http://www.gamersnexus.net/news-pc/ 3016-amd-moves-vega-56-e

>>61845911
say it with me: 1060 performance

>Although we were told we wouldn’t be sampled, following our defiance of AMD’s decision to express favoritism by permitting only select reviewers to publish early Threadripper data, we will still have Vega content ready on embargo lift dates. We’ve sourced information and parts elsewhere. Turns out that talking about thermal compound efficacy on an IHS is disallowed, despite the fact that TR’s thermal performance was already published by outlets expressly permitted to break embargo.

kek, AMD hates him now

>>61846448
Well, making clickbait AMD-related videos is THE way to get on mommy's blacklist.

File: pajeet koduri lab.jpg (1MB, 3264x2448px) Image search: [Google]

1MB, 3264x2448px

>>61846448
>talking about thermal compound efficacy on an IHS is disallowed

>>61846481
t. leather jacket man

>>61845911
Gamernexus never liked AMD, they shit on them every time they can and they were the ones making benchmarks of Ryzen with games at 240p just to make intel look good.

>>61846464
>>61846491
>samefag buttmad amdrone
>it's OK when amd resorts to shady practices

is curlyhair reaching hardocp level of butthurt?

>>61846582
hardocp is on amd payroll since ryzen
he even did the vega """"benchmark"""" with the broken memesync samshit uiltrawide meme monitors

>>61846582
his beef is pretty justified imo, it makes no sense to disembargo select reviewers for random benchmarks then turn around and cry when other reviewers release exactly the same information. it's pretty different to kyle who wore his bias on his sleeve and spat and pulled hair for no discernible reason every time AMD did anything ever.

>>61846968
it comes down to him being butthurt about less views, as a consumer I don't care who gives me the info if it's true
either way proper way is to cross check between several reviews

>>61846968
is steve a jew

he gives me that vibe

>>61846448
>>Although we were told we wouldn’t be sampled, following our defiance of AMD’s decision to express favoritism by permitting only select reviewers to publish early Threadripper data, we will still have Vega content ready on embargo lift dates. We’ve sourced information and parts elsewhere. Turns out that talking about thermal compound efficacy on an IHS is disallowed, despite the fact that TR’s thermal performance was already published by outlets expressly permitted to break embargo.

helps that you tried to be hardocp

>>61846513
>please also bench at 4K
>ooooooooooooyyyyyyyy veeeeeeyyyyyyyyyyyyyyy!!!!

>releasing 56 benches 2 weeks early
will there be better drivers in 2 week or something?

>>61847435
It really depends if they can *fix* memory bandwith issues for launch.

>>61847435
isn't 56 also coming out monday?
why two weeks?

>>61846870
>memesync
lol, intel adopted

>>61847736
According to (((Steve))) Vega56 launches on 28th of August.

>>61847780
it actually might be true because initial rumor was that 56 is aib only
who cares though, more importantly we will get our benchmarks monday

>>61847941
Yes, but 56 will be DOA once 1070 returns to MSRP pricing.

File: Capture.png (30KB, 606x249px) Image search: [Google]

30KB, 606x249px

Why did he delete this?

>>61848032
Looks like it's still there.
He means they fixed the bandwith regression?
That'd be very nice.

>>61848032
He was called out for leaking info to miners basically baiting them to buy it for mining instead of gaming which we all know is going to be a huige failure.

File: 9y5R5Pc.png (124KB, 900x775px) Image search: [Google]

124KB, 900x775px

>>61848060
The message is still there.
Also
>leaking
The dude is CEO of RTG you moron.

>>61848054
Why is twitter fucking with me? It wasn't there a minute ago.

>>61848060
>which we all know is going to be a huige failure.
we don't know anything yet.

>>61848070
Twitter is evil.
Anyway, just look at Vega's bandwith deficit.

>>61848107
We don't even know why it's effective memory bandwith hangs at 1080 levels. Its peak is 484GB/s, same as 1080ti, but effetive bw is nowhere near that.

>>61847123
Steve is a good Christian boy.

>>61848240
t. Steve

File: topkek.jpg (30KB, 902x68px) Image search: [Google]

30KB, 902x68px

>>61845911
>http://www.gamersnexus.net/news-pc/3016-amd-moves-vega-56-embargo-forward-prioritizes-over-64

>p-please review this card first, d-don't say too much about the other card

Wew this doesn't bode well for the big Vega card

>>61848267
Full Vega10 feeds like three lineups, gaymen, pro and instinct.
Selling it to gaymers is the last thing you wanna do.

>>61848279
What are you talking about you idiot, RX vega are gamer cards

>>61848286
Full 4k ALU Vega10 feeds three fucking lineups you retard. Selling it as low-margin trash to gaymers is the last thing AMD wants to do right now.

>>61848311
Do you live in al alternative reality?

We are talking about RX Vega 64 here, a GAMING card.

>hurrrr it's not a gaming card

Fuck of with your delusional bullshit already

>>61846509
Just because you love shitty products doesn't mean amd products are good.

>>61846509
You mean they don't pat them on the hat and treat them like the little kid that needs protection like alot of other reviewers do.

>>61848327
N/v/idiots really are retarded. The full sized Vega chip in Vega 64 is the same full sized Vega chip that is used in the much more profitable (for AMD) WX Vega, SSG Vega and MI25 Radeon Instinct cards. Since the profit margins on RX Vega 64 are the lowest of any use of a full sized Vega 10 chip, AMD would rather push Vega 56 to gamers as much as possible.

>>61848327
Holy fuck, full Vega10 feeds three fucking lineups, cutdown Vega10 is present in only one.
They want to sell as much cutdowns as possible.

>>61848279
Vega cater only three categories
1.amd drone mindless fanboy
2. Miners
3. Retards who buy imacs

>>61848363
Nobody cares you stupid cunt, that's irrelevant to what is being discussed here, the gaming performance of the RX Vega cards.

>>61848374
Hello Jensen!
Don't worry in 3 years you'll be able to compete with SSG. No worries!
Yours, mr. Pajeet

>>61848375
Holy fuck you are so fucking retarded.

>>61848113
wich is wierd as it has a way bigger bus and is bound directly to the gpu

>>61848382
Nvidia already DESTROYS your stuttering SSG garbage

http://www.tweaktown.com/news/58669/nvidia-beats-amd-real-time-8k-video-editing/index.html

>AMD's demo of the Radeon Pro SSG was stuttering in front of our eyes

>>61848399
No you, now fuck off

>>61845911
>RTG is finished in the gaming GPU market. Instead they are profiting from miners and professionals and they will have extra money flowing in not only from GPU sales but from the success of Zen related products. This means they will have plenty of cash for further development of Navi and other exciting GPU related products.

FTFY

>>61848404
It's magick.
>>61848406
Low quality bait.

>>61848412
>wait for Navi!

>>61848371
dont you mean as little? margins on cutdowns are way lower as the cost to produce is the same as they derive from the same wafer as the big ones
they might wanna sell as many as possible to reduce throwaway

>>61848406
>AMD has promised to send me a Radeon Pro SSG when it launches
That's some silly way to get review sample.

>>61848436
Back to /v/ Shoo! Shoo!

>>61848463
>trigged AMDrone

>>61846448
AMD is as fucking evil and awful as Intel and Nvidia but freetards and /v/ tier shills continue to claim otherwise. Thank god I'm not a fucking prosumer, having to choose between two electric furnaces in order to do your work has to be a nightmare.

>>61848267
they learned from 1800x marketing failure
if they pushed 1700 first it would've been much smoother launch

>>61848507
CPUs =/= GPUs.
You need to launch the biggest and the baddest GPU first to sell the rest of the lineup.

>>61848525
people look at prices not at "whatever $2000 gpu I will never buy" can do

>>61848382
Fuck your shitty Vega with us readyboost

>>61848525
Like the Vega fe flop?

>>61848545
>"people"
Gaymen GPUs are bought by (man)childer. Marketing is all that matters.

File: fuckinhell.png (285KB, 1009x953px) Image search: [Google]

285KB, 1009x953px

>>61848375
>Hurr Durr AMD's obvious reasons for why they are pushing Vega 56 over Vega 64 is irrelevant to why they are pushing Vega 56 over Vega 64

>>61846509
I love how prior to release of Poolaris you shills were naming GN as one of few trust worthy reviewers cuz of your delusional fear of conspiracy against your flopping products just so you can start calling Steve shill.

>>61848550
The Obvious reason they have for pushing 56 over 64 is because , as all the leaks have suggested, 64 adds very little performance at a huge amount of extra powerdraw.

Not your imaginery bullshit reason

>>61848573
Steve please.
>>61848577
Show me the benchmarks please.

>>61848054
He means that Linux drivers are absolute garbage.

File: Vega_FE_Memory_Bandwidth.png (129KB, 800x574px) Image search: [Google]

129KB, 800x574px

>>61848577
>Hurr durr I don't even realize what will happen if RTG manage to fix the memory bandwidth issues on Vega 64

>>61848609
>Just wait!

How about AMD just releases a product that functions well at launch

>>61848618
The entire Pro lineup works as well as it gets.

>>61848637
So why can't they get it right at launch with Vega

>>61848643
RTG are pretty smoll to do everything at once.

>>61848654
So Vega is a faulty product at launch

>>61848656
RX Vega is not even launched yet, anon.

>>61848609
Muh secret features

>>61848637
Like fe shit? They even cloned fe with some frontier shit name just to clone Nvidia fe. Fuck you knock off Indian products

>>61848569
I bought a gayman gpu for games. All I did was look at benchmarks from a couple of sources and then compare prices. I imagine many do the same, although I've seen a few people who believe GeForce cards are always just better at all price points.

>>61848065
For fuck's sake how does Vega have worse bandwidth efficiency than Fiji.

File: b3d_fillrates.png (62KB, 1016x1203px) Image search: [Google]

62KB, 1016x1203px

>>61848744
>Muh secret features
A glaring obvious memory bandwidth bug is not a secret feature. That being said, if RTG is able to fix it, Vega 64 will see significant performance gains as it appears to be totally memory bound in gaming performance.

>>61849157
It's probably software bugs because self-evidently that regression could not have been intended and if it was a hardware bug there was plenty of time for a respin.

There is one outlandish theory that its the result of RTG sandbagging like motherfuckers by using the (locked) BIOS to disable one of the two HBM2 memory controllers and thus artificially limit the memory bandwidth to a maximum of 307 GB/s, but I think that's a little far fetched (though it would be utterly hilarious if true).

>>61849258
That's outlandish as all hell because they'd be locking out half of the VRAM channels to L2 cache and starving the chip so hard we wouldn't be seeing 250-300w power use.

>>61849333
Yeah, I think there are BIOS issues rather than anything else. The real question is to what extent RTG can fix them.

>>61849333
c-states are also disabled
so power use in this range is normal for a chip pulling current constantly
same happens when you disable c-states and HPET for ryzen it goes in overdrive 200w
AMD reuses IP across departments all the time, it might have ryzen power delivery regulators which are pretty good

so I have a little bit of hope for vega for now because AMD themselves never stated official TDP on it

>>61846968
>>61846481
There was numbers, specifically loaded CPU temperatures. Steve likes numbers. Naughty naughty, no numbers allowed.

>>61849936
Lol

https://www.techpowerup.com/236055/amd-asks-reviewers-to-prioritize-vega-56-over-vega-64

>>61848497
freetards not anymore. fx was last free cpu

>>61850068
#rekt

http://www.sapphiretech.com/productdetial.asp?pid=AE163547-6889-4073-BBF6-9DB2C45BB233&lang=eng
>750 Watt Power Supply (Suggestion)

oh fuck that, I guess I'll stay with 390 Nitro

>>61849936
you're such a salty bitch you know that

>>61850473
oh apparently nitro has same power reqs on sapphire site,still keep a hope for 250w mark

>AMDs slides show up to 30% memory bandwidth usage reduction when DSBR is active. This clearly shows that their tile-based renderer and on chip-binning ("fetch once") is working properly.
>However we don't know whether the occlusion culling ("shade once") is enabled and/or working. Per pixel occlusion culling would directly reduce the number of pixel shader invocations, and would bring other gains (reduced ALU/TMU use) in addition to reduced memory bandwidth use. "Shade once" should improve performance also in games that aren't memory bandwidth bound.
Sasuga sebbbi.
So can anyone guess where exactly did AMD fuck up?

>>61850633
I believe the current best theory is RTG actually has implemented most of the bandwidth saving features in Vega (e.g. primitive culling and DSBR) but Vega is still being held back by ~20-30% regressions vs. Fiji in raw memory bandwidth, effective texture bandwidth, and texel fill rate.

The remaining question is whether the memory bandwidth issues can be fixed by BIOS updates, because as it stands Vega 64 is leaving a crap ton of performance on the table by being so heavily memory bandwidth bottlenecked. Based on what we know, if you increased Vega's memory bandwidth to match Fiji, Vega would instantly get ~20% faster in games.

>>61850723
Right now Vega has the same effective memory bandwith as 1080...
...and performs the same.
I wonder how exactly did RTG fuck up effective memory bandwith so much?
Something to do HBCC implementation?

>>61850748
Who knows? The one hope is that the fuck up is in software and could thus be fixed with BIOS updates sooner or later. We aren't going to find out much until Monday anyway, and even then we have to hope someone has enough sense to put it through the B3D Suite.

>>61850778
>even then we have to hope someone has enough sense to put it through the B3D Suite
G*rmans.
G*rmans will do it.
Hopefully.
The rest will be too busy publishing clickbait articles.

Why is Steve Burke so openly hostile against AMD? You can easily tell in any video in which he reviews anything AMD-related.

What is the real reason AMD can't hold a candle to Intel? Every time they hype something it's a fucking bust. i don't get it

>>61850815
He want's to be (((Kyle))) since Kyle himself sold his soul to mommy.
>>61850825
They literally crucified Intel at every price point in every market besides laptop one where they have no products yet.

HotChips soon, faggots.
Maybe Samsung will start talking their low=cost HBM again.

>>61850815
I never get that feeling from his videos.Why does AMD prop up linus for little kids? That vega video was like daycare

>>61846448
>Break embargo
>AMD doesn't send them samples anymore
>Write snarky paragraph about how "apparently" you're not allowed to break embargos
What fucking retards.

A stock Vega 64 air on the newest development driver apparently posted a Firestrike score of 22,996: http://www.3dmark.com/compare/fs/13332822

An overclocked result earlier today on the same drivers has a score of 23,948: http://www.3dmark.com/compare/fs/13331782

>>61851663
That's still shit.

>>61851681
A blower reference Vega 64 beating factory overclocked AIB 1080s is shit?

>>61851722
Oh yes it is. These chips are in different leagues given ALU count.
It's still bandwith starved.

>>61846448
Also don't forget how AMD explicitly told der8auer her could post his TR delidding video, then forced him to take it down a day later once everyone noticed that TR had 4 dies. AMD's marketing is as retarded as ever.

>>61851735
I'm aware of that, but these numbers should at least be enough to make the launch not a disaster even if they haven't managed to fix the bandwidth issues yet.

>>61845911
>early release
>amd is finished and bankrupt
lolwat

>>61851722
>J-just wait for the aftermarket cards that'll be overclocked for the 1080ti competitor!
Literally 500w housefires.

>>61848525
AMD is pushing 56 hard because they know it will be a 1070 killer. The 64 is less impressive as it will be the same as the 1080 in most games probably at the same price.

>>61851793
They just need to unchoke the memory.
You don't feed 4k ALUs with 250 gigs effective.

>>61851803
AMD can just trash entire consumer Vega10 and launch Vega11 that's 20mm^2 bigger than GP104 and it'll kill it like no tomorrow.
Right now consumer Vega10 makes no sense, given the bandwith issues.

File: buttmad1.jpg (209KB, 406x480px) Image search: [Google]

209KB, 406x480px

>>61849936
I wonder if the quote spam "kill yourself" amdrone is on medication.

>>61849258
You can't sandbag a GPU like AMD did with Ryzen though. People were getting pissed off at Intel, AMD released Ryzen at the right time. Nvidia is still incredibly popular and can generate massive amounts of hype with surprise launched. look at their Ti and Titan launches, Jensen just walks and stage and gives people GPU's, they don't even have to announce things anymore. How the fuck do you compete against that? Even an expert marketing team would have difficulty coming up with a way to dethrone Nvidia, and AMD has no marketing team at all.

>>61845911
AMD just gave up on g@ymen, all they want is to sell GPUs to miners

>>61851862
You punch Jensen the moment faggot gets cocky.
It worked very well for ATi before.

>>61851804
what do you know what bandwith is necessary, AMDrones relativising everthing

>>61849936
Say it with me

1030 performance

>>61851874
Vega10 needs roughly the same amount of effective bandwith as full GP102.

>>61846448
>>61846481
>>61846513
>>61846968
>>61848240
>>61848497
hes not entitled to free stuff from amd. no one is entitled to free shit from amd. he's a cry baby freak who hasn't learned how to take a shower and cut his air. he stumbles in half of his videos, looks like an awkward autistic fuck in his videos, and in teardown videos he clearly behaves like he doesn't know what the fuck he's doing. he comes off as an asshole snob who thinks he's better than everyone else. i never liked gamersnexus (really, what kind of fucking name is that?) and never will. never liked them long before ryzen, let alone polaris ever launched. my beef actually started over a nvidia article they wrote awhile ago.
actually i know why most of /g likes him. linus is the living embodiment of /v, gamersnexus (((STEVE))) is the living embodiment of /g. a bunch of autistic, smelly fucks who compensate for their laughable small penis with an ego.

>>61851862
>You can't sandbag a GPU like AMD did with Ryzen though.
why?

>>61851862
well, you could sandbag your product and let it nvidia shills know. let them shill for weeks until everyone believes their shit and everyone makes fun of AMD. so it's guaranteed that AMD has it's spotlight because everyone is a cocky punchdown fag nowadays and then BTFO the shills with magic drivers

>>61851939
>AMD
>magic drivers
almost got me

>>61851874
Maybe try actually reading the thread, n/v/idiot. Graphs from Beyond 3D Suite testing (Google it) have already been posted in this thread showing that Vega FE currently suffers from 20-30% regressions vs. Fiji in raw memory bandwidth, effective texture bandwidth, and texel fill rate.

>>61851975
how does a 25% regression against fiji indicate a memory bandwith problem? maybe vega is just architected for higher clocks and needed to sacrifice it. it's faster than the fury at higher clocks, so they got the performance improvement

>>61851966
what? omega>crimson>relive were magic drivers
amd been pumping magic drivers for 3 years now

>>61852018
because of ALU are redone, bandwidth has nothing to do with gpu clocks

>>61852044
what's the memory clock for?

>>61852018
Elongation of pipeline and addition of latency hiding mechanisms has nothing to do with memory bandwith.

>>61851862
I don't think AMD really sandbagged with Ryzen though, they gave a performance estimate and exceeded it. Then they got good clocks from the process which combined with the amazing yields means that anyone buying a 1600/1700 can reliably hit somewhere between 3.8-4.0GHz.

Other than the wonky memory support it was pretty much as good as a product launch they could've hoped for.

>>61852089
They've sandbagged with EPYC. Heavily.
Thats AES-128 engine attached to every memory controller is something.

>>61852018
>how does a 25% regression against fiji indicate a memory bandwith problem?
Because no one in their right mind would design a higher clocked, higher geometry perf evolution of GCN to have even LESS memory bandwidth than Fiji?

Because it's fairly self evident that using 484 GB/s memory and realising only ~300 GB/s of that clearly indicates something is wrong?

>>61852106
It's actually ~250GB/s which is silly.
You don't feed 4k ALUs with 250gigs.

>>61852106
isn't that new fancy cache controller not even working on FE? And it was advertised FOR gaming, so it might bottleneck whole soc?

>>61852116
I was talking about the raw memory bandwidth as shown in >>61848609

>>61852141
New fancy cache controller is for large datasets.
Gayms are not large datasets.

>>61852156
they first demoed it with DXMD, anon
i'm not talking about SSG
https://www.youtube.com/watch?v=85ProuqAof0

>>61852172
It's still for datasets larger than memory.
It does not cure bandwith starvation which leads to unsaturated ALUs.

>>61852192
we don't know how intertwined it is with rest of gpu, disabling it might be what kills bandwidth

>>61852238
I really-really doubt.
But whatever, Vega works in mysterious ways.

>>61852258
it's just that recent 4 generations AMD hyped the hell out of new arch, they are way too tight lipped about vega

>>61852292
Underpromise and overdeliver. That worked well with Zen.

>>61852312
>poor volta
>underpromise

>>61852417
AMD never made any conrete statements about Vega's perf. Silly marketing is silly marketing. And they're right anyway, poor (margins on) Volta.

>>61851903
hahahahahaha you're retarded

>>61852457
>the absolute state of amdrone mental gymnatics

>>61852471
Sopa de favela monkey.

I paid a little under 400 for a 1070 just before the crypto explosion. Should I KMS now that Vega 56 is coming soon

>>61852501
No. 1070 is decent. And no one knows how 56 will perform anyway.

>>61852417
>poor voltage
there was a sticker on 'ge', they just confessed power problems

I wonder if they'll make FP64 Vega for lulz. FP64 + HBCC must be very nice for HPC.

>>61852417
Once you understand that Vega10 is totally memory bound in games, it should become obvious what would (will) happen when RTG does a four HBM2 stack version of RX Vega with literally double the memory bandwidth of Vega 64. The idea that consumer Volta won't get BTFO'd depends on pretending to not know that Vega20 is still coming.

>>61850473
>Not knowing that ALL GPU's get labelled as needing 600W + PSU's
Its to keep retards from buying shitty chink 300w PSU and it blowing up on them.

>>61852585
Vega20 is FP64 thingy (according to rumors). It's not a consumer SKU.

File: AMD-VEGA-10-specifications.jpg (149KB, 1574x859px) Image search: [Google]

149KB, 1574x859px

>>61852570
Yes I know, but the only thing you'd need from Vega 20 is the four HBM2 stack configuration. Or maybe a single GPU version of Vega 10 x2 will allow a quad stack HBM2 set up. Hell, I don't know what the point of a dual Vega 10 card would really even be, since Vega 10 is already bandwidth starved.

>>61852585
>j-just wait for vega 20!

>>61853008
Oh, is Volta out already? I must have missed the launch.

>>61853142
My source says they're announced next month

About time too, pascal is getting old

File: faggot.gif (2MB, 480x270px) Image search: [Google]

2MB, 480x270px

>>61851903

>>61853199
>Jensen will kill his margins before 2018 hits
How about no.

>>61851862
>AMD has no marketing team at all
What are you talking about? Don't you remember "make some noise" and "poor Volta"?

>>61853294
We'll see next month :)

Volta would only get nvidia even more gpu market share right now

>>61852465
>>61853220
t.butthurt gaymursnexus fanboy

>>61853376
They already have all the marketshare. What they need is margins since volume is not going to increase. And Volta is literally lower margins.

>>61853199
>still no reason for anything above a 1070
>pascal is getting old

>>61853412
They already made mad margins on pascal and effectively bumped up each segment by 100 bucks. Getting people to upgrade to volta with some new gameworks shit or proper dx12 support is probably the next step

Once nvidia supports full dx12 everyone will use it

>>61853472
>still no reason for anything above a 1070

If you game at stone age refresh rates maybe

>>61853495
Volta literally means biggur dies thus lower margins. It's not gonna happen any time soon.
Probably late Q1 early Q2 2018.

>>61853472
>playing at 1080p 60hz

Pascal is well over a year old. A gpu cycle doesn't last two years friend, it's due a replacement soon

>>61853472
>400 bucks for a mid range card

Ummmmm na. Hopefully volta goes back to Maxwell prices

>>61853535
amdfags have different sense of time

File: 1463434655833.png (42KB, 653x726px) Image search: [Google]

42KB, 653x726px

>>61851885


1030 performance

>>61853530
>lower margins

Yeah that's what you think until the 1170 is 500 bucks

But it will be hailed as a bargain because it has 1080ti performance

The absolute state of the gpu market

>>61853516
>>61853535
>>61853553
1070 is perfectly suitable for 4k or 90+ fps

>>61853535
it does in 2017
moore has outrun the games

>>61853574
It won't have GP102 perf, since it's not going to be this big. Volta is not even a node shrinks, it's memes: the uarch.

>>61853593
> 4k or 90+ fps
At shit settings maybe

>>61853604
ultra is a bad meme

This thread has gone full n/v/idia.
Anyway GV104 will be ~400-420mm^2 big.

>>61853593
No it isn't. A 1080ti is just barely capable of 4k 60fps on max settings.

1070 just about passes the 1440p check but won't for long

>>61853631
>max settings
literally just placebo to make you feel like you haven't wasted your money on overpriced crap

>>61853620
I'm getting 80+ in the most demanding new games with a 1080

>>61853659
name a specific one, specific res and settings
poorly optimized ones don't count

also don't talk about FPS if you're CPU limited, if you don't know how to tell if you are just fuck off

>>61853677
>poorly optimized ones don't count
Yeah not even going to waste my time with you.

Any example I will give you will just say

>doesn't count reeeeeeeeeeeeeeee

>>61853702
if you post something like mafia 3 that just won't run well on anything it's not a good point

if you post anything that has supersampling or deferred msaa or other such crippling useless nonsense at 4k on max settings, same deal. game would look exactly the same without quadratically scaling performance hogs

>>61853739
Again, I can already tell the kind of person you are, an argument would be wasted.

But here we go, I'll give it one try

Mass effect andromeda , 1440p, everything on ultra, several settings on high (like shadows), hovers in between 55-90 fps.

>>61853773
>frostbite
>ultra
>known performance problem child
>doesn't say cpu or loads
>has no idea why it runs the way it does
>just buys a faster gpu

yep, confirmed retard

>>61853825
My cpu is a 6700k at 4.5 ghz

I'll await for you to move the goalpost somewhere else next