About gaming hardware benchmarks

UFOs, lost socks, discuss whatever you like here.

Moderators: Master_Kale, TNM Team

Post Reply
AEmer
Illuminati
Posts: 1490
Joined: Fri Jan 26, 2007 12:04 am

About gaming hardware benchmarks

Post by AEmer »

We've had some discussions on this forum about pc gaming hardware, what's good, what's bad, and how you tell the difference.

In particular, things like CPU's, memory and graphics cards have been tricky to really evaluate, because the statistics various benchmarking sites post don't reveal the full picture.

Often the numbers they put up give you some measure of how you can compare hardware to other hardware of the same category, but usually, that's about it. You get some idea of what's best, but you can't really ever tell if something is "good enough" - and so, it becomes very hard to determine whether a specific piece of hardware is worth the additional cost compared to the alternatives.

Fortunately, hardware reviewers have become privvy to this. A specific site I just came upon directly addresses the usefulness of benchmarks, and attempts to come up with a better benchmark for pc hardware gaming setups.

http://techreport.com/articles.x/21516/1

I figured that, with a bunch of hardware interested nerds hanging around here, it would be worthwhile to put it up, and see what you all think about their scheme: Measuring how long it 'usually' takes to render frames, and looking at how common and how bad the slowest rendering times are.
bobby 55
Illuminati
Posts: 6354
Joined: Wed Jun 24, 2009 9:15 am
Location: Brisbane Australia

Re: About gaming hardware benchmarks

Post by bobby 55 »

Interesting stuff. I have to admit to not knowing about multi GPU stuttering until reading that. The one thing I wish the various benchmarking sites would include is the best hardware for people who don't have lots to spend.
Growing old is inevitable.......Growing up is optional
User avatar
RedDynasty
Thug
Posts: 27
Joined: Thu Jun 21, 2012 9:55 pm
Location: Portugal( yes there are portuguese people in dx)

Re: About gaming hardware benchmarks

Post by RedDynasty »

very interresting, now i know how to get proper hardware :)
Image
Your not a gamer until you played this.

Cant....stop...listening...to.... RedSun......Menu music.
AEmer
Illuminati
Posts: 1490
Joined: Fri Jan 26, 2007 12:04 am

Re: About gaming hardware benchmarks

Post by AEmer »

Well, to me, this stuff is absolutely priceless.

What it means is, frame delay happens, and we have a way to measure it. We can see how many frames take more time to render than we'd like, and we can see whether that's a cpu or gpu limitation. We can also compare cpu's and gpu's, and we can work out what our goal should be; if we set a specific goal in terms of fluidity and latency, we can design a pc - high cost or low - that will meet those specifications, and we can optimize the price towards making sure we get a minimum of performance.

It's easy if you just want the best current solution - just get the Core i7 3770 and an nvidia gtx680 and ram clocked at 2133 mhz and then you're set. It's expensive as hell, but now we actually know that this combination has a specific benefit over other combinations, and if we use it as a baseline, we can say, oh, but I don't need the 99th percentile to be lower than 33 ms..so long as I get that, I'll be happy, since that guarantees 30 fps without any significant slowdowns.

For that reason, I actually wish they would take it a step further and give us the 999th percentile, since that's one skipped frame every 15 seconds.

The 99th percentile is a good measure though, combined with their measure of number of frames superceding 50ms rendering time.

As a result, if I were to design high performing budget pc's, I could actually use this review; I could say, look at the cpu's, choose one which, when combined with high power components, acts in this way. I could choose a graphics card in the same fashion, following the same 99th percentile specification. That way, I have two components that are equally good at insuring a minimum performance. Then, I'd choose ram running a similar frequency to the testers, then I'd choose a motherboard with a similar chipset to the testers, and then I could be certain that the system should deliver the minimum fluidity quality I'd specified.

If these measures become common, it will be very easy for me to build pc's that will perform as they should, and as a result, I'll help people save money and get a great experience :D
User avatar
ZeroPresence
The Nameless Mod
The Nameless Mod
Posts: 618
Joined: Mon Oct 17, 2005 7:29 pm
Location: Florida, USA
Contact:

Re: About gaming hardware benchmarks

Post by ZeroPresence »

AEmer,

This may not be very helpful because its so basic, but check out a site called Cyberpowerpc.com.

They custom build rigs for people at a rather large price, so I don't actually suggest purchasing here unless you just want a no hassle good computer... but their customization setup is wonderful for window shopping. They give you a quick read out of FPS on modern titles as you swap items and it is all guaranteed to be compatible so you can write down notes and what have you.
AEmer
Illuminati
Posts: 1490
Joined: Fri Jan 26, 2007 12:04 am

Re: About gaming hardware benchmarks

Post by AEmer »

that actually doesn't sound bad at all :-)
AEmer
Illuminati
Posts: 1490
Joined: Fri Jan 26, 2007 12:04 am

Re: About gaming hardware benchmarks

Post by AEmer »

Unrelated, but something else I stumbled accross at tech report

http://techreport.com/forums/viewtopic.php?f=37&t=81957
Post Reply