Lol, sorry, but our C++ projects have an average build time of 40 minutes on an 8-core/16threaded Xeon CPU in my workstation. Even using Fastbuild/SN-DBS it still takes 5-10 minutes. We'll take any number of cores we can get, thank you.
Lol, sorry, but our C++ projects have an average build time of 40 minutes on an 8-core/16threaded Xeon CPU in my workstation. Even using Fastbuild/SN-DBS it still takes 5-10 minutes. We'll take any number of cores we can get, thank you.
I'm a Python developer and I can work from laptop. I do a lot of data processing and my laptop can handle development and testing. On production, some of the scripts are running on 64 core machines and use vector operations via numpy and scipy.
In a company where I work, almost every front-end developer uses a laptop with an external display because it's a portable solution.
I currently use a 6-core mobile H-class CPU for my work that is a mix of Java and JS development, with a bunch of browser tabs and docker containers running, large IntelliJ project, etc. While my system often has a lot of processes/threads running, the main cost of those is in memory utilization rather than constant CPU performance.
I'm not saying that can't use more CPU horsepower, but it's certainly enough for my needs (for now). The main benefit of having more cores for me right now is just to maintain system responsiveness while running a bunch of background tasks, rather than raw throughput of large compile jobs.
Btw docker is only standard for CI systems. Never would I advise anyone to actually use docker for local development let alone 10 docker instances at once. It's just pure insanity.
There are many alternatives to docker which don't cripple your computer. Nix is the best one.
Whether it's having to run a lot of Docker containers, a large C++ project, or a large Webpack project.
Nitpicking and criticizing someone's dev environment is irrelevant.
If you don't like their example, I'm working on a JS project where I run a lot of local VMs to locally test the web client in different configurations. It would continue to be a bizarre response if you thought it was time to tell me to "just use browserstack.com" or something. Software isn't as easy as we think.
I also compile large, multi-compilation-unit C++ programs, and it's much, much better to have a 24 or 32 threads clocked at 2 GHZ than 8 threads clocked at 5 GHz.
The parent comment was talking about fast individual core designs, which are genuinely useful for gaming, and can't be approximated by adding more cores.
What generation and what frequency?
Those things can be quite dissapointing for what they cost compared to consumer CPUs.
I would expect the 5950X to be around 3-6 times faster than an 8 core Xeon.
It's fast, sure, but for that price tag I expected more. Currently evaluating buying a TR 2950x workstation. That seems to be amazingly fast though my research on the net is still ongoing.
The biggest companies in the world writing the largest software don't have build times that slow.