/v/ - Video Games

>>15375270

seriously whats next? GFE flopped so bad they removed the option to install it and their drivers have been dogshit since pascal

>If I’m a reporter, I probably would not sign it. I could pick on several elements but basically, it is intended to provide a strong control over work product of the reporters.

there's one group of people I hate an equal amount to jews

journalists

who are 99% of the time jews themselves.

they are kvetching that if they want to get a free early access version of the card so that they can write a review about it and shill it on their worthless opinion blogs then they have to sign an NDA.

There is an alternative to this, wait until the card comes out, and then purchase it, and then review it. They do not want to this because they are special, they are "journalists", (((they))) aren't supposed to be treated like the rest of the goyim.

>>15375301

lol pretty much but at least they gave us plebs some insight into how bad nvidia is getting.

>vetting and approving "kosher" reviewers to review your latest meme cards

At this point it seems like a rhetorical question, but can they even do that legally?

I mean, I guess they're always 'free' to review and benchmark the cards once they're out, but trying to clamp down this hard seems completely ridiculous and jewry of the highest order.

>>15375309

pricing the midrange 2070 and 2080 cards card out of average consumers hands is bad enough

nobody is gonna buy these

>>15375309

>vetting and approving "kosher" reviewers

that's not what they're doing. they're vetting and approving "kosher" reviewers who will get a free card well in advance of the official release so that they can have the reviews ready and written and get the page views.

this isn't "you can't review the card unless you sign an NDA"

this is "we are not giving you a free 2080 a month in advance to write a review about it if you don't sign an NDA"

>>15375321

t. nvidia

we're onto your bullshit

>>15375313

>nobody is gonna buy these

No one sane anyway.

I never paid much attention to what goes on with these corporations, but how is AMD doing product-wise? The image I have of them is they're basically the underdog on both the GPU and the CPU market, how much of that is true and why shouldn't people just go to AMD if they don't want to support jewish tricks like this?

Are AMD even any better?

The hardware market feels like such a shitshow these days I'm glad I won't be upgrading for some time.

>>15375321

Thank you for reiterating my post man. That's basically what I said had you cared to read the rest.

It means any review coming out for them before launch is not to be trusted because they're well and deep in nvidia's pocket.

My question is can they even legally exert that much control over others?

>>15375321

>this isn't "you can't review the card unless you sign an NDA"

That's exactly what it is you braindead retard. They flat out say they refuse to sign the rediculous overbearing NDA, so they aren't getting an early review card.

But the bigger issue here is that because a proper driver for the new hardware is also locked behind NDA, they won't even be able to source the part early somewhere else to be ready for launch. For example, Gamers Nexus obtained the Ryzen 2000 series CPUs early through one of their sources, which allowed them to do detailed testing in time for launch. There were able to get BIOSes for their motherboards that worked for the CPUs too, lucky for them.

Nvidia is flat out preventing this option for the non-approved reviewers that aren't guaranteed to suck their dicks.

>>15375263

> After its heavy-handed NDAs have sought to muzzle journalists

Oh no

not the journalists

democracy is coming to an end

nazis are everywhere

we must unite and #resist

>>15375353

>, but how is AMD doing product-wise?

fantastic

their gpu's are not just in consoles but alot of laptops apu's and even some intel systems

vega isnt so good at high end but shines at low-midrange suprisingly

>>15375358

>this isn't "you can't review the card unless you sign an NDA"

<That's exactly what it is you braindead retard

> so they aren't getting an early review card.

early review card

nothing is stopping them from reviewing a card they purchased themselves.

> they won't even be able to source the part early somewhere else

they can wait until release, but of course the (((journalists))) won't be able to goy harvest traffic nearly as much if they wait will they.

i'm not advocating anyone buy this stupid fucking card or install nvidia's botnet drivers, but these journalists are kvetching over not being able to get a card before anyone else, not that they can't review the card at all once it's released.

>>15375354

Professional cards typically go into a rack with a bunch of other GPUs and high airflow fans blow air into the front and out the back, and then the hot air is pumped into a cooling system based on condensation. This allows Nvidia to ditch the fan and just fill the space with more heatsink.

>>15375371

You seem to completely miss the point continuously. I'm not sure if you're doing it on purpose or if you're just that dense.

It ensures that any idiot dumb or dishonest enough to accept those conditions is going to have to review those cards the way nvidia wants them to.

If you thought getting free cards to review was already conflict of interest then this basically blows that out of the water since they may as well be nvidia marketers.

>>15375371

Nvidia can give them the card for testing and then require them to be returned when finished. This often happens with engineering samples for example.

They should have as many outlets as possible reviewing the product for maximum exposure and thus sales. To limit it to only a chosen few good goys shows who Nvidia really are. Also remember HardOCP are the ones that exposed the GPP, which was such a shitty idea that Nvidia had no choice to back out of it once everyone knew what they were doing.

>>15375397

Nice 7, faggotm

>>15375371

remember the geforce partner program shit tho?

fucked

>>15375382

lol

>>15375382

The way things are going they'll need every bit of ad money to buy those ridiculously priced cards to review them.

How much do you get paid to divert this hard?

>>15375370

>their gpu's are not just in consoles but alot of laptops apu's and even some intel systems

That's nice for them, but I was more looking for a more detailed rundown of where they stand. Can they finally properly compete with nvidia/intel on both markets? See my other questions.

>>15375429

>Can they finally properly compete with nvidia/intel on both markets?

Yes, you dipshit. Do your own research next time.

>>15375450

Shit man, sorry for trying to stir some conversation on viable alternatives. I'll know not to do that next time.

Not really, go fuck yourself.

>>15375467

lol im not the original guy you asked dont mind that autist m8

Yes Zen was a runaway sucess and they only just fired raja and that other vice pres idiot so now the company is in much better state after 2020 when the ps5 launches their next gpu will be a full on monster.

vega 7nm is apparently awesome but only for servers sadly no cut down desktop/laptop verison planned now that navi is done.

So whatever comes after navi will be insane.

>>15375270

The fuck even happened to her in the third shot?

>>15375429

>>15375450

>>15375467

Here let me answer for him:

AMD's Ryzen CPUs hit Intel hard. Intel was forced to speed up their product roadmap, rushing the 8000 series and the Skylake-X/Kabylake-X parts to market in order to stay ahead of the game, and is also forcing them to finally give the mainstream something other than quad cores. AMD is finally seeing profits in their quarterly reports. Threadripper's high core count is forcing Intel to bring their absolute best chips down from the server market and into the high end desktop market. The Xeon 8180 is a 28 core CPU that they sell for servers at $10,000 a piece, but now they have to bring it down to the $1500-2000 range on desktops (see: Computex demo this year), which is a huge profit loss for them. They are currently unable to produce CPUs with a higher core count while also maintaining good clock speeds, the yields on such monolithic dies are way too low. This is why Intel has hired Jim Keller, presumably to have him design multi-die processors like he did for AMD with Ryzen.

Vega was good but not good enough. Vega 56 traded blows with the 1070 nicely, and Vega 64 was around the 1080 but not quite as competitive. They tried to get the liquid cooled Vega 64 to compete with the 1080 Ti but there's just no winning that fight. In the lower end though, Vega is an interesting part. Interesting enough that Intel decided to get semi-custom Vega chips to put into their Intel NUC devices - tiny box computers that use lower end/laptop parts in them. Vega is much better at graphics than what Intel can currently provide, which is why they hired Raja Koduri to help them improve their graphics to the point where in 2020 they could launch dedicated cards into the market.

>>15375525

>AMD's Ryzen CPUs hit Intel hard

Beautiful, but its not enough. AMD needs to maintain the momentum even further beyond Vega to even start bringing the kikes to heel.

>>15375514

blood out of her mouth

>>15375539

vega was a flop on the desktop side but i think it was quite successful on the enterprise side

>>15375539

The EPYC 2 server CPUs in Q1 2019 will be the make-or-break point for AMD. The worst case scenario is that they are 48 core CPUs, and the best case scenario is that they are 64 core CPUs, neither of which Intel can compete with, unless they pull a fucking miracle out of their ass.

When those CPUs land, we will also get a preview of what will soon after hit the desktop that year. If they are 48 core CPUs across 4 dies, then the 3000 series will top out at 12 cores on mainstream desktop. If they are 64 cores across 4 dies, then the 3000 series will top out at 16 cores on mainstream desktop. The only other possibility is that EPYC 2 might not be made of 4 dies, but instead something like 8 dies, each still containing 8 cores, and they can fit due to the 7nm die shrink (courtesy of TSMC, the company that makes the Nvidia GPU dies).

>>15375563

>neither of which Intel can compete with, unless they pull a fucking miracle out of their ass.

they cant

the i7 9000 series doesnt even have ht on the i9 9900 lul

>>15375563

>48-64 cores

What the fuck kind of shit are they smoking at AMD?

>>15375566

>i9

Its a straight up retard tax for morons thinking "More Expensive = Better", I'm glad that shit is dead in the water.

>>15375571

>What the fuck kind of shit are they smoking at AMD?

jim kellers cum out of mumma su's vag mate

>>15375563

>64 core CPU

In their quest for more coredakka, did they ever think of what they'd do with all of that coredakka?

>>15375587

One does not wonder what one uses coredakka for.

One merely strives to acquire more coredakka.

For no matter it's form, there is no such thing as to much dakka.

>>15375263

Archive it, faggot.

https://archive.is/WhCrN

>>15375411

>remember the geforce partner program shit tho?

Remember when Threadripper had a bunch of restrictive embargoes which even included "unboxing"? This is why reviewers need to get their samples through other means that do not force them to sign a bunch of nonsense agreements. Even if it means waiting for the release.

The scramble to be the first to publish is sickening anyways. You see it when there's a shooting and these vultures are circling social media before the bodies are even in the ground or before the shooting is even over begging for comments and footage from anyone involved.

>>15375587

>>15375644

>64 core CPU

Sounds impressive but consider the 32 core 64 thread 2990WX. Single core memory throughput is shit since two of the four dies are not connected to memory or PCI-E thus have to go through the two that are. It made the 2990WX shit in a lot of tasks so even the previous generation 16 core 32 thread 1950X was beating it.

>>15375587

Cloud computing

>>15375980

this was my entire point. fuck nvidia, and fuck journalists, and "journalists" are a greater scum than nvidia jews.

>>15376031

Cloud folks don't want threadripper, the cores will starve as they can't be fed with ram fast enough for general use. They're for mustards seeking rice and possibly will have found a market in renderfarms and video production as those are massively compute-bound. I don't think they're any closer to taking the extremely valuable server market.

>>15375353

>how is AMD doing

Actually pretty well (for them). Zen is doing ANUDDA SHOAH in server space, which means money. I've heard that Vega for workstations is actually good and best bang fir buck, but even then it's not the most powerful card out there. For gayman, Vega is too little, too late, but supposedly when AMD finally finishes their wunderwaffen drivers it should be a big jump in performance. Too bad they can't make drivers for shit.

>>15376163

Nah that's bullshit. Vega's performing about where it should be, the problem is that it's starved for memory bandwidth because it was designed when Samsung was promising that the HBM chips would run at 1.6GHz and when they were released they could only manage 1.2

>>15375997

That issue does not apply to the 32 core EPYC which is running 8 channel memory, nor will it apply to a 64 core EPYC 2. That issue is exclusive to the 2990WX, and that CPU only.

>>15375313

>nobody is gonna buy these

You underestimate the goy tendencies of Mustard race faggots

>>15375313

>nobody is gonna buy these

Keep dreaming. Preorders sold out like a week ago

>>15376199

>>15375589

I fucking love that channel young scrolls a based

https://youtu.be/iR-K2rUP86M

>>15376352

It's actually not because nowadays all movie files are offloaded to dedicated decoder hardware instead of general purpose execution units

>>15376376

Next is 7nm workstation cards late 2018 or early 2019, new consumer cards aren't coming until late 2019 probably.

>>15376376

No idea I guess they are gonna wait until rtx launches and go from there

https://youtu.be/EGBbw2DLBEA

Back in 2016

Also this is interesting powervr had a 150usd mobile gpu that could out proform the 2070 and the engineers work at amd now

>>15376447

>Obviously they didn't buy into speculative execution and the hyper thread meme.

But they did… they just didn't fuck it up like Intel. Most CPUs since about 2010 have had speculative execution and AMD CPUs all have hyperthreading as standard (unlike Intel which only have it on their higher end CPUs).

AMD CPUs also don't have AVX512 yet which results in Intel cores being physically larger.

>>15375571

>What the fuck kind of shit are they smoking at AMD?

Its not as good as you would think, like the LGA3647 Xeon CPUs the Threadripper and Epyc CPUs are multiple dies on the one interposer. Because not all the cores are on the same die there are NUMA-related effects similar to having a multi socket system which will impact performance in heavily threaded applications.

>>15376419

>"A-At least it's not AMD"

>>15376552

AMD cpus are also notoriously short lived. They die before their warranty very often if you actually use them for video editing or playing games that come within the cpu limits.

>>15376552

>AMD CPUs also don't have AVX512 yet

Because literally no one uses it. We've been using AVX-512 and while it was faster than AVX2, using GPUs instead completely blew AVX-512 out of the water in terms of performance. Especially with HSA.

>>15376643

Man you intel shills are really reaching

>First-gen AMD Thunderbird overclocked to 1.4Ghz

>Used for video editing and encoding for 15+ years

>on stock cooling

>still runs fine

Meanwhile the P3 OCed to 800Mhz it replaced made it a year before it burned up.

>>15376680

Just so I'm not (1) and done. I became a poorfag after building that machine for many years and it was my primary machine until 2008 or so. So about 8 years of constant use with the cpu pegged at 100%. I was a fansubfag in those days and several of the Chinese cartoons so many faggots enjoyed from 2000-2006 were encoded on it. Later on when x264 came around I watched a lot of series on it which pegged out the CPU constantly until I eventually got a job and stopped watching media all together.

Like I said I had a P3 slot type processor before this that was also overclocked. I got a years use out of it doing encoding almost 24/7 until it crapped out. Same problem with an early P2 machine I built where the K6 I put in afterwards also still runs to this day. My current machine is an Intel again that I built for video work but I'll be replacing it with something from AMD once I re-purpose it for an arcade cabinet I'm building. My next shit posting machine will be a POWER9 or POWER10 though because I'm trying to ween myself off the botnet.

>>15376666 (checked)

>We've been using AVX-512 and while it was faster than AVX2, using GPUs instead completely blew AVX-512 out of the water in terms of performance

Satan, while you may be correct about GPUs having superior throughput compared to CPUs the downside is the higher latency and overhead of launching kernels. There are many applications where doing something on a GPU just isn't feasible because of the low latency required, or applications where it makes more sense to run some minor part of the compute on the CPU so the GPU can be utilized for the main compute load. Also not everything has a GPU, for instance storage nodes aren't going to have a GPU but the user might need to run a filesystem which does SHA2 which can be greatly accelerated by AVX512.

GPUs are good at what they are designed to do, but they aren't a silver bullet.

>>15376666

GPUs aren't a replacement for the kinds of things AVX is intended for.

>>15376919

Yes they are you dumb shit, why do you think GPUs are used in supercomputers, to play PUBG on medium settings?

>>15375382

>tfw they still can't beat adblock and can't do shit about element blockers

>>15375353

AMD is doing decent to good. Not in the negative anymore.

Ryzen was a massive success and put AMD in a very competitive postion.

Ryzen APUs are in a few laptops and are providing good performance at the 400-500$ laptop range. Intel/Nvidia combinations can't really compete in that range.

RX 570s and 580s got scalped to hell by miners and are/were only competitive if priced at MSRP. If MSRP, 580 was equal or better than the 1060, but no actual consumers are getting their hands on the fucking things.

Vega 56 and 64 are not too good. Also hunted by miners, but not a success in the consumer/gaming sphere. Not very competitive with 1070 or 1080 (Ti)

AMD Epyc server CPUs are doing good. Pretty high performance and competitive.

>>15378333

>Vega 56 and 64 are not too good. Also hunted by miners, but not a success in the consumer/gaming sphere. Not very competitive with 1070 or 1080 (Ti)

They might be soon should nvidia decide to gimp their drivers like they did with Maxwell cards previously.

Judging by the inappropriate amounts of shilling and jewery surrounding the Turing GPUs I'm inclined to believe they will.

>>15376968

They aren't. Stick to video games, kiddo.

>>15375358

How incredibly retarded do you have to be to write this post?

You quote

>this isn't "you can't review the card unless you sign an NDA"

and then explain

>this is "we are not giving you a free 2080 a month in advance to write a review about it if you don't sign an NDA"

While acting as if the post you're responding to has never written it.

Please kill yourself.

>>15381809

iirc, umatrix allows you a lot more control over what resources can be loaded, and which are blocked. It works on scripts, but also on XHR, frames, cookies, etc. You can blacklist individual resources, different types, or different providers (like googleadservices). It also allows you to whitelist in the same way.

It requires a little getting used to, because you have to get to know all the different resources and types responsible in loading a page, many of which are clean and just necessary for the page to operate.

>>15383097

>that can run 2k at 140hz and I'll be set for the next 10years

There is more to graphics rendering than just the amount of pixels and frames.

You can already play minesweeper on 8k with 200fps on an integrated graphics card.

>>15383322

He obviously meant intensive graphics, anon.

>>15383097

yeah nvidia as sandbagging massively.

7nm when amd launch their new mcm cards in 202x will make them get off their ass

It would be very bad if Nvidia left the industry. They've been trying to push AI tech and research lately. Not good

>>15384342

They left the gaming industry to rot on shit overpricedpascal for 3 years and replace it with a even bigger more expensive slower card with features that won't be commonplace for another 2 generations.

Mate the consumer gpu industry has been dead since 2013 we literally peaked then since it's just been cut down quadros and bigboi tech from vfx studios trickling down to pc

imagine paying nearly 40% more to get ~20% performance bump and ray tracing that isnt even mainstream until like 3 years later when it actually gets near 60fps in 720p on like 2070

every card costs going up by another card

so 2070 costs about as much 1080

great

>>15384608

1080 has dropped to around 400usd so it's effectively double

Can't wait for normalfags to bend over and take the 5gb 2060

>>15384631

The 2060 won't even have ray tracing or DLSS, further reducing it's already likely trash value.

>>15384673

Lol it's nvidias worst midrange card turing will be remembered as one of the worst nvidia gpus besides Fermi

>NVIDIA ever

How does this shithole manage to get worse every time?

>>15390068

They can get away with you.

>>15384631

Might see a rebrand of the 10 series. Rumors are being thrown around that Nvidia has contractually obligated all of their big AIB partners to take their massive overstock of 10 series chips if they wanted any of the 20 series.

It was only about 2 months ago that all of Nvidias AIB's had returned 10 series chips which seriously harmed Nvidia's bottomline so this, the changes to the reference reference "founders edition" is probably a retaliation for not covering Jensen Huang's next leather jacket.

We could see everything from the 10 series being sold very very cheaply.

>>15376447

I had a fever dream last night where you could buy an amd adapter that turned a 16 core ryzen into a quad core processor. Imagine 4 cores per core, hyperthreading on steroids.

35% for 1,300$

>>15399360

Sounds fair

>>15383454

Yeah, and that card is going to run those intensive graphics is obviously going to keep up for 10 years.

>>15383487

>there isn't even a GPU on the market that can render your fucking desktop at 8k 200fps

Of course there are, you're just not going to find a 200hz 8k monitor if that's what you're talking about, nigger.

>>15405908

You're one sick puppy

>>15376352

No, Netflix's DRM for 4k (much like UHD 4K BDXL drives) on PC has completely insane requirements, so strict that the only 100% legal way to play them is with an integrated GPU connected to an integrated monitor. As such, the only systems meeting their requirements are all Intel laptops.

>>15376419

A Xeon Phi-style GPU would actually be awesome for stuff like (genuine) raytracing, and Itanium's VLIW design would actually be well suited to GPGPU.

>>15383487

There are numerous 240Hz monitors on the market (plus, of course, many bigger CRTs back to the 90s could hit 200Hz at lower resolutions), and there are prototype 480Hz monitors. All of these are easily driven by any GPU under Windows or whatever:

https://www.blurbusters.com/4k-120hz-with-bonus-240hz-and-480hz-modes/

And the maximum frame size from a single GPU is already three 8K 60Hz displays

https://en.wikipedia.org/wiki/AMD_Eyefinity#Feature_overview_for_AMD_graphics_cards

>>15414870

What the fuck are you trying to say

>>15423824

He's making an ironic suggestion that consoles should exist, and a lowkey platform wars bait post.

>>15425440

People in Australia have always liked it up the bum.

If Quads, Nvidia will crumble.

>>15426663

One more roll, dammit.

Voodoo will crush the entire GPU industry.

>>15427013

Damn shame.