>people buy AMD GPUs for budget builds usually because they're supposed to be better value
>reviewers test the cards with the most powerful cpu available to regular consumers like the 5960x or 6700k
>people buy the budget AMD card with a budget cpu
>people suffer horribly from AMD driver overhead
What's the point of AMD? They're no good in the budget end because of the driver overhead, and they're no good in the enthusiast tier because Nvidia flagships usually perform better.
When will they fix their drivers? Or will they just force dx12 down our throats instead?
If they fix their drivers, both on windows and Linux, they will be a viable option for many people who understand the issue and go with nvidia because of it.
When will intel make a CPU fast enough that AMD's driver overhead isn't a big deal?
>Why do people buy AMD?
Tinfoil hats, autismos, hipsters, and fanboys.
There really is no logical reason to buy AMD, most people do it out of something stupid like "muh nvidia gimping meme"
>>55277635
>using GameWorks game to conclude AMD sucks
The RX 480
lol is /g/ really nvidia shilling now? times change huh
>>55277682
Here's an AMD sponsored title.
>>55277695
>actually expecting the same performance as benchmarks when it's a known issue that you won't
>>55277714
>AMD makes absolute shit so everyone buys Nvidia instead
>Somehow Nvidia is the shill company
O am I laffin
you nignogs are the ones posting your Rx 480 viral marketing threads and raiding every discussion related to the gtx 1080/1070
>>55277765
we're on a forum full of linux and gnu memes and nvidia isn't the shill company? 3.5 etc
you must be new here
>>55277776
>hirr durr 0.5gb of my ram is slower
Buy a 390 or 980 if vram is that important to you.
There is nothing wrong with the 970, it performs exactly as it should at that price point.
>>55277776
Even if youre a freedomfag its absolutely retarded to buy AMD
especially if you use Linux
>>55277802
Except that 0,5gb of slower vram will prevent games from using their vram management stuff properly resulting in poor performance. Fast 3,5gb > fast 3,5gb+slow 0,5gb
>>55277747
Is that the mantle version?
>>55277816
>Even if youre a freedomfag its absolutely retarded to buy AMD
especially if you use Linux
Lubuntu user with an AMD laptop here. It's a pain in the ass to get some games working. Had no issues with Windows 8.1 installed, but for some reason Linux doesn't work well with my integrated graphics. Windows was way too slow though, so I guess that's the price to pay.
>>55277870
Nvidia cards don't support mantle, so it's a dx11 benchmark to make it fair.
>>55277635
8370 @ 4.6ghz
290x TRI-X 8gb
3200x1800, all settings max. 60+ FPS
???
Benchmark accuracy when?
itt polack ""benchmarks""
>>55277882
TurionX2 with HD3200m here, open source drivers are actually decent.
>>55277852
And yet it's still manging to outperform the 290x/390 with its non gimped vram with ease.
That's how bad AMD drivers are.
>>55277953
This uses nvapi, so its unfair to AMD isn't it?
>>55277995
Source?
>>55278020
https://twitter.com/repi/statuses/383174148572196865
Specific nvidia only optimizations have been used without being officially announced in many games.
>>55277994
I know that's BS because my 380X + 8350 gets over 50fps
>>55277994
>unreal engine
>2k
Literally nvidia partnered game.
If you are ignorant, I'll remind you that they're withholding async shader on PC because of their nvidia partnership. They have it enabled on consoles due to AMD contracts.
They will not enable async shaders until Volta arrives or they will enable async shader that nvidia wants, aka one that benefits no one and decreases performance for all GPU on dx12.
>>55277635
Source? This is actually interesting.
>>55278189
Source is Gameworks games.
Anything with nvidia partnership means some sort of gameworks code or nvidia specific optimization running.
When gameworks games run on nvidia cards, their GPU can run the computations, freeing up the CPU for normal rendering. Because most games hardly max out GPU load 100%, this will mean nvidia GPU run better than AMD on same system. AMD GPU meanwhile won't be able to run proprietary gameworks code, so the game runs the code on CPU, which slows down the render time, thus lowers fps. This gives the perception that AMD drivers suck.
Granted AMD CPU are weak compared to Intel CPU, the nvidia gamework games compound this issue and creates a rift. If the same game were to run gameworks code on either CPU or GPU equally, the performance difference would be lot less and the difference between the two would be negligible.
tl;dr nvidia gameworks code reduces CPU's power by x% because nvidia wont allow AMD cards to run gameworks code on GPU.
>>55278268
>gameworks
What does that have to do with driver overhead?
>>55278268
Fuck off with your fanboy nigger shit. I asked for a source so I can make my own judgement and not because I wanted to here your buyers remorse butthurt.
>>55278307
AMD card not running gameworks title = "driver overhead"
So far, the examples posted are purely nvidia partner games or games that use gameworks.
>>55278338
Hear*
>>55278340
Like battlefield 4?
>>55277747
>>55278340
>bf4
>gameworks
>nvidia sponsored
amdrones confirmed for being blind retards
>>55278349
Uses nvapi, nvidia specific optimization api.
AMD optimization is mantle api, but using that is "cheating".
>>55278338
>>55278189
http://m.pclab.pl/art60000-21.html
>>55278364
>the drivers for graphics cards AMD Radeon more processor-intensive or less managed by its resources, so that the more he is working in a particular place in the game and they is weaker, the worse Radeon falls GeForce terms.Radeon R9 390 OC in conjunction with a very fast processor Core i7-6700K @ 4.7 GHz is able to establish an equal fight with GeForce GTX 970 OC-I.Unfortunately, in tandem with a weaker CPU noticeably losing distance.The problem in this case lies in the fact that it applies even overclocked to 4.5 GHz Core i5-4690K, which after all is still among the leaders in terms of performance.
Not comforting that AMD does not mention that intended to improve CPU utilization by drivers for DirectX 11 games.
Sorry for bad English, using Google Translate because I'm not polish.
>>55278364
I noticed all the games tested have nvidia slant.
>Crysis
Have been accused of crippling AMD cards with excessive tessellation for everything in game.
>Fallout
Gameworks title
>GTA
Gameworks title
>Witcher 3
Another Gameworks title
Not saying conspiracy, but is there a reason why these games are chosen for this "fair" test?
>>55277635
I've always considered AMD to be the poor mans card or for NEETs that had to save up but will end up in the buyers remorse threads and hating on those that were smart and went NVIDIA, NVIDIA is basically the Tesla of cars as to AMD which is like a Reliant Robin.
>>55277635
>graph literally says CPU test
Well what were you expecting?
Also
>using outdated graphs
>2016
I guess if you're going to go retard, go full retard.
>>55278410
Whoops, forgot.
>Crysis
Nvidia physX, aka another gameworks
Jesus Christ, these are all gameworks title.
>>55278410
you are right, all reviewers are shills, they should only test Ashes of Singularity, otherwise it is biased.
>>55278383
We know amd cards struggle with driver overhead. Only brain dead retards would think otherwise. It's a factual statement.
>>55278450
No, you're right DX12 is a meme. Ashes of Singularity is a meme. Async shaders are meme.
There are no DX12 games anyway. Long live nvidia.
>>55278410
Well then good thing I own an nVidia GPU with a 4770k.
>all the games ever are all gameworks & nvidiashilling engines !
>gonna keep buying amd that'll show em !
>>55278486
Short term, its fine. Long term, you have $1000 GPU that should be going for $400 or so. Unfair practice leads to monopoly, which leads to price gouging. If you intend to buy any GPU in the future, you'll be fucked.
Short term vs Long term happiness.
>>55278410
If every game ever is an nVidia optimized title, tell me why I shouldn't buy nVidia?
>>55278460
>Post Overclock vs Non OverClock
>Calls people brain dead retards
Just go to sleep man your night is over.
>>55278461
>Ashes of Singularity is a meme.
It is a meme though. Literally nobody plays it except benchmarkers or people who got it for free with their amd card. At least the people doing these tests are using popular games with hundreds of millions in sales as a reference point. What's the point of testing games nobody plays? All 4 of the games there were/are goty contenders/winners.
>>55277635
>>55277747
Doesn't the 290x outperform the 780 cards as of today due to the massive driver improvements that the GCN cards had? On top of Nvidia gimping 7xx card performance once the 9xx cards were out?
>>55278522
>being this stupid
It's not about the frame rate you dunce. Its about the frame times.
Did you get dropped on your head as a baby by any chance?
>>55278522
Does factory OC count?
Even then, that's a pretty embarrassing delta.
>>55278412
AMD isn't a bad option, but nvidia is definitely better.
I'm pretty satisfied with my i3 2120 and r7 260x, but that's only because it was so cheap. A$220 for the lot early january 2015.
>i3 2120
>8gb ddr3 1333
>300GB velociraptor
>150GB velociraptor
>asus r7 260x direct cu ii
Gives me 80fps on gta v with low settings, but some things turned up.
55-65 fps high settings battlefield 4.
1080p
>>55278527
Can you find a source that isn't using a 5960x or 6700k?
>>55278546
>game is optimized for Nvidia
>Nvidia card comes OC'd
>it performs significantly better
>pretty embarrassing
>>55278546
Wow you're getting stupider by the minute. You're saying because the Overclock was done at the factory and not by the user at home it should not count? Kid go to sleep.
>>55278508
No one's stopping you from buying either. Company engaging in dishonest tactics vs company investing in strong hardware. Both have advantages and disadvantages. Its up to you on how you decide with your money.
>>55277635
You swam all the way to Poland to cherrypick that graph? im impressed.
>>55278599
www.guru3d.com/articles_pages/gta_v_pc_graphics_performance_review,5.html
>>55278593
>vs company investing in strong hardware
Please don't tell me you're referring to AMD here. The nVidia cards run cooler and have better performance for anything that's not double floating point performance.
If I were looking to build a cheap bitcoin mining machine, or cheap AI system, I'd go with AMD. If I'm looking for performance/watt, I avoid AMD like the plague.
Not to mention LOLAMDDRIVERS.
>>55278599
The particular graph was for testing various CPUs with the same two graphics cards...
>>55278584
>gets btfo so hard that he decides to reply to some random who isn't even the op of the post instead of the actual op >>55278541
>mfw retard couldn't even understand the difference between frame time and frame rate in the first place
Yeah, your mom clearly dropped you off a balcony when you were a child. Some serious brain damage there.
>>55278599
The graph in the OP was testing CPU performance, you dumb fucking poorfag.
>>55278599
I'm not surprised an AMDrone tries to draw the thread away from the original discussion of driver overhead.
>>55278619
nvidia card runs "cooler" and have "better performance" is due to architecture and software direction of the games industry, which nvidia successfully courted. Not due to inherent stronger hardware. Their hardwares are infact weaker because of lack of certain features compared to AMD. They effectively cornered the gaming market with their partnership with benchmark maker, game engine maker and their acquisition of gameworks system.
You only need a small lead in the game to be considered the "King", then people will follow that regardless of how that title was obtained. Their lead with games industry is due to certain key maneuver, not out of actual performance increase, but through proper PR maneuver and marketing genius.
If you don't remember, nvidia drivers are the ones that have been killing the cards in the past couple of years. You were promised a "async compute" driver a year or so ago by nvidia, but this promise hasn't been materialized. Their Win10 drivers are lacking and you have crashes and unstable driver problems all the time.
Maybe bit more introspective will help you.
>>55278694
>>55278656
"driver overhead" aka gameworks code running on CPU instead of GPU
The proof of driver overhead is literally a nvidia meme
>>55277635
That graph doesn't show what you think it does.
In your graph, difference between 290x and 970 with 4460 is 12FPS/25%; difference between 290x and 970 with 8350 is 9FPS/25%.
Mine shows how poorly AMD CPUs work in GTAV with Nvidia hardware (i.e., removing AMD's GPU drivers from the equation).
Conclusion: No effect
>>55278711
>w10 drivers have been lacking
My gtx 680 performance has been roughly the same as it was on w7.
It should be between 5-10% better because of dx11.1, but it isn't.
That being said, until AMD fixes their overhead issue and stops trying to force dx12 to make up for their incompetence, they're not an option for me when I upgrade.
>>55278718
Got any proof? It's already a known fact that AMD drivers are limited to one thread while nvidia can use multiple threads.
Prove me wrong if you can, but there's plenty proof in my favor ITT, you've done nothing but shitpost in a pathetic attempt to defend AMD.
>>55278725
Your post doesn't even make sense, next time just post feet instead
>>55278744
I can't prove a negative. Its in your interest to show me that AMD has driver overheads. All the "evidence" thus far are purely gameworks title. Which in fact do not run on AMD GPU but rather on CPU.
>>55278711
nVidia can do async compute. It can even do so better than AMD cards for some queue depths.
>>55278711
>Their Win10 drivers are lacking and you have crashes and unstable driver problems all the time.
No they don't. I have owned a 980 since November last year when I upgraded from a 280x and I haven't had a single problem with it. I had more problems with the 280x like the cursor flickering shit than I have ever had with this 980. Now that I think about it, I haven't had a single driver related problem yet my friend who bought a 390 a few months before I bought my 980 is having nothing but driver problems with it. Not like the ones I had on my 280x but new ones like the marquee tool on Maya not working at all. It's a major issue and he was forced to downgrade to 2 year old drivers which obviously causes big problems in games. I don't think anyone I know has had a pleasant experience with amd drivers.
Also note that I'm speaking from experience and from factual information given from close fiends and not shit I read on WCCFtech like you're obviously quoting.
>>55278527
Test
>>55277747
This shows that the game is CPU limited, not that AMD's drivers are CPU heavy
>>55278778
Why does this "nvidia can do async compute" fail to show on all the DX12 games? Including the nvidia DX12 title, Rise of Tomb Raider.
>>55278782
I own a a GTX 770 on my side machine and the Win10 driver updates have been hit/miss. I always have to check threads on forums to see whether the drivers are safe or not due to how unstable the Win10 drivers are.
>>55278762
I mean I like feet, but why are you on the hardware board if you don't understand my post?
>>55278782
Gtx680 guy, me amd a friend (he has a 960) have the same problem since upgrading to 10.
Shadowplay will make the display drivers stop responding one in three times we use it.
My friend with a 980 ti, and another friend with a gtx760 don't have the issue we do so it's weird.
>>55278796
Except that it does show that because AMD cards take a bigger hit when the cpu is the bottleneck.
Did you even check what you wrote before posting?
>>55278824
>that performance gap on AMD
Jesus...
>>55278824
Cod aw
>>55278812
This is another gameworks title, so probable explanation I could give you is the gameworks code running on CPU taxing the system harder than with nvidia GPU.
>>55278848
explain cod >>55278838
and Rise >>55278824
>>55278694
There may be extra driver overhead, but these graphs sure don't show it
>>55278812
>>55278824
>>55278838
More evidence of the amd driver overhead. Look at those tragic frametimes on the amd cards compared to the nvidia cards. Literally unplayable.
>>55278859
or just explain this one (I've cropped the dx12 mantle results cause we're talking about dx 11 here) the actual results will be a little better on amd with latest drivers (as far as i know they have less overhead now but are far from nvidias level yet)
>>55278820
>Except that it does show that because AMD cards take a bigger hit when the cpu is the bottleneck.
It doesn't, retard. Do you even math?
>>55278859
Ryse is running crytek game engine, heavy nvidia leaning partner. As well as Gameworks title.
CoD AW, not sure. You may have something, but thats still only 1 possible valid instance.
Why would anyone use an AMD cpu anyways?
They're absolute dogshit.
Just get an i5 or i7 and you won't suffer from any driver overhead.
>>55278932
Pretty sure that Rise was a GAMING EVOLVED amd™ tittle still the overhead one here >>55278886 shows some bad numbers for amd on dx11.
Does amd supports dx11 command lists yet?
>>55278838
Could have cherry picked a bit and show how it performs with an amd apu as cpu.
Spoiler -52% perf
>>55278949
Because some people are on a budget, and you still get overhead on an i5, especially a budget i5 like the 6400, or an older one like the 2500.
>>55278862
Can you actually contribute instead of shitposting?
Or do you literally have no way of proving your point so you shitpost instead to try and persuade the weak minded sheep that maybe you have a point?
Fucking degenerate.
>>55277747
I dare you to run the same test with the latest drivers
>>55279033
If you really are on a budget you don't buy a 390/x
>and you still get overhead on an i5, especially a budget i5 like the 6400
Just get a 6600k which is a couple of bucks more and overclock it to heaven.
>>55278553
my shitty gayming laptop with a gtx850m and an i7 runs gta v at 55-60fps with things like textures on very high, only turning down shadows, reflections. i keep lighting at high and post processing on. runs at 70-75 average with spikes to 80
>>55279095
70-75*c ***
>>55279047
I would if I had those parts.
>>55279090
You're right, buy a 380 or 370 instead, where they'll suffer from the same overhead issue but no one will care because it's $10 cheaper than its nvidia counterpart.
Many people simply don't have the budget for a 6600k. I have a friend who went with an i3 6100 and gtx 960 because he didn't want to spend more on his pc.
So far it's working out great for him.
>>55279095
What resolution?
>>55277635
>gta 5
find a game that isnt a shit console port
>>55277661
>"muh nvidia gimping meme"
how many times do we have to prove it before you will stop being a cuck
>>55277765
put your tinfoil hat back on nvidiot shill
>>55279413
GTA v was actually pretty well optimised for a port.
>>55279413
Got any proof for the gimping, or just speculation.
Updated drivers have fixed all issues you AMDrones have pointed out so far.
>>55279413
>gta 5
>shit port
Are you stupid? Gta 5 has been one of the best ports in ages.
Amdrones confirmed for not actually playing games.
>>55279413
Kill yourself
>>55277635
>Or will they just force dx12 down our throats instead?
oh yeah because every LOOOVEs rendering pipelines that havent been redesigned since the mid 2000s
>>55277635
>AMD SUCKS BECAUSE THEY CANT IMPROVE THEIR DRIVER
>DX12 SUCKS BECAUSE IT IMPROVES AMD DRIVERS
REEEEEEEEEEEEEEE
>>55279489
It's what they're doing now.
>>55279509
Pure autism.
Kys.
>>55279047
Crimson didn't improve performance on older games.
It didn't do shit for overhead, it's just a reskinned catalyst.
>>55277635
>clinging to the past
Do you enjoy shitposting on your i386 with onboard video since you don't need future tech?
>>55279792
Driver overhead is still a issue for AMD.
It's why they're pushing so hard for low level APIs like directx 12, and donating mantle for vulkan and dx12.
I was told the 8350 would be enough for gaymen
>>55279913
It is if you use an nvidia gpu
Don't forget games get performance patches and most of these are launch benchmarks.
>>55279413
>shit console port
Son...
>>55278527
>gimping
Amds relative performance increasing 2 years after their cards come out because they are still sorting out their shit drivers means nvidia gimps their cards.
Nice meme
AMDs best bet right now is to get a LOT of cards into peoples homes at all cost, and offer devs something similar to gameworks. They probably can't afford all that though
>>55277655
around the same time amd does it themselves. it's not happening.
>>55277635
>AMD BTFOing every nVidia mainstream GPUs in 2 days
>envytard damage control threads 24/7
>amd fag needs argument
>only has dx12
>dx12 has no games, and requires botnet.
kek.
>>55280020
AMDs biggest weakness is drivers, once they sort that out they should work on gameworks, tesselation, etc.
>>55280041
>>55280028
I wish these shit bait posts would leave the thread, you're not even contributing anything.
>>55278711
Enjoy your theoretical performance.
>>55279047
But i thought amd having shit drivers they need to work on for years to sort out was a meme?
>>55280058
It's an exaggeration, they've been improving their drivers a lot lately but they decided to place their bets on dx12 instead of fixing their drivers.
>>55277635
>cheapest gtx 960, gainward phantom A$231
>cheapest r9 380, asus strix A$279
So GTX960 costs less and performs better while consuming less power overall?
Why is the 380 shilled so much on /g/ then?
>>55280047
amd's biggest weakness is relying on dx12 to save them. hardly anyone is using it and no one wants the microsoft upgrade.
the bait posts you think are bait are calling out amd's bullshit. get the fuck over yourself.
>>55279934
So if AMD cards suffer from drive overhead and have worse frame times than Nvidia, and Intel CPUs that have half the cores and running at 50% lower clock speed is faster than AMD CPUs...
Does that mean that it's stupid to get AMD/AMD or even AMD/Nvidia or Intel/AMD since one way or another the AMD part will drag down the other hardware down?
AMD is therefore by definition a bottleneck.
>>55280047
>drivers meme
>contributing
>>55280289
Pretty much.
AMD is only a good idea for SUPER BUDGET builds, where you are trying to save every penny. Even then it's probably not worth it because you will have more hardware failures and it won't last as long.
>>55280176
Those prices are wildly inaccurate, are you retarded?
>>55280327
But then I'm just gonna buy a cheap ass i3 since it have built in gpu if I'm going super cheap.
And since even i3 have good IPC if I then get more money I can boy a mid end GPU from Nvidia later.
>>55280339
AU$, not U$A
>>55280400
Integrated graphics are still awful for any decent game. You're better off getting a cheap AMD GPU if you want to save the most money possible.
>>55280176
>cheapest r9 380, asus strix A$279
>>55278505
What a fucking meme post.
Get a card that performs significantly better nie or one that will be slightly optimized vy drivers in a few years when it's time for me to upgrade anyway.
>Futureproof buying tech
well memed
>>55280502
4 years is future proof
280x is still viable option 560ti is dead
380 is viable option 770/960 is dead
290/390 is getting better while 970 is getting worse
>>55278410
>Have been accused of crippling AMD cards with excessive tessellation for everything in game.
Wrong game. Unlike the previous Crysis games, Crysis 3 is an AMD sponsored game, part of the gaming evolved.
>GTA
GTA V is both gameworks and gaming evolved.
So AMD gets 1.5 games, nvidia gets 2.5.
>>55280530
None of this is true
>>55280530
My 670 still runs most games on high. Now I want to max games, so I'm getting a 1070 because I want the performance now and not in 3-4 years.
>>55277882
which cpu are you using?
>>55277635
While you are right. 290 and 970 aren't buget cards.
>>55278762
Who is this semen demon?
>>55280467
Check MSY.
>>55280530
>280x vs 560 ti
>380 vs 770
Why are you comparing cards from different generations?
>>55280706
guess I should have changed my 770 for 670, it runs better apparently
>>55277747
why the fuck they test amd cpu @4.7ghz in gtaV test but @4ghz in battlefield test ?
>>55279287
1920x1080
>>55281210
But that one has oc and non oc.
>>55280554
>1.5
>2.5
You are one autistic fuck.
>>55278059
well thats kind of damning.
>>55281789
That AMDrone in the comments though
>>55277995
The NVAPI can detected if a non nvidia GPU is being used as well as what kind of architecture is being used in case of Nvidia, if the game uses NVAPI for AMD cards the game wouldn't even start at best you get a game that would start but crash all the time.
AMD also has his equivalent to NVAPI which in fact you need to get your game to work on their GPUs because the huge difference between architectures and how they handle task you can't use NVAPI on AMD cards without huge problems and vice versa.
Now this is where things get bad for AMD, lets start with the DX API which is designed by Microsoft. Microsoft releases the API with the features which are considered a task to be done either in software or hardware mode together with examples of how it CAN be done and what exactly has to be done by the task, meaning this feature (task) should do "exactly this", however it is up to the hardware manufactured to decide how they want to do the task, either they stick strictly to the Microsoft way or they go their own way, which doesn't matters as long as the task does "exactly this"
The problem is that if they want to go the Microsoft way which is the standard way they would need to change their hardware architecture to get it done the standard way, in some cases their own way is even better than Microsoft's way, if you can get the task to do "exactly this" and do it faster than the standard way without having to redesign your hardware then by logic this becomes the best way of doing it even if it's not the standard way of doing it.
This is the problem AMD faces, their "own" implementation is worse than Nvidia's and I'm not talking about NVAPI here I'm talking about their own tools, even their "standard" way of doing it ala Microsoft has problems according to different game developers who have made the occasional comments about this problem while Nvidia is always praised for having great tools and support.
>>55282081
It's a shame AMD can't into software, their GPUs are beasties with big potential.
>>55282100
No kidding, none of the currently released AMD GPUs are fully OpenGL 4.4 complaint let alone 4.5 even though their hardware supports and they have working extensions, 4.4 was released two years ago.
Now people will say it doesn't matter since barely no game uses OpenGL but you can still use it as an indication of how poor their software team is, even Intel got it.
>>55282124
Doesn't doom use opengl?
>>55282157
Yeah, 4.5 on Nvidia and 4.3 on AMD.
Like I said they work some working extensions but none of their cards are fully compliant while on the Nvidia side even old as fuck shit before Fermi is fully complaint.
>>55282168
>before fermi
Like my gtx260?
>>55282191
Couldn't tell you, I can't find the compliance list on the OpenGL website and I don't feel like going through google just for this.
>>55282221
Opengl 3.3 only for 200 series nvidia
>>55282249
Well give me a break, I was pretty close.
I fell for the AMD meme
Cheap CPU and Cheap GPU, both seemingly not going to bottleneck each other in any way..Yet terrible performance?
These fucking 8 core memes honestly useless price to performance is WORTHLESS when your actual performance per core is terrible.
I beg you do NOT buy an fx8320/50, they are not worth it. 4 of your cores are absolutely useless for most games and the other 4 work about as good as an intel CPU from 7 years ago. Spend the little extra and buy intel I promise it's so much better
It's so bad that DISABLING 4 cores in dying light gave me 15 fps increase
>>55277635
This.
AMD needs to fix their shit before they actually become a brand to use, whether budget or high end.
>>55277747
I have an R9 280 and fx 6300 and I never drop below 60fps on ultra in that games...
>>55282341
It's not bad if you use an nvidia gpu.
I have a friend with an overclocked fx6350 and gtx960. He's happy.
>>55282360
I was using it with a 760 and I couldn't get 60 fps on lowest settings on so many games, only games that even worked remotely well were ones where a toaster could run it.
I'm talking LOWEST settings like going into the configs and setting things to abnormally low below low and having to change things in nvidia control panel to performance just to get sub-par performance.
>>55282388
He mostly plays bf4, gta v, and steam games like planetary annihilation.
>>55277816
i bought an AMD GPU because their FOSS drivers are better than nvidias ones and won't break when i upgrade my kernel. it is just less of a hassle.
>>55280018
It's not a meme that Kepler really got fucked over when Maxwell came along. Example: TW3 with those ridiculous GameWorks settings. OTOH i've read that kepler lacked on the arch side regarding some modern features making it horribly outdated at launch. Or blame optimizations for consoles with AMD arch. Probably a combination of those two.
Will be interesting to see if my Maxwell will face the same future now that Pascal is out.
>>55282544
It will, no doubt about it. We're talking consumer electronics, a realm where planned obsolence is the norm.
>poorfags still thinks 480 is going to launch at $200
>>55282168
wasn't the beta using 4.5 on amd? then a hot fix made it work on amd?
>>55277953
If I had an AMD gpu I wouldn't use the dx11 version just to be 'fair' to the nvidia gpus......
>>55283196
no, they hot fixed 4.5 to 4.3 on AMD.
AMD just doesn't support 4.5 properly.
>>55278762
this
>>55281273
My 980m could do high/ultra settings hitting 75 and it's about 4 times more powerful. So I'm calling bull.
i like radeon cards, but even i'm not stupid enough to buy an and cpu
>>55283272
not him, but remember that GTA V is a CPU (and memory) intensive game which means it's an outlier nowadays. could very well be your processor bottlenecking your lappie
>>55283326
*amd cpu
>>55278338
its good to see you addressed his argument instead of insulting him, now that would be the actions of a true nigger
>>55282544
It's not so much that keppler got gimped, but rather that the hardware just isn't as capable as maxwell for modern gameworks features like 64x tessellation.
>>55277747
I have an R9-290 and a Xeon 1231v3 and the game runs at 80-100fps on Ultra, 4xMSAA using Mantle.
>>55283955
Thread discussing driver overhead on dx11 people still trying to talk about mantle and dx12.
Yeah we know that mantle dx12 has less overhead but it'll be nice if amd can fix dx11 on their side too.
I think that nvidia has done something that allows theme to use dx11 command lists even in games that are not using this feature while amd has a half assed command lists implementation that it's just enabled in some titles (civ 5)
>>55284055
Mantle is the amd equivalent to nvapi, but game developers didn't use it. Review sites review mantle games as "seperate" entity. Users have no demand for it. Meanwhile, its ubiquitous for people to use nvapi api for nvidia optimizations in games.
Its good that they scraped the mantle and went with Vulkan/DX12.
>>55277635
I got a 280X and a i7. Don't know how many frames and don't care. Runs great on highest settings...
>samefags shitposting non-stop
nvidiots are gonna kill themselves come june 30th
>>55278138
>>55284127
Also: Mantle hasn't been scrapped - its powering AMD's VR technology (and equally its why AMD is ahead of Nvidia for VR latency).
>people buy flagship nvidia GPUs
>Have it gimped to midrange AMD levels a year later
>Have it further gimped to entry level amd levels a year after that
nvidia dindu nuffin wrong
AMD fanboy lies. NVIDIA SUPPORTS ASYNC COMPUTE.
>>55284206
Kill themselves from laughing so hard
>>55284507
>>55284555
what is the source of all of that spilled salt?
>>55277635
Did you guys saw the 80c and 88c on the R480 benchmarks?
>>55283669
i can sense sarcasm in your post, strange because i literally wrote "ridiculous gameworks settings".
>>55282081
if amd would work on optimizations for games with their cards then nvidia wouldent be the only one, but amd is lazy and relies on shills like you to prove they are being opressed so u should buy their cards becuase you feel bad for them
>>55284564
No idea - I found it on OCN but fuck knows where it originally came from.
>>55278725
\thread
>>55278812
is this with the crimson driver?
>>55277635
GTAV is nvidia optimized and running it at those settings is murder for an AMD card.
Otherwise my 7850 worked just fine when I had my 750K CPU
>>55277747
The 780 is only getting 57 fps here, how in the hell is their score so much higher?
>>55284942
and an even wider gap appears with slightly lower settings
>>55284942
>>55284958
>that resolution
Even when the 290x was new 1080p was extremely popular so it makes little sense to test at 1680x1050 unless you are trying to demonstrate differences between cpus.
>>55280176
the 380 is overall faster than the 960
cheapest 380 is $164
https://pcpartpicker.com/product/fsZ2FT/asus-video-card-strixr9380dc2oc2gd5gaming
960 $169
https://pcpartpicker.com/product/ZNGj4D/pny-video-card-vcggtx9602xpb
Stop posting
>>55284989
didn't even notice that, fuck tech spot
>>55284942
>>55284958
>>55284989
>>55285003
1920x1200 because tech spot is retarded
290x is still in the lead
>>55285014
same story again disabling MSAA
>>55285014
>>55285034
Intel Core i7-4770K (3.50GHz)
x2 8GB Crucial DDR3-2133 (CAS 11-12-11-24)
Asrock Z87 Extreme9 (Intel Z87)
OCZ ZX Series 1250w
Crucial m4 512GB (SATA 6Gb/s)
Microsoft Windows 8 64-bit
Nvidia Forceware 331.65 Beta
AMD Catalyst 13.11 (Beta 7)
http://www.techspot.com/review/734-battlefield-4-benchmarks/
with an even slower CPU than what's used there, so the other website is seemingly doing some shady shit
>>55284942
>>55284958
>>55285014
>>55285034
>GTX TITAN
HAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHA
>>55285080
The titan has never been the ideal gaming card, the titan X is a horrible abomination, The Titan Black comparably actually has more double precision performance than it and was a pseudo workstation GPU
>>55285080
As a mini quadro that card has aged like wine.
As a gaming card its aged like milk (as do all of Nvidia's cards).
>>55285190
Don't they still force it to use the geforce drivers and not the workstation drivers?
at least with the Pro Duo you have the option
>>55285202
Yups - its under the GTX branding and uses the normal drivers for their gaming cards.
The pro duo is a really niche card but its a monster at what it does.
>>55285324
AMD only made it to say they have the fastest dual slot GPU
>>55285364
They didn't need to for that - the 295x2 still has no competition on paper. The pro duo is essentially aimed at VR development.
>>55285752
>The pro duo is essentially aimed at VR development.
kind of
and the 295x2 is beaten out by a 1080 no? Or the devil dual 390 card potentially
>>55277635
Seems like a pretty decent argument and one that people buying AMD should definately consider if they are running a budget CPU.
Ofcourse AMD fanboys will just foam at the mouth, but take a minute to realize that this is a valid argument to consider if you have a budget cpu
>>55285817
The criticism that benchmarkers use high end CPUs is still a retarded criticism.
Yes, just run benchmarks with different CPUs all the time, so we have no idea if the different results are from the graphics cards or some other part
And to keep all parts but one the same means you need a high end CPU to not bottleneck the higher end GPUs you're testing.
Benchmarks are about isolating the GPU's performance so you can buy the best one for your money.
>>55285932
It's not retarded if the average consumer uses much less powerful cpu's in which case the driver overhead on AMD cards can become a problem
>>55285786
>and the 295x2 is beaten out by a 1080 no?
There is a reason why I said on paper - nothing except the pro duo within the radeon or gtx lines has enough compute power to compete against the 295x2.
Its the number AMD cited when (at the time) claiming the 295x2 is the fastest consumer gpu in the world. Its a technically correct statement as it beat out even the titan z for single precision compute.
>>55285932
So how would you know if your CPU is fast enough?
Look at the CPU reviews.
Where they isolate the CPUs and keep as much as possible the same, including using an unrealisticly (usually) high end GPU setup, to not bottleneck the CPU.
If CPU A gets 50fps, and CPU B gets 30fps, and GPU A gets 50fps and GPU B gets 30FPS, you don't buy CPU A and GPU B together
It's pretty fucking simple but over and over again people complain that test setups are unrealistic
OF COURSE THEY ARE
>>55285953
See
>>55285985
You're a level of retarded where euthanasia is a mercy to humanity
No idea what you are saying op. My hd7850 has been working just fine for the past 4 years. No problems at all
>>55285985
Right, so the conclusion is, do not buy an AMD card unless you have a high end CPU
>>55286034
The driver overhead thing is a dumb meme
Seriously
You won't hear a damn thing about it outside of /g/ because it's not a real thing.
Nvidia's drivers have overheads too, and they're very similar. The differences come when AMD's cards actually try and perform on goyworks- it doesn't increase CPU overhead, it just gimps the card.
>>55286034
Times are changing.
>>55286095
>its just a beta, nvidia driver will fix async compute
>its only one game
>its only some games
>who needs async compute anyway
>dx12 is stupid, who cares
>>55286063
Some guy on overclock.net did an exhaustive test of several games on a 290x and 780 Ti by matching each with an 8350 and a 4690k to show differences in driver overhead.
On average, the weaker gaming CPU (8350) was about 10% slower with the 290x relative to having that same 290x paired with a 4690k. So, it wasn't even a big deal 2 years ago and Win 10 actually did improve performance across the board for AMD. They've tightened things up since then and DX11 is only going to be relevant for so long.
>>55286191
>driver will fix async compute
>>55286191
>muh ASOTS
>every time
>literally a game that nobody plays
>>55286215
Do you mean this?
http://www.overclock.net/t/1534128/vishera-vs-devils-canyon-a-casual-comparison-by-an-average-user
>>55286095
>fury x can't even get 60 fps in the most heavily favored AMD game
What a turd
>>55286232
Given vishera is quite a bit older than skylake this is a damn good showing.
>>55286260
Don't worry, a 1080 barely outperforms a fury x.
>>55286242
No. Found it pham.
http://www.overclock.net/t/1495236/amd-vs-nvidia-cpu-overhead
>>55280176
>being australian
wew lad back to /s4s/
>>>/s4s/
>>55277802
>There's nothing wrong with the 970
AHAHAHAHA
>>55277635
>>55277747
>>55278599
>>55284942
>>55284958
>>55285014
>>55285034
>Old outdated benchmarks
>No 390 or 390X
Fuck off
>>55286280
In muh cherrypicked benchmark maybe
>>55286378
what game is that
>>55286577
mirror's edge catalyst
>>55286411
just add 10% to the 290/290X
>>55286612
Yeah >>55286378
>>55278728
>It should be between 5-10% better because of dx11.1, but it isn't.
Nvidia doesn't support Dx 11.1.
>>55286555
Actually if you understand what the chart is showing you'll see that both the fury x and 1080 are held back by the cpu.
>>55286555
all directx 12 games look like that
http://www.techpowerup.com/223669/geforce-gtx-pascal-faces-high-dvi-pixel-clock-booting-problems
