Tom's Hardware Graphics Charts: Performance In 2014

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

InvalidError

Titan
Moderator

He was writing about the point where bandwidth starts causing significant performance degradation. Right now, most GPUs are still mostly fine with 1.0x16/2.0x8/3.0x4 but at the higher end of the spectrum, 2.0x16/3.0x8 are starting to become necessary - the penalties are becoming steeper than the 3-5% they used to be.
 

bin1127

Distinguished
Dec 5, 2008
736
0
18,980
I would've thought 1440 or 1600 is a resolution Toms would test. QuadHD seems like just enthusiast curiosity, but 1440 is pretty widely used.
 

InvalidError

Titan
Moderator

When NTT originally coined and demonstrated QHD many years ago, it meant 7680x4320.

I just hate when acronyms get re-defined a bunch of times between the time they are first used and the time people settle on one specific meaning. HD was much the same with many outlets slapping the HD label on anything capable of more than 480i, some trying to be more transparent by labeling those ED or whatever else so we ended up with the FHD label for displays genuinely capable of 1080 lines or better... just like QHD got bastardized from 1280x720 "HD" so now we also have QFHD for the real thing and a shift to 2k/4k/8k naming to disambiguate the bunch-of-letter-nobody-agrees-on-a-standard-definition-of mess but is still plagued by tons of variants of its own.

As far as this sort of article goes, people who complain about the lack of different resolution testing need to keep a couple of things in mind:
1- this is only a general performance ranking thing
2- testing at extra resolution takes tons of extra time - particularly when re-testing all the cards due to OS, software and driver changes to update rankings
3- testing at lower resolutions adds little value since the GPU will be limited by fill rate so lower-resolution results can be extrapolated relatively easily
4- testing at uber resolution also adds little value since different GPUs' performance varies drastically between titles, multi-GPU setup, settings and other variables so people interested in beyond-HD resolutions, uber details would never be happy with the results from a ranking chart no matter how thorough the underlying testing may have been since that weighed-average chart result will include tons of results that are of no importance to them

More testing may produce a more "accurate" overall value but a more accurate performance index does not tell you anything particularly useful: the moment it includes results from games, resolution and quality settings you do not care about, the ranking becomes worthless beyond giving you an at-a-glance relative performance comparison and you have to hunt down individual benchmark results to get any more details than that.

So, for the purposes of generating an overall performance chart, only doing 1080p sounds good enough to me.
 
G

Guest

Guest
Wouldn't Grid 2 be a better choice than Dirt 3? It is from the same company, but is much newer, is better coded/optimized, and comes with a built-in benchmark (as does, Dirt 3, as far as I remember). I mean, Dirt 3 is old news which, even when it was around, being a rally game, didn't have as much mass appeal as a regular GT racing game.
 

Leah-Alaine

Reputable
Jul 11, 2014
10
0
4,510
I'm looking to upgrade my computer, I am not great with conputers but I'm willing to pay a fair amount to get it working for games. Bf3, MW3..games like that.
My specs (emachine)

Windows 7
-AMS Athlon 11x2 235e dual core processer
-NVIDIA GeForce 6150SE integrated
DVD super multi drive
-750 GB multi drive
-6 GB DDR2 memory
... Thanks
 


You're probably better off posting a thread.
 
Status
Not open for further replies.