4 ms·

What are people using CPUs for? I mean I have a crappy 7 year old CPU that I use for single threaded dev, and it works for 99% of cases. In the 1% of cases, I

What are people using CPUs for? I mean I have a crappy 7 year old CPU that I use for single threaded dev, and it works for 99% of cases. In the 1% of cases, I threw on multithreading and its fine. Now my daily driver has a GPU and it unlocks new possibilities. Its been over 10 years since I've really spent time thinking about a CPU.

jibe 3y ago
A lot of games still bottleneck on CPU.
- wilsonnb3 3y ago
  especially if you are trying to get high framerates
  - brucethemoose2 3y ago
    Or trying to run large maps in sim games. Speaking as a 7800X3D owner, I don't think any CPU on the planet is sufficient for modded Rimworld yet.
josephg 3y ago
Compiling rust code is slow and cpu bound. I have a fast cpu in part to improve compile times while I’m programming.
- tupolef 3y ago
  Did you experience a big gap between an older CPU and the current one? I did not change mine for 6 years, but looking at some benchmarks, it looks like compilation times would not get more than a few percents faster per generation. I expect to get maybe 20% faster with a recent CPU in the same price range. Until 2005 I was updating the CPU every 2-3 years, then every 5 years, and when I'll get the next one I plan to keep it 10 years easy.
  - rwaksmunski 3y ago
    I went from 2013 MBP to M3 Pro MBP and the performance improvement in Rust / rust-analyzer was staggering.
  - AdrianB1 3y ago
    There is a huge difference between my laptop and my desktop in using VS Code with a few basic plugins. The difference of the hardware between the 2 machines is mostly CPU and SSD performance, the SSD makes a big impact opening projects with many small files, but the CPU is used for IntelliSense or even when you copy/paste a few thousands of LOC from one file to another (laptop fan gets to max for tens of seconds). This is the biggest difference I see today. In the previous generation it was a huge boost in everything just becauase the old CPU was really old. I also keep CPUs 5 years or more.
  - doublepg23 3y ago
    I’d be curious if your old cpu uses an NVMe or SATA drive. I’ve seen a 20% increase on big projects from NVMe alone - I suspect due to the big I/O queue depths of NVMe.
  - dist-epoch 3y ago
    A new CPU would probably have double the core count (16c/32t) of your current CPU. C++ compilation can use all the cores.
  - the8472 3y ago
    Modern CPUs also have more cores. Make and Cargo can keep a threadripper fed for a few minutes with a task that would run an hour on a laptop.
gwill 3y ago
I do astrophotography and stacking images is the only time I notice CPU performance. It can take close to an hour depending on how much data, so I just go do something else.
doublepg23 3y ago
With the release of Ryzen in 2016/2017 the bare minimum upgrades Intel was pushing YOY no longer worked in the market. Since then the CPU space has been very exciting, you've been missing out.
spacechild1 3y ago
Gaming, video editing, music production, 3D modelling, compiling C++, etc.
phkahler 3y ago
>> What are people using CPUs for? I compile C++ code with an AMD 2400G (zen1+) and parallel builds are way faster than single threaded. I also used the same box to add a bunch of parallel paths to the code (using OpenMP) which improved performance of the app by 3x-4x in some cases with little effort. Compiling a full build is 30-60 seconds for me. If I turn on LTO it's several minutes and that's with parallel builds -j I'm aiming to get a Zen4 or Zen5 with 8 cores one of these days. That should give about 4x speed boost. The new GPU has AV1 support too!
- fransje26 3y ago
  > and parallel builds are way faster than single threaded. Not only that, but parallel builds on newer CPUs is way faster than parallel builds on 7 year old CPUs, even at lower clock speeds.
hwbunny 3y ago
That's why most of the review sites never show older, just current, maybe previous generation of cpus with reviews.
- ogurechny 3y ago
  This is and article from Phoronix, a site with more test results than you'll ever need. You can probably find some coffee maker performance in Tux Racer in the database (if you figure out how to use it).
pi-rat 3y ago
Whenever bazel prints that it’s compiling one of the rust parts of the codebase at work, then I mutter something about needing more faster better CPUs, and go fetch a cup of coffee..
- karolist 3y ago
  Why not remote builds? https://bazel.build/remote/rbe https://bazel.build/remote/rbe
ActorNightly 3y ago
I have a laptop with a 4080 and AMD chip. With Nvidia Prime, you can basically use the chip integrate graphics that are a lot more power efficient, and make the battery last way longer.
brucethemoose2 3y ago
Actually... machine learning. One would think its all GPU. But you'd be suprised how single-core-speed limited parts of certain tasks are.
deeth_starr_v 3y ago
Modern typescript based web-app … you’re going to want the fastest single core speed cpu & nvme possible to make it not suck.

jibe 3y ago

A lot of games still bottleneck on CPU.

wilsonnb3 3y ago
especially if you are trying to get high framerates
- brucethemoose2 3y ago
  Or trying to run large maps in sim games. Speaking as a 7800X3D owner, I don't think any CPU on the planet is sufficient for modded Rimworld yet.

josephg 3y ago

Compiling rust code is slow and cpu bound. I have a fast cpu in part to improve compile times while I’m programming.

tupolef 3y ago
Did you experience a big gap between an older CPU and the current one? I did not change mine for 6 years, but looking at some benchmarks, it looks like compilation times would not get more than a few percents faster per generation. I expect to get maybe 20% faster with a recent CPU in the same price range. Until 2005 I was updating the CPU every 2-3 years, then every 5 years, and when I'll get the next one I plan to keep it 10 years easy.
- rwaksmunski 3y ago
  I went from 2013 MBP to M3 Pro MBP and the performance improvement in Rust / rust-analyzer was staggering.
- AdrianB1 3y ago
  There is a huge difference between my laptop and my desktop in using VS Code with a few basic plugins. The difference of the hardware between the 2 machines is mostly CPU and SSD performance, the SSD makes a big impact opening projects with many small files, but the CPU is used for IntelliSense or even when you copy/paste a few thousands of LOC from one file to another (laptop fan gets to max for tens of seconds). This is the biggest difference I see today. In the previous generation it was a huge boost in everything just becauase the old CPU was really old. I also keep CPUs 5 years or more.
- doublepg23 3y ago
  I’d be curious if your old cpu uses an NVMe or SATA drive. I’ve seen a 20% increase on big projects from NVMe alone - I suspect due to the big I/O queue depths of NVMe.
- dist-epoch 3y ago
  A new CPU would probably have double the core count (16c/32t) of your current CPU. C++ compilation can use all the cores.
- the8472 3y ago
  Modern CPUs also have more cores. Make and Cargo can keep a threadripper fed for a few minutes with a task that would run an hour on a laptop.

gwill 3y ago

I do astrophotography and stacking images is the only time I notice CPU performance. It can take close to an hour depending on how much data, so I just go do something else.

doublepg23 3y ago

With the release of Ryzen in 2016/2017 the bare minimum upgrades Intel was pushing YOY no longer worked in the market. Since then the CPU space has been very exciting, you've been missing out.

spacechild1 3y ago

Gaming, video editing, music production, 3D modelling, compiling C++, etc.

phkahler 3y ago

>> What are people using CPUs for? I compile C++ code with an AMD 2400G (zen1+) and parallel builds are way faster than single threaded. I also used the same box to add a bunch of parallel paths to the code (using OpenMP) which improved performance of the app by 3x-4x in some cases with little effort. Compiling a full build is 30-60 seconds for me. If I turn on LTO it's several minutes and that's with parallel builds -j I'm aiming to get a Zen4 or Zen5 with 8 cores one of these days. That should give about 4x speed boost. The new GPU has AV1 support too!

fransje26 3y ago
> and parallel builds are way faster than single threaded. Not only that, but parallel builds on newer CPUs is way faster than parallel builds on 7 year old CPUs, even at lower clock speeds.

hwbunny 3y ago

That's why most of the review sites never show older, just current, maybe previous generation of cpus with reviews.

ogurechny 3y ago
This is and article from Phoronix, a site with more test results than you'll ever need. You can probably find some coffee maker performance in Tux Racer in the database (if you figure out how to use it).

pi-rat 3y ago

Whenever bazel prints that it’s compiling one of the rust parts of the codebase at work, then I mutter something about needing more faster better CPUs, and go fetch a cup of coffee..

karolist 3y ago
Why not remote builds? https://bazel.build/remote/rbe https://bazel.build/remote/rbe

ActorNightly 3y ago

I have a laptop with a 4080 and AMD chip. With Nvidia Prime, you can basically use the chip integrate graphics that are a lot more power efficient, and make the battery last way longer.

brucethemoose2 3y ago

Actually... machine learning. One would think its all GPU. But you'd be suprised how single-core-speed limited parts of certain tasks are.

deeth_starr_v 3y ago

Modern typescript based web-app … you’re going to want the fastest single core speed cpu & nvme possible to make it not suck.