you legitimately need to recompile vast swathes of the codebase at once.
I think it depends on what you’re working on. If you’re working on some JavaScript web app you could say that CPUs are “good enough”. But even then larger more complicated apps will get annoyingly slow to “compile”.
It’s when you are working with larger and more complicated Rust or C or whatever code bases that compile time matters.
This all being said for me CPU important is a good thing. It was good in the 90s and it’s good now.
But even then larger more complicated apps will get annoyingly slow to “compile”.
It’s when you are working with larger and more complicated Rust or C or whatever code bases that compile time matters.
If you are in the middle of doing a unit of work, iteratively making small changes to the code, compiling, and testing them, those compile times should be small too. If a small change in one file triggers your entire project to recompile, you fucked up the Makefile or structured the whole program poorly or something like that. There’s something wrong that a faster CPU will only mask, not fix.
C’mon, faster compilation never hurts. It’s not just build scripts - think of development where it eats plenty of seconds each time you start debugging.
In most cases, you should only need to recompile the particular file you’re working on because interfaces should be changing a lot less frequently than implementations.
Any single file should not be so large it takes a long time to compile by itself.
If other files are getting recompiled anyway even though nothing about them actually changed, the dependency resolution in your Makefile (or whatever) is screwed up and you need to fix it.
Point is, routine long compilation times after a small change are a code smell. There’s something wrong that a faster CPU will only mask, not fix.
you can often just slap compiler cache on a project and get a 20-150x speedup, but when the original compile time was 45 minutes, it’s still slow enough to disrupt your workflow (though, I suspect you may be talking about some manual method that may be even faster. But are those really common enough where you would call the lack of it a code smell?)
If the new chips were actually faster than than current line. i.e. Ryzen 9 5950X has higher frequencies and Ryzen 7 5800XT increases boost only by 0.1GHz over Ryzen 7 5800X while Ryzen 9 5900X has more cores and same boost clock.
I hardly see anybody upgrading due to 0.1GHz, I’m curious though, what is your rationale? Perhaps you have a slower CPU and you would upgrade since the prices came down?
I am speaking about in general. Newer faster CPUs are something worth being happy about.
This being said there is more to CPUs than raw clock speeds. A major component is IPC uplift. For example if a new cpu has a 20% IPC improvement over an older CPU when they both run at say 3.5Ghz the newer CPU will do things 20% faster. It’s pretty cool stuff.
In the case of the CPUs you mentioned it looks like there was a 16% IPC uplift over last gen and last gen had a 18% uplift over the 5xxx series. I would be happy to have my code compile 16% + 18% faster.
That would be great if that was the case. However you were probably mislead by the title - that IPC increase applies only to 9XXX beasts (Zen 5), not these two AM4 refreshes. At least that’s how I understand it.
I was not in fact mislead. I actually went and found that article to help explain the concept of IPC uplift to you.
I was speaking of why new CPUs are a good thing worth being excited by in general. You were the one who said new CPUs aren’t worth being excited about in general.
As a bonus I showed how new CPUs even at the same speed have advantages. This all being said hey if you want to be a downer here go for it.
For me they are. Faster compile times means I can iterate and test things faster.
If you do dev work on a day to day basis a newer faster cpu makes a major difference.
This ain’t the '90s anymore; CPUs have been fast for a while now. If your mid-development partial compilation time isn’t basically negligible with the CPUs we already have, your build script is probably fucked up or the module you’re working on is way too large. You should rarely be working on something with such cross-cutting concerns that you legitimately need to recompile vast swathes of the codebase at once.
I think it depends on what you’re working on. If you’re working on some JavaScript web app you could say that CPUs are “good enough”. But even then larger more complicated apps will get annoyingly slow to “compile”.
It’s when you are working with larger and more complicated Rust or C or whatever code bases that compile time matters.
This all being said for me CPU important is a good thing. It was good in the 90s and it’s good now.
If you are in the middle of doing a unit of work, iteratively making small changes to the code, compiling, and testing them, those compile times should be small too. If a small change in one file triggers your entire project to recompile, you fucked up the Makefile or structured the whole program poorly or something like that. There’s something wrong that a faster CPU will only mask, not fix.
C’mon, faster compilation never hurts. It’s not just build scripts - think of development where it eats plenty of seconds each time you start debugging.
I am thinking of development.
In most cases, you should only need to recompile the particular file you’re working on because interfaces should be changing a lot less frequently than implementations.
Any single file should not be so large it takes a long time to compile by itself.
If other files are getting recompiled anyway even though nothing about them actually changed, the dependency resolution in your Makefile (or whatever) is screwed up and you need to fix it.
Point is, routine long compilation times after a small change are a code smell. There’s something wrong that a faster CPU will only mask, not fix.
you can often just slap compiler cache on a project and get a 20-150x speedup, but when the original compile time was 45 minutes, it’s still slow enough to disrupt your workflow (though, I suspect you may be talking about some manual method that may be even faster. But are those really common enough where you would call the lack of it a code smell?)
Yep sure mate, have a go at an enterprise scale mobile app or. Net and then let’s chat
If the new chips were actually faster than than current line. i.e. Ryzen 9 5950X has higher frequencies and Ryzen 7 5800XT increases boost only by 0.1GHz over Ryzen 7 5800X while Ryzen 9 5900X has more cores and same boost clock. I hardly see anybody upgrading due to 0.1GHz, I’m curious though, what is your rationale? Perhaps you have a slower CPU and you would upgrade since the prices came down?
I am speaking about in general. Newer faster CPUs are something worth being happy about.
This being said there is more to CPUs than raw clock speeds. A major component is IPC uplift. For example if a new cpu has a 20% IPC improvement over an older CPU when they both run at say 3.5Ghz the newer CPU will do things 20% faster. It’s pretty cool stuff.
In the case of the CPUs you mentioned it looks like there was a 16% IPC uplift over last gen and last gen had a 18% uplift over the 5xxx series. I would be happy to have my code compile 16% + 18% faster.
https://www.pcgamer.com/hardware/processors/amd-announces-zen-5-and-the-ryzen-9000-series-with-a-16-ipc-uplift-shipping-in-july/
That would be great if that was the case. However you were probably mislead by the title - that IPC increase applies only to 9XXX beasts (Zen 5), not these two AM4 refreshes. At least that’s how I understand it.
I was not in fact mislead. I actually went and found that article to help explain the concept of IPC uplift to you.
I was speaking of why new CPUs are a good thing worth being excited by in general. You were the one who said new CPUs aren’t worth being excited about in general.
As a bonus I showed how new CPUs even at the same speed have advantages. This all being said hey if you want to be a downer here go for it.
The new CPUs for AM4 socket. Aren’t we talking about AM4 as per post title?
Edit: Probably I should have made it clearer…new new AM5 ones are of course welcome and cool.