Guide to Performance Gaming PC Components for the best Value

Hardware Hub Other Hardware
1

I have finished tuning up my PC and I want to use a new thread to dispel some myths that I believe have taken hold in regards to Gaming PCs.

Firstly - CPU Intel is better @ 1080p.

This isn't really true. I've run a variety of benchmarks on my system with a 8370e. The Unigine Engine gives a good comparison to what gaming would be like on a system. This is the Overclockers.net comparison, I've chosen it on purpose because it shows Overclocked systems against Overclocked Systems, so apples to apples without Cherry Picking and those stupid graphs that make no sense. - My result is the TastyRabbitStew one, number 13 above the 5960X Titan X combination.

Capture.PNG530x579 139 KB

Overclockers.net Scores can all be found here.

Secondly - You should buy Custom GPUs with fancy coolers (presuming no LNO).

I'll demonstrate that this is not a good idea below, the money you save from a Reference design will allow you to go for a custom loop. Don't worry about reference being worse overclockers. The Chips all come from the same factories and it is down to chance unless the chip has been binned, which is very unlikely as the cost would be prohibitive and my result demonstrates this is not the case. The VRM's can all handle the voltage and Wattage you are putting through, and there is no effect to OCing unless on LNO. In the EU the warranty is guaranteed on all cards and electronics you buy as well as SOGA in the UK.

A fancy 980 Ti in the UK is priced at £650, a Ref is £509.

The fancy 980 Ti Chart from Guru3D-

untitled-1.png581x590 25.6 KB

My Reference - 5057

This beats all the customs at £150 cheaper.

So now you saved so much money on your GPU and CPU you can buy a custom loop and actually OC to get decent performance out of your stuff.

4K Tables:-

Capture.PNG661x659 125 KB

Your cheap system is up with the big boys for much less cash, it would be at the bottom without water. Remember these are probably OCed systems. They all perform about the same in reality regardless of 1080p or 4K.

Decent air coolers aren't worth the cash on GPUs, unless constrained by form factor etc. The reviews make no sense... You shouldn't be buying Intel 8 Cores etc for gaming, or Custom GPUs, save the money and get a custom loop. Universal Blocks on GPUs do just as well as Full blocks. Magicool make very cheap rads (£30 for the dual Here), and there are hundreds of super cheap Pump/Res combos for like £30 Here a 240v one or get a 12v for slightly more. You should be able to watercool for about £125 ish to under £200 if you go mad. Your saving £150 straight up just off the GPU.

The case is the same in the US I believe, but I'm too lazy to price it up as well.

Anyway, I hope this thread will stop me from Hijacking other peoples when I'm trying to explain why that G1 980 Ti + 5960X are probably a bad idea!!! I didn't even make it to Motherboards and that can of worms....

3 Likes
2

Good post, valiant effort.

The greedy wheels of marketing will keep churning for as long as there are people not informed enough to buy into the fairy tales.

It makes me chuckle each time I see a company release an X99 gaming motherboard. I don't have anything in particular against X99 solely for gaming, it's just an example.

1 Like
3

Good post, but the GPU thing is a little off. A reference cooler is designed to be used with the GPU it is designed for, and works perfectly fine, until you decide you want to game without a jet engine in your room. Then you have an issue. Many of the designs that the other companies come up with are much more efficient and keep your rig far quieter than what a stock cooler would give you. Some may not care for the increase in performance, they are just seeking the quietest rig that they can build. Not everyone wants to build a custom loop either.

4

@VXAce A custom card will not perform as well as a custom loop and will probably be louder than it at peak.

There is basically no reason to not buy a custom loop on a flagship card... I literally can't think of why you would by the £650 cards instead. The custom more efficiency thing is not really an issue all will draw about the same (425 Watt). The VRMs on the cards make no difference... Custom loops fit mITX now as well. The VRMs on the cards make no difference.. All GPUs are over-specced anyway to fit EU consumer law (unless LNO). Also the Chip is the same... what your saying seems to make sense but in reality it doesn't, that's what my post was all about.

I used to think the same too, but the reality is its an impossible rational argument. You can subjectively say you prefer air and are willing to spend extra on it though.

5

I can think of plenty. Custom loops can leak, they take much more planning and experience with building things than some people have, they are an utter pain to work with sometimes when they go wrong, you don't have access to the tools need, don't have the time, don't live somewhere where a stationary rig can sit for years, etc. And if you want to get a custom loop to fit in an mITX case (though something like the Node 202 or the SG-13 would beg to differ), you better know your way around a lot of tools, and have access to them.

Your situation is not defining reality, I can say subjectively that I prefer air, and I do. I'm not arguing that it is better, I'm say that there are other situations that people may be in that differ from yours or mine.

1 Like
6

Ok, I'll admit custom loops are not for everyone.

But for the vast majority of desktop users aren't moving around, and I think for value, they actually make sense where as the Custom Cards don't. I can't really understand the custom cooler thing!!! I think its a mental block to me, paying more for noisier less powerful cards. Setting up a custom loop is not very difficult.

Sorry if it comes out a little harshly though. The performance hit you take is quite large, like 25-40% on air.

7

nice topic.

But this is exaly the first wrong point in it allready.
Intel haswell i5's / i7´s are exaly better cpu´s for gaming, at lower resolutions like 1080p and also in some 1440p gaming.
.
The per core performance and ipc of intel haswell cpu´s are way better, then AMD FX cpu´s, which still makes allot of sense in some games today.
Especialy MMO's, RTS and such kind of games.
But also in cpu + gpu demending gaming scenario´s like heavy multiplayer.

So 3D firestrike doesnt realy tell me that much at all.

8

I dont think that making claims based on just firestrike scores,
realy gives you a full correct image of what is going on.

i woulld suggest to go and do some research, dig up some recent 1080p and 1440p gaming benchmarks, with higherend gpu´s.
This will give you some better perspective.

This is discussed tons of times allready, so i´m personaly not going to botter to start this discussion again.
But i have done tons of researching my self over the last 3 years, just about this debate.

9

@MisteryAngel I've given Unigine and 3DMark scores.

There is no difference between the CPUs scores @ 1080p or 4K.

You can check the Overclockers.net link at the top of the page.

I am making the presumption that the results there are OCed results as well, which I think is fair.

The number show there is no difference.

X ONLY IN GAMING X (I haven't run straight benches).

10

My thoughts exactly, synthetic benchmarks only mean so much. Actual gaming benchmarks mean a lot more.

11

cpu load doesnt realy change that much wenn you up the resolution.
How higher the resolution, how more GPU limmited you are going to be.

12

Unigine is a fair gaming assessment. Or as real as you will get.

The numbers are there for everyone to read.

@MisteryAngel I've given 1080p numbers too.

13

You dont get full the picture.

Highend gpu´s at 1080p can push allot more frames, then on 4k.
so at 1080p the cpu will have a harder time to keep up with the GPU, then it is on 4k.
Because on 4k you will be mostly GPU limmited.

Since AMD FX chips have slower cores, they will have a harder time to keep up with a highend gpu on lower res, then an haswell cpu, which has better per core performance.

Ofc it highly depends from game to game now days.
But still it doesnt change the fact, that intel haswell cpu´s are better allround chips for gaming.

14

No, it really isn't. Unigine doesn't have 300 players moving around on a map at one time with tens of thousands of rounds being tracked along with shadows, lighting, audio and just about everything else. Or 64 players with hundreds of rounds being fired and an ungodly amount of explosives going off.

A gaming benchmark is totally different, it factors in the insanity of an inconsistent chaos.

And I haven't even mentioned if a game is optimized for more than one thread.

15

I don't know what else I can do to convince you...

I thought there would be a difference too, I wasn't expecting the results to be like this. I thought my i5k would be better.

I think I've shown to a reasonable level that it doesn't make a difference on heavily Overclocked hardware.

On normal hardware the results would be very different.

16

http://www.pcper.com/reviews/Systems/Quad-Core-Gaming-Roundup-How-Much-CPU-Do-You-Really-Need
Benchmarks galore... really in depth testing, not just quad core stuff. There is a benchmark for practically every combination of current hardware, with Mantle results thrown in as well. TONS OF INFO.

1 Like
17

That I5 4440 was calling to me. :) This chart

fire_strike_sorted_0.png602x1504 557 KB

1 Like
18

The point I'm trying to get across is these are all from non overclocked, or poorly overclocked systems.

The true difference is really far smaller when you take the CPUs and bench with custom loops, actually I'll go further and say there's no difference (within error) my number prove that.

That's the point I'm making. The Overclockers.net results are far better because they are all from enthusiast systems, putting up the results for free, and most Overclocked to a high level with no axes to grind.

I was hoping to lay the myth to sleep, but I've failed spectacularly.

I wanted to help people understand they didn't have to spend a fortune for a high spec - nearly the highest single gpu system by using lesser GPUs (reference) and cheaper CPUs + custom water.

19

i realy like these kind of topics in which there is a civil discussion about intel vs AMD.
I also realy appreciate the afford that topic starter has put into this sofar.

Keep it going! ☺

1 Like
20

That seems like a good chip. If I ever decide to build with Intel, I think I might choose that one. But for now my FX-8350 does everything I need it to... only real upgrade path for me is on the GPU side.