Computer Questions Not Worth Their Own Thread II

True if you count the i7's. But at $100-$150, you're well below an i7's price. At that price point, AMD has the processors with the highest GHz count.
 
Ok, that makes lil' more sense. But... What are the missing factors of CPUs top speed then? CPUs memory (what does that even do btw)? AMD's example has 4Mb against Intel's 3Mb. What does make Intel's product better then? Why that part has better top speed in the test?
 
OK. I tried to make some more research and I found one hard-to-read-for-non-native-english-speaker kinda article and I think it tried to say that these many little factors that affect CPUs speed affect to how many work is done per each cycle a.k.a. MHz. So my previous way of thinking was this: the AMD guy clocked 20 pull-ups in pull-up competition against Intel dude who scored 15 pull-ups. AMD guy wins. My new way of thinking: I look at these sportsmen and I see that AMD is 75kg fit guy and Intel is 125kg wrestling/strongman guy. So if we think about that we see that Intel got more work done because he lifted 50kg more weight per 1 lift aka MHz.

So why they don't count it like how much work they do per minute? Not like how fast their CPU clocks. What if it doesn't do work at all per MHz? I don't get it..
 
OK. Thanks. I understand that you have helped so good as you can. Still this situation is too difficult for my brains. I just don't get it; why they don't put the benchmark numbers to their products then so average guys can see if the machine is real deal or nooby piece of poo poo. Instead of this they think that average guy knows about cache, threading, pipelines, donald duck and rocket science and the bronze medalist of Bosnian TaekWondo 2nd class contest 1987.
 
There's no way to get competing manufacturers to agree on an objective benchmark standard - different products perform differently relative to each other depending on application. (e.g. AMD products perform on average relatively better in highly threaded environments. AMD products generally have better integrated graphics, except compared to the Iris Pro, and integrated graphics aren't relevant if using a dedicated video card. )

Again - think of cars - how would you ever get BMW, Ford and Honda to put a benchmark number next to each of their products? Does a Ford F150 have better performance than a Honda S2000?
 
You can quite literally take entire courses in computer architecture and the relative performance of various components (CPUs, but also GPUs, memory architectures, etc.) and only cover part of it. You could probably take an entire course on branch prediction, which is itself only one of many components affecting CPU performance. So, CPU performance is a difficult and complex topic if you want to really understand what's going on.

But, just as you don't need to understand how a car's engine is built to understand whether it will be a good fit for you, you don't need to understand all the details of a CPU to figure out which one is better for you. Car review sites will have reviews covering aspects such as acceleration, top speed, grip, and I'm sure for trucks, how much they can pull how effectively. And sites such as Tom's Hardware and Anandtech will have CPU benchmarks - what's relevant for most people is figuring out which benchmarks apply more to them, and then seeing which CPUs do well in those benchmarks and fit your price range. Which can itself be a bit of a challenge, but it's less complicated than understanding everything going on behind the scenes. Similar to how you can read Car and Driver and get an idea of which car/truck will work for you, without reading up on the details of how the engine is built.

A relatively high amount of technical academic papers on CPU performance is available to those who wish to read it, and some sites' reviews also go quite in-depth on that. But even in the academic papers, benchmarks are often used to compare the practical benefits (or disadvantages) of the discussed topics - though the academic papers tend to go into more theoretical detail than many review sites. However, as mentioned, these are pretty technical and not Sunday-at-the-beach reading. Good for upper-level undergraduate and graduate school computer science courses, and actually interesting if you're into that sort of topic, but not what you're likely to find on the "New and Popular" shelf at the local library.

Edit: I'm not sure how many of the academic papers are actually freely available. When I was in college, it was easy to access them, but searching for one of them online now, it looks like it isn't actually free to the public. Unfortunately, a lot of academic papers wind up being like this - researchers do research with public funds, but only people who are actually in a university system have access to them. It looks like I actually have paper, not electronic, copies of most of the papers I read in college.
 
CPU performance, TL;DR version

GHZ do not matter unless comparing same-generation/nm processors.

Architecture of the CPU rules all (22nm CPU is fitting almost 50% extra components in the same space as a 32nm CPU). This is, among other reasons, why a weak 2014 4th-gen dual-core i3 at 2.2ghz can be better than a 2007~ish AMD X4 Phantom quad-core at 3.6ghz.

A 4th-gen i7 at 3.6ghz is better than a 4th-gen i7 at 3.4ghz

a 3rd-gen i7 at 3.8ghz is not necessarily better than a 4th-gen i7 at 3.6, since architecture has improved.

L2 and L3 caches are important, bigger the better, although I'm not sure on the specifics here.
 
OK. Thanks.

Problem solved: Check benchmark lists for the purpose you need that processor to get the best answer to what is the CPU of your choise.

Sorry that I didn't listen to your Tom'sHardware tips at first.

Maybe I come back soon to learn about GPUs :p!
 
OK... I've now thought about the current CPU situation for a few weeks now but it is still hard to tell what is the best CPU for me. I'd spend max. 150€... What would that be in dollars?... Well, its about the same :p!

That CPU should be good for games. I dislike Intel i3 because it has only dual core system. That pretty much sucks in the long run because games will use more than two cores in near future (at least many people think so). However these guys performed pretty well in benchmark tests of brand new games. Something like Starcraft II and Arma III, they were beating AMD's octa core FX buddies really bad. They fit my budget.

AMD FX guys have many cores and in some games they didn't get totally owned by i3 guys. However these are starting to have sooo many cores that I'm afraid I my games won't benefit having them. Like.. u know they can't use all the cores properly or something... Also I've been into quite com friendly games like Civ and LoL and AoE but also some faster games like BF Heroes. However my current rig is cheap laptop and it lags in all above games. And I surely will buy new games from time to time and I don't want that they lag all the time.

i5 series would be cool I guess but these bad boys are a bit above from the budget. However we finnish people tend to order tech stuff (and many other stuff) from lovely Germany to avoid our insane taxes and benefit EU rights so I guess I'd get a nice i5 from a german super sale :D!
 
multi-core support for games is next-decade level "Soonᵀᴹ". It probably just won't happen at all. That is crazy difficult to program.
Dual-core support is somewhat here, in the sense that it actually helps - offload OS to one core, gaming etc. to another.

Dual-core i3 is the most price-efficient gamer CPU for under 150 USD. In single-threaded applications (games. all games. ever.) it's up to 50% faster than comparable 6 or 8-core AMDs, while using less power. It's the obvious choice unless you happen to be a photoshop enthusiast, videographer, etc., using actual multi-thread apps that require real power.
 
multi-core support for games is next-decade level "Soonᵀᴹ". It probably just won't happen at all. That is crazy difficult to program.
Dual-core support is somewhat here, in the sense that it actually helps - offload OS to one core, gaming etc. to another.

I think you are still stuck on the state of one or two years ago.

And it does make sense that the move to decent multi-core support across the board is happening about now;
PS4 and XbOne have to make do with "8-core" CPUs with anemic single thread performance.

PC only games obviously still can afford to run basically on a single main thread, though.
 
See? This is why I'm confused. One person tells something and other says something different... But what I see know is that maybe I should focus more to single core performance too.

Let's hear what other CFC guys think.
 
Fortunately, getting an i5 isn't meaning sacrificing single-thread performance, it will perform well at any application.
In a few cases a i7 or 8-core FX might perform slightly better, or a i3 might perform as well, sure.
But a i7 (Xeon) is at least $/EUR 50 more expensive, and there are enough cases where a FX or i3 are totally outclassed by a i5.

But these days, on a new system primarily meant for gaming, it's the most sensible choice, unless those $/EUR 50 more than a i3 are totally busting the budget.
 
Okay. Makes sense now.

The cheaper members of i5- series are much better for my purposes than FX- or i3- series, so when the time is right (xmas maybe) I'll grab an i5 from german super cheap sales. This might mean a little sacrifice in GPU ground but I will probably be okay anyways.

Now I know all that I need from CPUs. Moving forward to MoBos. So what might be a cheap option for 1150-socket i5 CPU. I won't overclock much because I'm inexperienced in that. I want to learn how to do that but I probably don't need that at least for a long time. I think that the computer would be in a quite difficult place so WLAN would be cool. That is MoBo's property, or? I like cheap options to this because I don't like sacrificing GPU too bad.

Soon I've figured out my dream computer's parts. It will be ~600€ budget, +30€ shipping to Finland.
 
Back
Top Bottom