Is there a legit website to compare GPUs

[Marcelo Delicia]

is there one that does not always say nvidia is better blah blah blah

[thekeemo]

benchmarks do not lie

[minibois]

I always use hardware.info, not sure if the international version site is good too

[GlassBomb]

is there one that does not always say nvidia is better blah blah blah

 

I would just look up benchmarks and reviews from reputable websites.

 

Such as PCPer and AnAndTech to name two. (linking didn't properly work for me, sod it.)

[themaniac]

is there one that does not always say nvidia is better blah blah blah

yeah here just ask for some benchmarks of said cards and we will give you some

[Atmos]

I swear to god no one say GPUboss, I've already done an in-depth report on them, and what it shows... oh dear lord.

 

HWComparison shows stats, but just stats, not benchmarks. And it doesn't have a particularly high sample pool of cards to draw on, so it's pretty lacking. But at least it's not horrendously biased.

 

http://www.hwcompare.com/

[CyberneticTitan]

http://www.anandtech.com/bench/GPU15

 

Performance oriented comparisons from Direct X 11, Compute, Synthetic, and Noise/Power/Temps.

[CydiaDash]

http://www.videocardbenchmark.net/ I use it all the time

[gojira]

I use GPUboss. sometimes it's wrong, but definitely has been reliable for my needs.

[Marcelo Delicia]

yeah here just ask for some benchmarks of said cards and we will give you some

r9 380 vs 970 gtx from nvidia

[Bim Job]

GPUBawce

[Marcelo Delicia]

I always use hardware.info, not sure if the international version site is good too

what do you mean

[manikyath]

the thing is you're comparing the taste of an apple with the taste of a pear, and numerically deciding which is better.

 

lets say we have AMD GPU A and nvidia GPU N, and say they're similarly priced within margin of error.

 

in game 1 they perform near identical, 1-2fps difference.

in game 2 A suddenly outperforms N by 5-10%

in game 3 N suddenly outperforms A by 5-10%

in game 4 theres special features for N, which A doesnt have

in game 5 theres a special API for A, which N doesnt have, making N perform notably worse.

---

 

what i'm trying to reach here is you need a way of numerizing a multitude of different factors of a GPU in one number.

 

its best to just look at as many different benchmarks you can find, where both cards were tested in the same scenario, and make a generalization of which one will be better.

--

that said, theres also other things we need to look at, for example GTX900 series' low energy usage & thus low heat output.

*currently* AMD seems to have a big advantage over nvidia in terms of Dx12, which to some users may matter a lot, and others only play older games, so they dont care at all.

 

 

if you're really unsure which team to go, go look at benchmarks of your favourite games, look at features you particularly like (like nvidia gamestream)

will you be doing multiple VRAM heavy things at once? maybe go for the 390X's 8GB VRAM at a very low price compared to nvidia's similar offerings.

 

will you be going for a quiet media center that needs to at least perform moderately on graphics? check nvidia's (rather pricy for what they are) multimedia GPUs (x10-x40)

 

that said, NEVER go to websites like passmark, because if you look closely, they're HIDEOUSLY off from reality, where a 980ti is actually winning significantly over a titan X, the 290x and 390x are placed under one banner, and the GTX970 is outperforming a fury X (and not by a small amount)

EDIT: as i rant about passmark, someone ninja's me with passmark's GPU website.

[othertomperson]

r9 380 vs 970 gtx from nvidia

 

Different tiers. 380 = 960. 970 = 390.

[minibois]

what do you mean

http://us.hardware.info/

[AddonCustoms]

I swear to god no one say GPUboss, I've already done an in-depth report on them, and what it shows... oh dear lord.

 

 

GPUBoss....  naaa I actually only scrolled down the comments to see who would recommend it, and I am pleased to say it looks like people are learning not to recommend it.

[Marcelo Delicia]

Different tiers. 380 = 960. 970 = 390.

i already know 380 crushes 960

[othertomperson]

i already know 380 crushes 960

 

It doesn't though, they're about the same.

 

If you pick cards that are in entirely different price points like the 380 and the 970 you can reasonably guess which tier is better (not always, but generally). When it comes to cards in the same tier, such as the 380 and the 960, you really need to look at benchmarks for the specific games that you are wanting to be playing. Because strategically AMD and Nvidia deliberately make very little difference between the cards. There is no sense in them over-engineering for a price point, (it also keeps the door open for a 380X or 960 Ti in future) but there are huge differences between specific games for any one of the cards and some are going to favour AMD and some are going to favour Nvidia.

 

So instead of listening to the fanboys who are going to say that their favoured card is going to "crush" the other team's, look at benchmarks of the two in the games that you are actually going to be playing with it.

[Atmos]

GPUBoss....  naaa I actually only scrolled down the comments to see who would recommend it, and I am pleased to say it looks like people are learning not to recommend it.

 

You might then be warmly inspired by the people here, that knew not to trust them as well.

 

http://linustechtips.com/main/topic/413757-gpuboss-lets-see-just-how-bad-it-is/#entry5566715

[SmittyShmee]

I'm kind of surprised no one has mentioned http://www.tomshardware.com/ yet.

[juretrn]

I use GPUboss. sometimes it's wrong, but definitely has been reliable for my needs.

PLease be a troll.

[gojira]

You might then be warmly inspired by the people here, that knew not to trust them as well.

 

http://linustechtips.com/main/topic/413757-gpuboss-lets-see-just-how-bad-it-is/#entry5566715

 

forgot to type /s

 

but thanks for that link anyways, I didn't think they we're that corrupted.