This was inspired by my inability to run Diablo 2 Resurrected, but I’m not specifically asking about that. I thought that the problem might be that my desktop PC is about 5 years old, so I thought that I’d look into replacing it.
I checked out Best Buy and Office Depot and found some lower-end computers for ~$400 - $700, but I can’t make sense of the processors. [Back in the day it went 386, then 486, then Pentium…] I googled and found this site: Best Processors - January 2023
It seems that it’s way more complicated than just i9 > i7 > i5 > i3
What I have now is a Lenovo with “Intel(R) Core™ i5-6400 CPU @ 2.70GHz 2.71 GHz” which seems to be #177 on that link. I found a PC at Best Buy with an i5-10400 for $679. At #90 on the list, it doesn’t seem like much of an upgrade, although, if we go by the 3DMark Physics Score, it’s about 2.5x better.
I’m starting to think that I WAY overestimated the age of my PC (is there an easy way to check this?), or Moore’s law has been broken. Or I just don’t understand this stuff at all. I used to, but I’ve been out of the technology game for about a decade.
Surely there must be an easier way to computer shop.
Except for very specialised apps more than 4 cores/CPUs is little if any advantage. What matters, especially for games, is clock speed. Your current 2.7ghz is not too bad in that respect but there are ones up to say 3.5ghz for affordable prices.
Whether the model of CPU will help performance depends on the game or other app. If it is not maxing out your current i5 then going to a higher model, i7 or i9, is not likely to help any.
BTW you say you cannot run Diablo 2 Resurrected, yet your CPU exceeds it’s requirements, if not by much:
Or is it some other factor that stops you runnng it, say not enough memory or GPU power?
Computer power has never been possible to measure in a single metric. Even benchmarking with a specific work set is impacted by scalability.
The balance between raw CPU, memory performance, and I/O has been a minimum, with communication speed becoming important as well. But it is really much more nuanced.
Cache performance has a huge impact and is really the dominant part of what makes modern machines performant.
The memory hierarchy of a computer becomes a major determinant of performance. Where we draw the lines changes over time, but the principle remains. Nowadays the top layers of the hierarchy all live on the CPU die. L1, L2, sometimes L3 cache, memory controller. Off chip you get the memory itself and beyond that we really need to include flash based storage. Flash storage has been the biggest bump to PC performance in recent times. Advances in raw CPU have been much less important.
But that ignores GPU performance. Budget PCs are reliant on on chip GPUs, and there is a wide range of performance here. Much wider than the host CPU.
Intel’s iX range of CPU branding is mostly marketing. For many workloads there is no useful improvement in speed between different iX chips of similar configuration. The more advanced and expensive chips gain accelerated performance on a narrow set of workloads. Maybe that helps you, but often it doesn’t.
You may find that the higher spec on-die GPU in a newer device is the most important gain over an older one. But it isn’t trivial.
As the OP said, the 3dmark physics score is much better for the 10400 ? why ? because the modern 3D software is multithreaded and makes use of all available cpu cores. You do need the graphics card that allows use of the library, but the modern 3d drivers are able to allocate 3d function to CPU emulation routines or pass them off to GPU , so five unusued cores in the main cpu do a fair bit of 3d graphics work…
So a game that makes use of the same physics library (GPU has to be compatible with that.) is obviously making use of the extra cores in the 10400.
10400 is 6 core, 6400 is 4 core.
10400 can run at up to 4.3 GHz, 6400 can run at up to 3.3 GHz
10400 has faster RAM access, about 40% faster
10400 has twice the L3 cache, 50% more L2 cache, three times more L1 cache .Cache sizes have diminishing returns. (so the change from double to triple is only a few % overall performance improvement. So dont worry about cache sizes unless its the only difference in the cpu. )