Why people buy Nvidia and Intel (Gamegpu.ru review)

[BrightCandle]

I go to gamegpu.ru on a regular basis. Its in Russian so you can't exactly read the reviews without a translator but they do some of the widest CPU and GPU benchmarking of any site. The main issue I have with what they do is usually the scene they use for the benchmark itself, its usually simple (but repeatable) and quite often its the in game benchmark which often isn't all that representative of gameplay. But they are also one of the few sites that tests a wide variety of CPU and GPU hardware across a vast list of games (probably thousands at this point). Today they released a performance test of the latest version of warthunder.

 

What I think is interesting is two graphs (you don't need to translate the page to read gamegpu.ru, just look at the graphs). The first is that on GPU performance at 1080p (although the same thing plays out at 1440p and 2160p as well):

 

 

The second picture is that of CPU usage and performance.

 

 

Nvidia and Intel absolutely dominate, its not even funny how poorly AMD competes here. Its obviously not like this in all games but there is a common trend that keeps coming up on gamegpu.ru and other sites, the less well known games run much better on Nvidia and Intel, considerably so. Gamegpu.ru often go back and play old popular titles (like this and Arma and World of tanks and a variety of other war games they presumably like) and there is this common theme, the lesser well known and older but still very well played titles always show a huge advantage for these companies. We don't see the same results in AAA games, but in the A games or the free to play titles its pretty common to see this.

 

I have two machines, the 970 based on and the Fury based one. I like to play minecraft with a tonne of mods, the Fury runs awful because AMDs openGL drivers are terrible, you have to turn the graphics down on a 4770k based machine because of the Fury card in it! An Nvidia 970 however runs it so well the GPU doesn't even wake up from sleep and where its basically maxed out on AMD and needs settings turn down, and the game is mostly CPU anyway. Its not one of those titles you will ever see tested on a big gaming site yet its the best selling games of all time. Why does AMD have a performance problem in such a big game? Gotr me, I never understood why I had white snow where the anti aliasing should be for years in Arma 2, its bugs like that which drive people to buy the other guys card.

 

This is the reason people buy Nvidia and Intel, its the reliability and consistency. Its probably for a variety of reasons that this dominance actually happens but its hard to ignore that this just keeps happening and has been happening for a long time. Its probably the companies involved just optimising for what they had or were given but it has a profound impact if you play anything outside of the AAA games. I thought the graphs were interesting and I think people should have a read of the site and see it for what it is and isn't, but its got compelling data for 6 core CPUs for games and it shows just how often AMD seems to have below par performance on its GPUs on less well played games and older titles.

[Aytex]

or you get the best bang for your buck....

 

also one benchmark for one game does not tell you if it is truly better than its competiter

[PlayStation 2]

Their CPUs sure, but on the GPU side, it's more of Nvidia's advertising when AMD is nine times out of ten the (normally vastly) superior choice.

[Nacho Marco Segui]

War thunder is Nvidia optimised,so... 

[App4that]

I can personally call bull shit on this as I worked for Gaijin and my 290 hit 128fps in the ground forces benchmark. This is just propaganda and should be ignored as such.

[manikyath]

so.. with big scope benchmarks like these, i generally ignore the actual numbers at first read, and search for obvious red flags.

 

one obvious red flag to look for is the performance of dual-gpu cards. if they are present in the listing, but sli/cf isnt supported in the benchmark (passmark does this with CF, its hilareous...) thats a HUGE red flag. its either they want to tak down on the specific dual-gpu card, or its that they didnt bother checking.

once again pointing to passmark, who is VERY guilty of this, and to note, is also very clearly more optimized to the nvidia side, with factual slower nvidia cards outpacing amd cards in the benchmarks.

 

while these guys dont seem *that* bad about it, theres certainly things here that make me discard it as a whole, because they clearly didnt even bother to cross-reference their own scores to make sure it even made sense.

[DXMember]

What the hell even is that game?

Googling it returns a huge GameWorks banner...

[Glenwing]

Every AMD card above the 290X is conspicuously missing from that graph.

[App4that]

Biggest red flag is the lack of listing which benchmark they used. War Thunder has two.

[Sauron]

Yeah, ONE BENCHMARK is definitely proof that nvidia and intel are the only solution.

[DXMember]

so.. with big scope benchmarks like these, i generally ignore the actual numbers at first read, and search for obvious red flags.

 

one obvious red flag to look for is the performance of dual-gpu cards. if they are present in the listing, but sli/cf isnt supported in the benchmark (passmark does this with CF, its hilareous...) thats a HUGE red flag. its either they want to tak down on the specific dual-gpu card, or its that they didnt bother checking.

once again pointing to passmark, who is VERY guilty of this, and to note, is also very clearly more optimized to the nvidia side, with factual slower nvidia cards outpacing amd cards in the benchmarks.

 

while these guys dont seem *that* bad about it, theres certainly things here that make me discard it as a whole, because they clearly didnt even bother to cross-reference their own scores to make sure it even made sense.

The review site is actually pretty good, I'd say it's on par with anandtech (when they actually do reviews) if not better.

They have THE best GTA V performance review on the network.

Didn't really read into this particular but I suppose they analyze and write down why the performance is as it is, at least most of the times with big games they do so.

 

I'd say the huge GameWorks banner is at fault this time.

[OddsCrazyStuff]

Fail benchmark, Fury X is missing.

[manikyath]

Every AMD card above the 290X is conspicuously missing from that graph.

remind me again if 300 series released before or after 980ti, and when the fury x happened

 

it might just be from before 300 series, they might just not give a duck because based on the 295x2 being in there kinda makes me think they're rather biased.

[manikyath]

The review site is actually pretty good, I'd say it's on par with anandtech (when they actually do reviews) if not better.

They have THE best GTA V performance review on the network.

Didn't really read into this particular but I suppose they analyze and write down why the performance is as it is, at least most of the times with big games they do so.

 

I'd say the huge GameWorks banner is at fault this time.

well.. if its an nvidia title that doesnt support CF, remind me why they bother testing the 295x2...

[Glenwing]

remind me again if 300 series released before or after 980ti, and when the fury x happened

 

it might just be from before 300 series, they might just not give a duck because based on the 295x2 being in there kinda makes me think they're rather biased.

In the OP he says they released those graphs today.

[manikyath]

In the OP he says they released those graphs today.

excuse me while i go cry in a corner...

[App4that]

well.. if its an nvidia title that doesnt support CF, remind me why they bother testing the 295x2...

It supports crossfire, and as I said I tested my Vapor-x 290 while I worked for them and got a score of 128fps. I also know the inner workings of the game, intimately. I know the priorities. This graph is udder bullshit.

[manikyath]

It supports crossfire, and as I said I tested my Vapor-x 290 while I worked for them and got a score of 128fps. I also know the inner workings of the game, intimately. I know the priorities. This graph is udder bullshit.

but... if it supports cf... why does the 295x2 perform worse than the 290?

 

i is all the confuse right nao...

[App4that]

but... if it supports cf... why does the 295x2 perform worse than the 290?

 

i is all the confuse right nao...

Because it's BULLSHIT :lol: You have no idea the drama that goes on. (NDA stops further comments) I can tell you the favorite GPU over at Gaijin is a 280x, so. 

[Nena Trinity]

This info is false...  (its funny how my FX-8370 with GTX 970 gets better FPS than that FX-9590 with GTX 980Ti...)

 

And even if was true we all know its clearly game breaks fault!  (GPU side...)

 

Edit:

I also forgot to mention I played War Thunder on Linux which Linus says perform so terrible, oooooh shots fired! (benchmark lie double true!!!)

[don_svetlio]

[App4that]

I love you man.

[don_svetlio]

I love you man.

[Dabombinable]

Let me see, the entire FX line is shit and it takes an FX9590 to compete with an i3 in most games, I could afford an i5 4440 and upgrade to a 4790K-both of which are far better for games, with the i7 being better at everything. And at the time the GTX 970 cost less than the R9 290X while performing as well and even better at times.

[don_svetlio]

Let me see, the entire FX line is shit and it takes an FX9590 to compete with an i3 in most games, I could afford an i5 4440 and upgrade to a 4790K-both of which are far better for games, with the i7 being better at everything. And at the time the GTX 970 cost less than the R9 290X while performing as well and even better at times.

nope and nope.