Just curious to know which one is typically faster to compile on Linux, Intel or AMD CPUs?
Core count is the more important parameter, in order to exploit parallel building of crates. Following that, using a faster linker like mold can really help.
If you're building a desktop for developing software, I'd be more worried about compatibility. To my knowledge (I may be mistaken, do your own research), AMD is typically better at supporting Linux than Intel/NVIDIA. On the other hand, Intel/NVIDIA are much more common, so you might choose them to be able to better target that hardware. It depends on your goals. Just make sure you have 8-16+ GB of RAM and a fast internal SSD if you really want the best performance when compiling.
Linux support for Intel has always been stellar in my experience; I don't know why you're grouping them with NVIDIA.
That's fair, I can't say I've had issues either. I've got some vague ideas in my head about partially supported iGPUs or network cards, but I don't have anything concrete.
They had some weird CPUs with PowerVR cores (e.g. Nexus Player used one) that had very poor compatibility, but after their attempt to make something usable for smartphones failed they stopped using these.
Extremely different attitude from nVidia.
About 10%-20% faster on Intel processors, however YMMV. I am comparing i9 to Amd 9500. I have many friends with Nvidia, so if you have a particular question, I can forward it to them.
The best way to pick which CPU is faster is to look for compilation benchmarks on Linux, such as Timed Linux Kernel Compilation Benchmark - OpenBenchmarking.org
From looking at it, the fastest desktop CPUs are:
| CPU | Time to compile (smaller is better) |
|---|---|
| AMD Ryzen 9 9950X3D 16-Core | 45 ± 3s |
| AMD Ryzen 9 7950X3D 16-Core | 48 ± 1s |
| AMD Ryzen 9 9950X 16-Core | 49 ± 4s |
| Intel Core Ultra 9 285K | 50 ± 2s |
| AMD Ryzen 9 9900X3D 12-Core | 53 ± 1s |
| Intel Core i9-14900KF | 54 ± 3s |
| AMD Ryzen 9 9900X 12-Core | 58 ± 5s |
| AMD Ryzen 9 7950X 16-Core | 60 ± 9s |
| Intel Core i7-14700K | 63 ± 6s |
| AMD RYZEN AI MAX+ PRO 395 | 66 ± 2s |
To be clear: each of these CPUs is expensive. Very. Once you get into the more mid-range components, things get a little more complicated.
Compilers like rustc and clang are basically giant hash table benchmarks, so more cache helps more than in most benchmarks -- thus the X3D chips doing so well -- and of course for large, well-factored projects it all comes down to more raw power from more cores.
So in prosumer chips, chiplet-based AMD processors tend to be best. I don't know in more "normal" consumer chips, though, nor in datacenter-tier chips.
@Joe232 Are you considering a laptop or a desktop, or, do you prefer it to be quiter or you don't mind a fan noise? (This is not your original question, but since compilation speed is your goal (as is for many of us)...)
If a laptop or a low noise desktop, you want metal chasis and good thermal disspation design. Otherwise the CPU and GPU speed gets throttled automatically.
The RAM speed and the bus speed don't get auto-throttled. Could the laptop/low noise sweet point be in CPU with a high number of cores, even with lower maximum speeds?
If you want a laptop and you don't need extra fast graphics, integrated GPU saves a lot of energy and hence heat, so (if all other parts are equal) it would auto throttle less, hence CPU can run at higher speed more.